Fools!-I've built the first INSTANTHistory product. I've put together the daily history for monthly RS-IBD from 4/84-12/99. The master file, including all graphs and position/portfolio breakdowns, is +27Mb large. Compressed it is 9Mb. Yeah, pretty big, but it is 16 years of daily data.....It shows taht someone that only invested in the Position #1 stock for that period had a 163,100% return. Here are the returns for the various portfolios:Top 1 127,700%Top 3 22,000%Top 5 46,600%Top 10 17,100%DJIA 910%S&P 830%Naz 1,500%-I attempted the first stop loss research for a monthly portfolio for the long-term. I tried something that I thought would be actually pretty conservative. What happened to be a good hypothesis turned into an execution disaster.I decided to try this for a stop loss screen. For a stock that is in the #1 position, when the cumulative return since the time of investment goes over 10%, the holder of the screen sells out of the stock before the end of the month when the stock drops below half of the return the highest closing after that point. If the owner sells out before the monthly update then cash is held until the next update. Then the stock is reevaluated for appreciation from the new purchase price.Example: You buy your stock at the beginning of the month at $100. It goes up within two weeks to a maximum of $115. That is a 15% return from the initial investment. Half of that is a 7.5% gain. That reflects a minimum stop price of $107.5 that is the stop price. If the stock closes below $107.5 before reaching a new "ownership" high, then you sell at that closing price. You own cash until the next monthly update. If it reaches a new high then you update the new 50% cut limit. There is no limit if the stock price while you own it does not go up above 10%. If you hold the stock for more than a month then the stops carry forward until reaching a new high (in which the stop will be reset with 50% of the new high), you hit the stop price, or you change stocks with a monthly update. I tried this on the 168 months of the RS-IBD #1 position for 4/84-12/98. By doing this +10%-50/50 return screen your return was reducted by a factor of 6.4x!!! Simply owning the #1 stock during this period returned a factor of 1632 (163,100%). Using those floating stops caused the portfolio to sell out 24% of the time before the end of the month resulting only in a return factor of 254 (25,300%). That is a devistating low return for owning one stock and dropping out of the market in an attempt to perserve the gains. It appeared that the straight version vs. the stop loss was the same through the 80s, but as market volatility increased the screen started stopping out more often and losing out in the wipsaws.I'm going to continue to look at this as I increase the number of holdings for monthly RS-IBD. I'm also going to look at RS-13wk to see if this effect is only for the heavily screened variables or if it affects the simplier screens too. I'm also going to look at something like a 85% limit screen too. I'm not impressed so far. I thought a conservative stop loss would help the screen but it appears to have almost fatally damaged its ability to function. We'll see how future results go.Sux
Sux,I continue to be amazed at the work that you and others put in on this board - you amaze medan.
<<I thought a conservative stop loss would help the screen but it appears to have almost fatally damaged its ability to function. >>It makes sense to me. Everybody who watches their MI stocks on a daily basis (twice daily? 10 times daily??) knows one thing for sure: They are extremely volatile, sometimes swinging up and down 5%-10% within a single trading day.I think what you have described is not a "conservative stop", but rather is a "hair-trigger sell order". Maybe it would be convervative if applied to the typical Dow Jones stock (excepting MSFT and INTC, of course!), but surely not to PEG-type stocks.Ray
I decided to try this for a stop loss screen. For a stock that is in the #1 position, when the cumulative return since the time of investment goes over 10%, the holder of the screen sells out of the stock before the end of the month when the stock drops below half of the return the highest closing after that point. If the owner sells out before the monthly update then cash is held until the next update. Then the stock is reevaluated for appreciation from the new purchase price. Example: You buy your stock at the beginning of the month at $100. It goes up within two weeks to a maximum of $115. That is a 15% return from the initial investment. Half of that is a 7.5% gain. That reflects a minimum stop price of $107.5 that is the stop price. If the stock closes below $107.5 before reaching a new "ownership" high, then you sell at that closing price. You own cash until the next monthly update. If it reaches a new high then you update the new 50% cut limit. There is no limit if the stock price while you own it does not go up above 10%.Sux, it's great that you're going systematically after stop lossess, but I think you started out in the wrong direction.Think about what it means for a stock to go up 10% within a month. That's a HUGE winner. If that's a pre-condition for triggering the strategy, then it isn't a stop loss strategy, it's a stop win strategy. It's selling the best winners as soon as they retrace a little, to "only" a 5% monthly gain or more, which is an 80% CAGR.I wouldn't consider any stop loss until a stock had dropped at least 10% below the purchase price. Then we're beginning to look at a significant loss of momentum.Also, I doubt that stop losses will be effective for monthly holding periods. There's not enough time for a trend to develop, and if you trigger a stop loss it may be so close to the end of the month as to make no difference. BWDIK? A backtest is what counts. I would just recommend looking at longer holding periods too.Elan(posted and emailed)
Sux,That's an awesome database. I had proposed this same kind of "save gain" strategy in post 72707, but didn't have any database. Could you run through a case where you set the cutoff much higher, say maybe a 30% gain during the month, to set the sell stop at 25% gain? My thought was only to start putting stops in if you have a REALLY good month going (e.g. +2 sigma).Tails
I tried this on the 168 months of the RS-IBD #1 position for 4/84-12/98. By doing this +10%-50/50 return screen your return was reducted by a factor of 6.4x!!! Sux A Stop Loss screen's objective is not to maximize CAGR, but to minimize the probablity of significant losses (i.e. minimize the worst drawdown). As an individual about to go into retirement, my greatest fear is an October '87 type drop followed by a '72 to '86 flat market. The initial punch knocks your portfolio worth down by 40% and the second strangles it for >10 years. If I knew the hits were going to be followed by great rises, which is the scenario we have experienced the last 15 years, I wouldn't be worrying.DGM
I wouldn't consider any stop loss until a stock had dropped at least 10% below the purchase price. Then we're beginning to look at a significant loss of momentum.Also, I doubt that stop losses will be effective for monthly holding periods.+++++++++++++++++.Sux, Good idea for a start but I agree with Elan that you are killing a chance for big winner. Rayvat's RSV board's idea might be a better idea for a stock that is off like a Visx or qualcomm. I realize that Elan'z 10% stock loss is a good idea when you pick up a dog or one week wonder. Did you see anything for monthly screens where the momentium corrects downward and you watch your 40% or 50% profit fade out the week before you were due to trade down to nothing or a loss on rebalance day? How many people get caught in this situation each month? Did you see any stop loss ideas that covered this situation? (You had mentioned a 50/10 in one of your posts last week.)Keep up the great work!doc
<<<I'm not impressed so far. I thought a conservative stop loss would help the screen but it appears to have almost fatally damaged its ability to function. We'll see how future results go.>>>My humble suggestion is related to Kuperman's PEG stop loss research. Maybe re-evaluating the 20-30% stop loss "sweet spot" that he found? I also think a stop loss might show a better result in quarterly or longs holds.My basic premise and doubts still hold however. Will this increased effort, monitoring, recalculating of prices, etc, statistically improve the returns? Especially when the increased time is factored in? What do I do in a couple of months when I'm in mainland China? True, I won't be that far from Hong Kong but as far as internet access I might as well be half a world away. Wait a sec, I WILL be half a world away.While I applaud the research, and it is a question that begs answering, I do so with caution. The primary premise of Foolishness is that there are much more important things in life than watching a bunch of ticker symbols. We might be becoming too Wise if we're not careful.JLC
I continue to be amazed at the work that you and others put in on this board - you amaze medan. Thanks. Nothing like a little ego boosting to start off Monday!Sux
I tried something that I thought would be actually pretty conservative. What happened to be a good hypothesis turned into an execution disaster.Sux, I applaud yor efforts.I wonder if some of the difficulty might be that you were often being stopped out when the whole market was dropping, and the scheme might work better if you were triggering the stop when the % difference between the stock and a common average (S&P?) dropped. I don't even know if such a concept is possible within the mechanics of your system, but if this idea has any appeal, could you perhaps check the hypothesis by running the stop loss program on QQQ itself, or SPX, and seeing whether you get a comparable number of "sells"? If so, triggering on the difference should produce fewer sells.Hankemailed and posted
It makes sense to me. Everybody who watches their MI stocks on a daily basis (twice daily? 10 times daily??) knows one thing for sure: They are extremely volatile, sometimes swinging up and down 5%-10% within a single trading day.I think what you have described is not a "conservative stop", but rather is a "hair-trigger sell order". Maybe it would be convervative if applied to the typical Dow Jones stock (excepting MSFT and INTC, of course!), but surely not to PEG-type stocks.RayWhat would you recommend? I was going off of what you mentioned at MICon about your 50/50 rule. I just used the +10% threshold as an "activation point" for the rule. I believed that would be an acceptable point in wanting to perserve gains. Do you think using that type of methodology on a short-term screen is not good, or is it more that my bound is way too tight? Should I be looking at cumulative loses instead of perserving gains?I need to figure out a way to get that Excel graph of period returns by positions on my board. You can see how the cumulative gains and loses mount for each period. Maybe you'll see something I don't--I'll admit I'm new to stop-loss strategies. Sux
I'm not sure this will help the discussion, but here is some further information on this topic. A few months ago, partly out of boredom and partly out of curiosity, I began experimenting with various rules for trading VL T1/2 stocks, rules completely divorced from MI. As part of that excercise, I also developed stops for a number of the models developed. A few of the models did better without stops, just using sell rules that were the obverse of the buy rules. But most did better with stops, either using just stops or adding them to sell rules. Stops that were effective were of two sorts: stops designed to preserve capital and stops designed to preserve profit. Of the former, rules that took losses of 20% versus initial purchase price worked best. That sounds large, but anything tighter just didn't work. Using a stop of that size generally resulted of a loss of 3-8% on a portfolio of 10 stocks.Stops of the latter type, stops on profit, generally worked best when set to keep 80% of profit after a profit of 20-30% was hit. Anything tighter didn't work.These tests were run using daily data on T1/2 stocks for the past two years. That fact, combined with the fact that the models used were not based on MI screens, may make the above comments less useful. But they are offered for what they may be worth.arezi
I wouldn't consider any stop loss until a stock had dropped at least 10% below the purchase price. Then we're beginning to look at a significant loss of momentum.Also, I doubt that stop losses will be effective for monthly holding periods. There's not enough time for a trend to develop, and if you trigger a stop loss it may be so close to the end of the month as to make no difference. BWDIK? A backtest is what counts. I would just recommend looking at longer holding periods too.ElanWell, how about pointing me in a good direction to try out? I've got the daily data for monthly RS-IBD--I can probably test something like a 15% down limit from original purchase price (or how about a 15% per period--you could end up owning the same stock for multiple periods even though it is still dropping--CPQ in '89 devistated the RS-IBD bullet by dropping 15% each of 3 months in a row while it was ranked #1!), but I'd like an idea or direction to go in.Again, not very familiar with "the rules" except for O'Neil's 8% strategy, which, looking at the data so far, looks way too tight.Sux(sent everywhere)
Sux,That's an awesome database. I had proposed this same kind of "save gain" strategy in post 72707, but didn't have any database. Could you run through a case where you set the cutoff much higher, say maybe a 30% gain during the month, to set the sell stop at 25% gain? My thought was only to start putting stops in if you have a REALLY good month going (e.g. +2 sigma).Tails I'll put it on the block for testing. Remember I have all the data, but it takes a little bit of time to set up the portfolio you want. What do you want to use? 1 stock? 3 stocks? 5 stocks?How about that 30% gain/25% stop? Cumulative over the life of owning the stock (you could have owned BBY and DELL for over a year in the 5-stock version in '97 and '98, respectively) or just monthly reset every "purchase" day? Stops are a lot more complicated mechanically than I thought. Sux
While I applaud the research, and it is a question that begs answering, I do so with caution. The primary premise of Foolishness is that there are much more important things in life than watching a bunch of ticker symbols. We might be becoming too Wise if we're not careful.JLCIt reminds me of when ethics and morality finally push their way back into and in front of our scientific gods and technology-centric society. We can continue to push the envelope of technology, but it doesn't mean we should use it. Right now that debate is for biotechnology whereas 50 years ago it was for nuclear research. Atoms being split or genes being spliced are not immoral in and of themselves, but how they are used for the contribution of society or its detrament deems the act moral or immoral. I think the stop-loss research will be like that. Research that is done and knowledge is gained, but it is not used because it is counter to our accepted philosophy of MI. A decision by the group not to use what is known--willing restraint, if you will. Some will, but it will be a minority of participants. Sux
I wonder if some of the difficulty might be that you were often being stopped out when the whole market was dropping, and the scheme might work better if you were triggering the stop when the % difference between the stock and a common average (S&P?) dropped. I don't even know if such a concept is possible within the mechanics of your system, but if this idea has any appeal, could you perhaps check the hypothesis by running the stop loss program on QQQ itself, or SPX, and seeing whether you get a comparable number of "sells"? If so, triggering on the difference should produce fewer sells.Hank Three things come to mind--tracking the index direction, which index, and use a Beta correction factor to correct for volatility of the stock vs. the index?The data is out there. My INSTANTHistory file has daily data for the three indices, so I could do a directional indicator as a "trade permitter" for the SL to activate and execute. That might reduce some.Which index? If you have a portfolio of Naz and S&P stocks with an occasional Dow which do you use? They have become very divergent lately.I guess I'd have to have a number of recommendations before trying something like that. Interesting though.Sux
These tests were run using daily data on T1/2 stocks for the past two years. That fact, combined with the fact that the models used were not based on MI screens, may make the above comments less useful. But they are offered for what they may be worth.arezi Hey, doesn't hurt to look, eh?Would those be cumulative gains or just monthly gains? Again, that's a pretty important whether you "reset" every time the stock reappears in consecutive months.Sux
Sux: Sounds like a trailing stop.Could you try a simple 30% trail stop?IIRC,this is Ray's stop for his emotional picks.TIA Wayne/BassAttack
Sux asked: Would those be cumulative gains or just monthly gains? *******************************************************Cumulative.arezi
Which index? If you have a portfolio of Naz and S&P stocks with an occasional Dow which do you use? They have become very divergent lately.I started out using the S&P 500, because that seemed to be the yardstick against which most mutual funds measure themselves, and I wanted to beat most mutual funds. It seems to have worked in the screen test I have been posting, even when the Nasdaq tanked in the spring.Hank
Well, how about pointing me in a good direction to try out?Several people have already made suggestions. Here's what I would do, starting from the simple and going to the more complicated.First, test a simple stop loss based on the purchase price. Assume that the money from the sale is kept in cash until the next rebalance. Try selling at -5% from the purchase price, -10%, -15%, etc., maybe down to -40%. See if you can identify a sweet spot range. Let's say you find it between -20% and -30%. If you see such a range, try to zoom in at 1% increments.The lower your stop, the less stocks will be sold. I suspect that the number sold at -30%, let's say, in a monthly holding screen will be so small as to be statistically meaningless. Try to test the same idea on a quarterly, semi, and annual screen.The next step might be to set the stop off the latest high. e.g. Instead of selling at -25% from the purchase price you sell at -25% from the high for the holding period.Next step, like Hank suggested, might be to decide based on a drop relative to the market index. Try the S&P as the most obvious benchmark. The stops would probably be much tighter. Dropping 10% more than the index might be enough to trigger the sale.Later, it would be interesting to see what happens if you buy another stock instead of going to cash. It may change the complexion of the test entirely. How do you decide what to buy? That's an open question in my mind.Overall, it's important to change one variable at a time as you go along.There's probably enough here already to keep you busy for three months if you don't automate the process. So I'll stop here.Elan
". . . someone that only invested in the Position #1 stock for that period had a 163,100% return. Here are the returns for the various portfolios:Top 1 127,700% . . ."___So is "Top 1" a typo for Top 2?
Which index? If you have a portfolio of Naz and S&P stocks with an occasional Dow which do you use? They have become very divergent lately.Sux, your question got me thinking about my knee-jerk reply (That's one of the things I really like about this board). I've been using the S&P (because I use the S&P), but in this market, what I really want to do is beat the NASDAQ. I think I'll set up another spreadsheet for my screen test that finds which stocks have beaten the NASDAQ over the last three months, and see if I get better or worse returns than the ones that have beaten the S&P.I didn't really want to go on that mororhome trip next month to see my son's new house, now, did I?Hank
Sux:Sounds like a trailing stop.Could you try a simple 30% trail stop?IIRC,this is Ray's stop for his emotional picks.TIA Wayne/BassAttack Ok. I'm not sure which you mean. 30% down from where you bought it originally, 30% down from the peak of your holding, or 30% down from the last evaluation? Big differences there. Again, clarifications are needed since the screens will hold stocks in different positions for long periods of times. Several periods can pass before a falling stock gets off the screen. I get confused easily. Sux
I'm wondering instead of using a stop loss, it would have been better to avoid stocks over a previos period of time had advanced to much. This year, 2 of the worst performing stocks, QCOM & CMGI, both losing around 60%, were among the best performers in 1999. QCOM increased around 20 times and CMGI increased nearly 10 times. An idea would be to look at the previous 12 months, 6 months or 3 months and look at how a picked stock had done. For example if a stock had advanced 5 times in the past 3 months, then we'd skip that stock. Another idea would be to look at sometype of moving average; such as the 200 or 50-day moving average. The idea being to avoid a stock that is selling at a multiple to its moving average i.e. 3 times its 200-Day moving average or 2 times its 50-Day moving average.Just some ideas to ponderGary
So is "Top 1" a typo for Top 2?No. There was careful wording there and it's something to pay attention to. Maybe I wasn't clear so I appologize.The Top 1 portfolio is usually the #1 ranked stock and is often called The Bullet Port; however, the Top 1 stock may hold 1, 2 or even 3 stocks (as in early 1999) because they are in statistical ranking ties with one another. Mechanically, you must hold multiple stocks b.c. they are tied with one another and they are both or all truly the #1 stock. I did the backtest where ties for Top 1, 3, 5 and 10 were in proper proportion to the rank and number of tied stocks at the cutoff point. A tie at 10th place between two stocks means that each of those two 10th stocks were 5% of the total portfolio (the other 1-9 position stocks were 10% each per normal method).The #1 position stock is just the first stock listed on the rankings page. This stock will be in the Top 1 portfolio, but it does not mean that there is not another stock tied with it. RS-IBD and Formula 90 have many ties due to the ranking system so the #1 position stock does not necessarily translate (although it often does) into the Top 1 stock or Bullet Port. I just chose to run my first crude experiment on the #1 position stock without ties so as to not have to manually look for stop losses for two or three other stocks simultaneously. The results will differ if I do include those stocks when there were ties for #1, run the stop on them and blend their returns, but I'd already spent all Saturday finishing the daily data collection for monthly RS-IBD and all Sunday running that stop to see if it was somewhat effective or finding time-based differences (slight--more stops occur in 90s due to volatility).It was a rough and crude experiment, but effective to spark a lot of activity here.Sux
Sux,The idea of a 10% sell rule is an interesting one. I proposed something similar several months ago, and Peter recently agreed to backtest it. Essentially, it would give a ranking to the stocks based on the current price relative to it's 52 week high. (Ideally, this would be a very recent high.) Anyway, it would modify the RS-x like this:In any screen, substitute the RS ranking by this two part ranking:1. Compute a new variable %52wkHi equal to current price/52 week high.2. Create 7 "buckets" based on %52wkHi such that:%52wkHi Bucket85-100% 670-85% 555-70% 440-55% 325-40% 210-25% 1 0-10% 0This can be done using the formula:Current Price------------- + 0.04999952 week High------------------------ = Bucket .15then truncate the result (e.g.6-6.999999 would become 6). With out the + 0.049999, the top bucket would contain only the top 10% of stocks, while bucket 0 would contain the bottom 15%.For 10% brackets as you had suggested, the formula would simply beCurrent Price------------- - 0.000000152 week High-------------------------- = Bucket .10The "- 0.0000001" prevents a stock which closed at its high from getting into a bucket of its own.3. Replace the RS sort with "Sort descending by bucket, breaking ties with RS".Would this accomplish you goal, also?FF
%52wkHi Bucket85-100% 670-85% 555-70% 440-55% 325-40% 210-25% 1 0-10% 0
Current Price------------- + 0.04999952 week High------------------------ = Bucket .15
Current Price------------- - 0.000000152 week High-------------------------- = Bucket .10
<<<Well, how about pointing me in a good direction to try out?>>>I know what I'm about to mention has been talked about before, but it's amazing what you can recall when you're at the gym working out and "turning off" your brain.These screen are more volitale than the indexes, so why not key the stop losses to the index? If the index is flat or rising, set the stop loss at 20-30% of highest price since purchase. If the index is falling, add that fall to the stop loss, i.e., index down 10% then stop loss of 30-40% instead of 20-30%. To avoid numerous recalculations, maybe this stop loss could be reconfigured once a week after the Friday close. JLC
As an individual about to go into retirement, my greatest fear is an October '87 type drop followed by a '72 to '86 flat market.Do you mean a '72 to '82 flat market? The DJIA roughly doubled between '72 & '86, not including dividends (which were very significant). And the fear you're concerned about distills to the biggest 1-day drop in history, followed by the worst 2-year market ('72-'74) in the last 70 years. Kinda double counting. Frankly if these kind of things concern you (and it is not unreasonable that they do), you should think twice about the more aggressive MI screens.--Terry
I wonder if some of the difficulty might be that you were often being stopped out when the whole market was dropping, and the scheme might work better if you were triggering the stop when the % difference between the stock and a common average (S&P?) dropped.Some prior discussion to spark your imagination on exit/stop loss stratagies now that there seems to be enough data (?) to possibly backtest some of these ideas.Best Regards,- MikePS - Note below: 1-3-99 – Sparfarkle – post 13420. Sparfarkles correlation to market stop loss comments introduced by 1-3-99 - AhabIshmael - Post 13416 on the MI board. Perhaps its backtestable now.**********************************************************************Exit Strategies, Discussions (Some exit strategies based on personal criteria beyond the mechanics of the screens, others are suggestions to add certain criteria to screens for exiting purposes) 12-25-98 - This subject keeps coming up on different boards. I've been a MF follower for almost 2 years. Used to have much thought on this subject, too. NOW, I'VE COME TO A PERSONAL CONCLUSION: Rayvt – post 12562http://boards.fool.com/registered/Message.asp?id=1030013003075015&sort=id12-26-98 – WADDAFOOL RESPONDS TO - “ONE ISSUE I WANT TO BRING UP IS SHOULD WE BE IMPLORING AN EXIT STRATEGY ON ONE OR ALL OF THESE SCREENS.” - This is exactly the point I tried to make a months or 2 ago but didn't get much of a response. A yearly hold seems arbitrary and even though bactested results have been positive, having a system to dump losers earlier could easily boost performance. WaddaFool – post 12617http://boards.fool.com/registered/Message.asp?id=1030013003075027&sort=id12-26-98 - However, if you are going to say that you sell if another set of criteria ISN'T MET YOU NEED TO KNOW EXACTLY WHEN THAT CRITERIA ISN'T MET. MoeBruin- post 12625http://boards.fool.com/registered/Message.asp?id=1030013003075028&sort=id12-26-98 - That's where I DEVELOPED ONE EARLY EXIT RULE THAT I RARELY, YET SOMETIMES, USE WITH MIKEYPORT. It is if a stock declines within 1 month of original purchase (assuming you use monthly trading like I do) Sux2BeU – post 12628http://boards.fool.com/registered/Message.asp?id=1030013003075030&sort=id12-26-98 - As to an exit strategy, I'VE MORE OR LESS STUMBLED INTO SELLING ONLY IF SOMETHING TRULY DIRE OCCURS, LIKE ACCOUNT IRREGULARITIES, A DOWNTURN IN EARNINGS. Gamxwell – post 12662http://boards.fool.com/registered/Message.asp?id=1030013003075032&sort=id12-28-98- Exit Strategies to the various screens - WHAT MAKES SENSE IS TO SELL IF THE FUNDAMENTALS OF THE STOCK HAVE CHANGED OR IF THERE ARE BETTER STOCKS TO PURCHASE. This is the beauty of mechanical screens. PKPet- Post 12796http://boards.fool.com/Registered/Message.asp?id=1030013003131000&sort=postdate12-26-98- THE PROBLEM IS YOU REAL NEED WEEKLY STATS TO VERIFY ANY TYPE OF EXIT STRATEGY. Quarterly definitely doesn't due and even monthly is weak. The reason is simple. If you … MoeBruin- post 12625http://boards.fool.com/Message.asp?id=1030013003075028&sort=postdate12-28-98 - TECHNICAL ANALYSIS TO DETERMINE WHEN TO EXIT IS NOT PERFECT. I've used MACD to get out of positions and I've gotten out too early and often not early enough.Factors that affect the price of stock are not often quantitative. For example 3 months ago LU was at 65 now it is at @ 110 and the fundamentals of the company are unchanged. – PKPet – post 12796http://boards.fool.com/Message.asp?id=1030013003131000&sort=postdate12-30-98 - As for the stop loss rule though, I HIGHLY RECOMMEND AGAINST A 15% STOP LOSS RULE--that's too tight for some of your screens. Screens like PEG and Spark routinely see periods where they lose 15% only to come rocketing back. – sparfarkle – post 13063http://boards.fool.com/Message.asp?id=1030013003179001&sort=postdate1-2-99 - When to exit a stock? Is it better to hold what may become a 700 bagger or cut your losses if your beloved goes in the tank? I've taken the PEG stocks from '87 to the present and the Spark5 since '89 and looked at how they behaved during the recommended one year holding period. VIRTUALLY ALL STOCKS WILL DROP BELOW THEIR ORIGINAL PURCHASE PRICE SOME TIME DURING THE YEAR. Only 4 out of the 85 never went below the purchase price. Even the best (annual gains greater than 200 percent) dropped as much as 10 percent- dmiddlebrook – post 13275http://boards.fool.com/Message.asp?id=1030013003220000&sort=postdate1-3-99 - Not quite as simple, but perhaps a useful refinement, WOULD BE TO SELL WHEN THE NEGATIVE SPREAD BETWEEN THE STOCK AND SPDRS BOUGHT ON THE SAME DAY EXCEEDS SOME THRESHOLD. More complex, but – continued – Jnelson – post 13443http://boards.fool.com/Message.asp?id=1030013003220005&sort=postdate1-3-99 - The market behavior this year was very atypical. It had the shortest bear market in 40 years AND I AGREE THAT JUMPING OUT IN THE LATE SUMMER/EARLY FALL DROP COULD HAVE CAUSED MISSING BIG GAINS. To see if this would have been the case I compared two approaches. The first is buy and hold for the full year. The second used the 20 % stop loss exit strategy. – dmiddlebrook – post 13425http://boards.fool.com/Message.asp?id=1030013003220004&sort=postdate1-3-99 - In your case, you've defined an intelligent exit strategy that correlates the market's behavior with a stock's behavior and calls for exit only in an extreme circumstance--based on a MECHANICAL rule. – Sparfarkle – post 13420http://boards.fool.com/Message.asp?id=1030013003179004&sort=postdate1-3-99 - I'VE NEVER YET SEEN ANY KIND OF MECHANICAL SELL-STOP THAT SEEMED TO DO MORE GOOD THAN HARM. One problem is that, - continued – Rayvt – 13528http://boards.fool.com/Message.asp?id=1030013003075040&sort=postdate1-3-99 - One could eliminate this objection to a stop loss (i.e the objection that the stock is merely following a downturn in the market) by setting up an Excel program which yields the stop point at, say, 75% of the purchase price PLUS a factor that takes into account the movement of the market as-a-whole. For example, set the stop-loss at STOP = (.75 + z)(Purchase Price)where z = % change in the SPY since the date of purchase. - AhabIshmael - Post 13416http://boards.fool.com/Message.asp?id=1030013003179003&sort=id1-4-99 - I must admit on these ocassions I get emotional but WHAT I DO IS LOOK AT THE SCREEN AND IF I THINK THE DROP MIGHT HAVE KNOCKED IT OFF THE SCREEN THEN I BUY THE NEXT STOCK ON THE LIST. If I think it will stay on the screen I don't. – MoeBruin – Post 13609http://boards.fool.com/Message.asp?id=1030013003279001&sort=postdate1-14-99 – OUTSTANDING - IT GOT ME TO THINKING ABOUT THESE SCREENS AND HOW WELL THEY WOULD MOVE US OUT OF A POSITION AS SENTIMENT CHANGES. Since we use Relative Strength as a measure of momentum, how would a near term drop in price be reflected on the screens. I know, over time, these screens bactest well. Still, there is the fear of sticking with the screen when all other evidence says, "Get Out!". – ProphetWise – Post 14639http://boards.fool.com/Message.asp?id=1030013003509000&sort=id1-22-99 - BEFORE TRYING TO DEVELOP A "SYSTEM", YOU MIGHT FIND IT PROFITABLE TO DISTRIBUTE ON UPSIDE AND BUY TO DECLINE. Nevertheless, I would encourage you to start looking at historical numbers of price movement, volume, and number of days for rallys and declines. Standard deviations for price ranges at given levels of volume, over specified time intervals, is very helpful. – nowise – post 15328http://18.104.22.168/registered/Message.asp?id=1030013003623001&sort=postdate1-22-99 - THERE HAS YET TO BE ANY PROVEN EXIT STRATEGY THAT HAS CONSISTENTLY WORKED. While I have been trying with better success lately to pick non-screened stocks, one of the reasons I like screened stocks is because I learned the number one enemy of successfull stock investing is YOURSELF. - Mcernick – post 15329http://22.214.171.124/registered/Message.asp?id=1030013003623002&sort=postdate3-5-99 -* NOTE*: NOT SO MUCH AN EXIT STRATEGY AS OPPOSED TO AN ENTRANCE STRATEGY ON A DECLINING STOCK YOU EVENTUALLY WANT TO BUY. I've dropped just about all my use of stops except for one method: A trailing buy-stop-market about 10% above the price of a down-trending volatile stock that I want to buy---but only when it starts a strong upward move. - Rayvt - Post 18711http://126.96.36.199/registered/Message.asp?id=1030013004232006&sort=postdate4-12-99 - WHEN I BUY A STOCK BASED ON THE SCREENS I HOLD ON TO IT FOR SIX MONTHS NO MATTER WHAT. If it's off the screens at six months I sell. If not, I reevaluate again once every three months, but not in between. Each stock may have a different buy and sell anniversary. I normally let winners run without rebalancing. Elann - Post 23339http://boards.fool.com/registered/Message.asp?id=1030013004925007&sort=postdate7/8/99 - I took PEG4 Jan-Jul semi, and checked in April and October for losing stocks. If any stock was a loser, I switched for the highest ranked Apr or Oct PEG4 to substitute. Here's the results: Out of 96 Jan/Jul stocks picked in 87-98, 15 were losers after three months. A remarkable 15.6% loser rate!For the 15 substitutions, the remaining 3 months had an average return of 2.8%. Then I checked how the 15 original stocks fared in months 4-6. Their return was 22.1%Conclusion, premature switching stinks and PEG4 stocks that are down, should be persued. Perhaps buying the loser works in this case. - klouche - Post 30744http://boards.fool.com/Message.asp?id=1030013006215004&sort=username8/8/99 - I use it with my PEG screen. Before I get flamed for not being truly mechanical, here's my reasoning...(probably not very reasonable, but here goes) I DROP A STOCK THAT DROPS MORE THAN 20% AND REPLACE IT WITH THE TOP STOCK IN THE NEW PEG SCREEN THAT WEEK. My reasoning is that the stock drops like that because it missed (or large investors assume it will miss) the growth estimates the PEG screen is based on.This way I avoid the minimum of one dog in the screen that seems to pop up every quarter and drop a minimum of 30%. - davidMN - Post 33810http://boards.fool.com/Message.asp?id=1030013006720004&sort=id8/14/99 - IIRC, we have been talking and looking for a sell-stop strategy here in the Workshop for about 3 years now. AND SO FAR, NOBODY HAS COME UP WITH ONE THAT ACTUALLY WORKED WHEN TESTED RIGOROUSLY. Not that this proves that there isn't one, just that if there is it will be devilishly hard to find.- Rayvt - Post 34371http://boards.fool.com/Message.asp?id=1030013006779014&sort=postdate8/23/99 - Conclusion: There is a sweet spot around a 30% Trailing Stop Loss where you can squeeze out 1.5% to 2.0% per six months or 3.6% to 5.0% compounded over a single 12 month period. Thus, using a 30% stop loss in the PEG5 Strategy may boost your return 3-5%. Avg. # Add. 6 Month Arith. Weighted # of Stop No Stop Days Add. Average of AverageStop% Stops Ret. Average to Stop Returns Combined Stops 1 755 0.90% 0.00% 1 0.12% 1.02% 1.02%2 755 1.30% 0.00% 3 0.28% 1.58% 1.58%3 755 1.58% 0.00% 5 0.49% 2.07% 2.07%4 755 1.89% 0.00% 7 0.74% 2.63% 2.63%5 755 2.36% 0.00% 11 1.08% 3.44% 3.44%6 755 2.34% 0.00% 14 1.41% 3.75% 3.75%7 754 2.70% 44.82% 19 1.89% 4.59% 4.64%8 753 2.79% 22.95% 23 2.30% 5.09% 5.14%9 753 2.88% 22.95% 28 2.73% 5.61% 5.66%10 750 3.47% 49.13% 34 3.32% 6.79% 7.07% - kuperman - Post 35214http://boards.fool.com/Message.asp?id=1030013006937010&sort=id9/13/99 - Rayvt has written that half the gains belong to the market, and she usually takes her share. I like the way this sets the frame so I don't panic over declines from recent highs. My plan is built around this vision.My proposal: With some regularity, calculate my total gain from this stock. When the total gain declines by 25% from its peak, take another 25% off the table by selling. Restart the mechanical strategies for increasing and decreasing stake at this point.- WilliamLipp - Post 37560http://boards.fool.com/Message.asp?id=1030013007260000&sort=id9/23/99 - FIRST OFF, DROP ANY NOTION OF BASING ANYTHING ON YOUR "TOTAL GAIN". [[ small digression into economics here.... When it comes to making a decision, your gain is irrelevant. It is based on your purchase price, which is a sunk cost. It is always a fallacy to base decisions on sunk costs. You must always and only make decisions based on the CURRENT STATE OF AFFAIRS. Your purchase price only determines how much of a profit you end up with. ]] - Rayvt - Post 38788http://boards.fool.com/Message.asp?id=1030013007260006&sort=postdate10/17/99 - I BELIEVE THAT DROPPING A STOCK BECAUSE OF A FIXED % LOSS IS A POOR WAY TO GO. It is not taking the market or the behavior of the screen into effect.TAKE THE FOLLOWING 2 EXAMPLES.Screen AStock 1 +80%Stock 2 +30%Stock 3 +20% Stock 4 +20%Stock 5 +5%S&P500 +15%Screen average = 31%Stock 5 is underperforming the screen by over 20%, the 2nd worst performer in the screen by over 10%, and the S&P500. I think it should be sold.Screen BStock 1 +25%Stock 2 +5%Stock 3 -10%Stock 4 -20%Stock 5 -25%S&P500 -15%Screen average = -8%Stock 5 is underperforming the S&P500, but is within 20% of the screen average, and within 10% of the next worst performer. The market is obviously in a downturn and there is no need to panic about Stock 5.Peter, is there anyway to test this? - piranha1 - Post 41825http://boards.fool.com/Message.asp?id=1030013007894008&sort=postdate10/21/99 - Rocket Dismount Report - My scheme was to keep half the profits by taking money off the table as fast as Lady Market did - ANYTIME THE GAINS DROPPED BY 25% I WOULD SELL ENOUGH TO TAKE A MATCHING 25%, THEN RESTART. Next time I'll set the sell points based on 25% of the position value instead of 25% of the profits. I've set my fifth sell point based on this revision. - WilliamLipp - Post 42234http://boards.fool.com/Message.asp?id=1030013007994000&sort=postdate10/29/99 - Summary: Purist MIs will phthphttt these results because they're affected by emotional decisions. Excellent!! You're right!! I believe the screens' performance can be improved with some rational, emotional decisions based on knowledge of the general market and the individual stocks picked by the screens. I believe sometimes you have to stop the bleeding or lock in gains when the market is not going straight up. That said, if you screw up, ain't no blaming the screens. If you're willing to take the chance, good. If you're not, fine. For those interested , these are my ideas: - Any RSIBD stock up over 50% becomes a potential for sale pre-turnover date, especially if it is not the #1 or #2 stock in the screen. Yes, I've had an itchy trigger finger on JDSU since Tuesday. It's up 30% since then. Lucky I didn't pull the trigger.- If an RSIBD stock was up over 30% and gets crushed on a strong market down day more than the market, I will probably sell it the next day. This is because in a truly correcting market, the leaders break down faster.- If an RP4 stock seems to have matched or outrun what I think is its reasonable appreciation potential for the year, it gets sold - e.g. Caterpillar. I'm happy with a 31% contribution to the RP4 screen from old CAT. Am I going to really care if it goes to 35%? No. Am I going to kick the @##@ out of myself if it crashed or went back to 0-5 without selling it? Yes. So it gets sold.A 4 stock RSIBD monthly is "clearly" better than 5. Evidence? None empirical, sorry to say. But anecdotally, I've seen almost every month here the #5 stock is a former high flyer on its way out (leaving you holding the bag), or a weak sister that has topped out and ends up getting pushed off in a month with no gain, or a loss. This kept me out of VISX after locking in a 30% gain, protecting me from a subsequent loss. - Mwynne - Post 43312http://boards.fool.com/Message.asp?id=1030013008167000&sort=postdate10/30/99 - From a guy who's been tinkering around with RSIBD selling strategies with my own real bucks on the line here's what I've settled on. (Not backtested he he ;-))1. Watch earnings...sell immediately on disappointments-- The Whisper number is the one that matters with RS stocks2. Volatility is a fact of life particularly around earnings release dates-- Just ride with it3. Graphs are helpful-- The basic line should always point to the upper right!-- Big parabolas are bad...when you see a nice arc going down that peaked over a month to 45 days ago...time to sell4. Downward price trends are pretty much meaningless unless (assuming no bad news) they last at least a month5. Using the above all you really accomplish is getting out a bit earlier than you otherwise would of.Guess what? You're no longer MIing!P.S. After a year of the above here's what I'll probably do in 2000 … continued in post … - KBGlenn - Post 43324http://boards.fool.com/Message.asp?id=1030013008139008&sort=postdate11/3/1999 - It sounds like you bought TMF emotional picks instead of your own emotional picks otherwise you probably wouldn't be asking this question. For my emotional port I use TMF techniques to buy my own stocks. My suggestion look at why you bought the stock, whats your goal with the stock and is it performing if not sell. Next time you go emotional write down why your buying the stock and what will cause you to sell the stock before you buy the stock. This will keep you disciplined in your emotional picks. - MoeBruin - Post 43807http://boards.fool.com/Message.asp?id=1030013008238001&sort=postdate11/3/99 - Maybe we could use existing data to find whether you should sell part of your longer-hold screens at a month or three or whatever, but what are you doing? It's not really an exit strategy so much as inventing a new monthly screen with a strong component of "keep what was there before if there's no reason to dump it". At it's base, it's still just a fixed-hold screen.So this brings me back to the beauty: the screen is the entrance strategy is the exit strategy. It's a floor wax and a dessert topping. - gritton - Post 43860http://boards.fool.com/Message.asp?id=1030013008240008&sort=postdate11/3/1999 - Why don't you look at Average True Range (ATR)?First you take the absolute maximum difference between:1) Today's Low - Yesterday's Close2) Today's High - Yesterday's Close3) Today's High - Today's LowThe absolute maximum difference is your Today's True Range.Next, calculate your Moving Average True Range. This is typically your 14-day average of each day's True Range.Next, you compute your Next Day's ATR Stop.Next Day's ATR Stop = Today's Low - ( Moving Average True Range * 1.382 ) - MGBloomfield - Post 43889http://boards.fool.com/Message.asp?id=1030013008240012&sort=postdate11/4/99 - NOTE: TMFElan requested later in the thread that the author of this post his results for the message board to examine, the author for whatever reason did not reply.The only exit strategy I recommend is a sell if a stock closes more the -15% from the buy price and the overall market is not in a serious downtrend. Many MI stock picks plummet and if you can remove them from your portfolio, your average can go up.I backtested by hand all PEG-5 for 12 years doing this system and the average went from 46.7% per year to about 51.5% Ok, not stellar but it did work and I had only a few sells (4 I think) on stocks that went back up. I got rid of several that would have lost 30% or more. Also, the overall yearly return was more even and I had no losing years.- ttrible - Post 44051http://boards.fool.com/Message.asp?id=1030013008240027&sort=id11/10/99 - Using the standard approach, each $1 grew to $12.06 (it was a good 47 months).The sell strategies I looked at were:a) sell the weakest and no reinvestmentb) sell the weakest and reinvest in the strongestc) sell the weakest and move to the second weakestd) sell the strongest and no reinvestmente) sell the strongest and reinvest in the 2nd strongestf) sell the strongest and reinvest in the weakestOf the six variations only two performed better than the standard, e and f.E resulted in each $1 growing to $13.11F resulted in each $1 growing to $13.03These differences may not be statistically significant, but it does indicate that it doesn't pay to sell the weakest performer in your portfolio part way through the holding period. - DrBob2 - Post 44992http://boards.fool.com/Message.asp?id=1030013008411000&sort=postdateA new technique for selling stocks near their peak High-flying growth names need extra wiggle room as they zigzag to new heights. A stop-loss method cooked up by Web Community members makes it simpler to sell before zoom turns into kaboom. - Jon D Markman - Money Central Articlehttp://moneycentral.msn.com/articles/invest/models/4858.asp?special=msn11/17/99 - Along those lines, I have been toying with the idea of doing small tweaks of my quarterly screens (buy highest of PEG26 and PEG13 not already owned; highest RS13 and RS26 not already owned both with preference to overlaps). Since I keep 10 stocks that make up these 2 screens averaging $5,000 per stock, I thought about dropping the single stock in each screen that has the lowest ranking (or absent from all rankings) and buying the highest not already owned. I figure by limiting it to 1 stock each screen per month I keep the incremental commissions down to 4 trades ($32/mo times 8 months for $256/year or max increase of 0.5%/year until I transfer to Brown soon) when it is not a quarterly transaction date. This is somewhat different than dropping the poorest performer. I prefer to let the screen tell me what to trade rather than price performance since my purchase. - buckaroobonzai - Post 45888http://boards.fool.com/Message.asp?id=1030013008532007&sort=postdate11/24/99 - Please notice I mentioned...sustained selling. There's a difference between knee jerk reactions usually lasting a couple of days and a continual one lasting a couple of weeks (or more). And again if anyone has paid attention to earlier opines of mine ref this subject, I only expect to exit a screen a bit earlier than the screen would have done eventually. In addition, there is no doubt this is a judgement call. Corollary to the mantra...you can always buy # 6. - KBGlenn - Post 47018http://boards.fool.com/Message.asp?id=1030013008689027&sort=postdate11/22/99 - When it comes to monthly screens, I agree w/Gritton, who said "the screen is the stop." But on longer time frames, particularly annual holds, I wanted to explore a different approach.What I looked at was selling a stock when it is no longer on the screen rankings for double the amount of positions held. As an example:1. Buy KEYSTONE Annual Five on Jan 1.2. Check rankings every week. When any of the five stocks purchased falls out of the Keystone 10, sell it.3. Use the proceeds from the sale to purchase the highest ranked Keystone stock that you don't already hold. I used 10 positions rather than 5 to minimize trades and whipsaws. Some stocks will drop to six or seventh position one week, only to pop back into the top five the next. I didn't want to sell until they fell off the top 10. - lafong - Post 46712http://boards.fool.com/Message.asp?id=1030013008672000&sort=postdate11-23-99 - I have a couple of points that keep nagging at me, with regard to early exit strategies:1) There is much research that says that people make inferior decisions when they stray from a set of proven mechanical rules. (JOS quotes some of it.)2) Ken Fisher echos this, with many discussions about regret and pride. "Anyone who deploys pride and regret mechanisms hurts themselves in markets and keeps themselves from becoming the best they can be. Your goal is the reverse, to shun pride and accumulate regret."3) Newbies always try want to sell at the first whiff of a profit. If a neophyte thinks he has discovered something that experienced pros have overlooked, his idea is almost certainly wrong.4) My dad giving me many lectures about being "too clever by half". And pointing out how I had cleverly shot myself in the foot. Again.5) You can work yourself to death worrying about trying to find the "right" time & conditions to exit early.- Rayvt - Post 46758http://boards.fool.com/Message.asp?id=1030013008665002&sort=postdate11/23/99 - You can do somewhat of a backtest on Jamie Gritton's bullet backtester, which allows you to rebalance when a a stock drops down to a certain rank. The comparison to a standard backtest of the same stock might tell you something about whether it pays to hold past the rebalance date.From what I've seen in playing around, it decreases returns on the stocks that were highly ranked when purchased, but could increase the returns on ranks that haven't peaked yet. Here's a sample:Monthly=======RSW rank stnd. buy ascending, sell #5 desc.======== ===== =========================== 1 32.5% 32.2% 2 40.4% 26.0% 3 18.5% 33.2% 4 4.1% 41.4%- Geocorona - Post 46860http://boards.fool.com/Message.asp?id=1030013008684001&sort=postdate11/23/99 - I have been following, for some time, a simple sell rule (with monthly strategies) that could have got you out of SPGLA early. I sell any stock whose price declines three days during which the market (NASDAQ) is up. - Damakan - Post 46903http://boards.fool.com/Message.asp?id=1030013008689005&sort=postdate11/26/99 - I look at screens as imperfect tools. I expect them to give me some winners and some losers. So, when I buy a screen, I immediately place a stop at 12% below the buy price on each of the stocks. I review this each week, moving the stop to 12% below the stock's high since purchase. - jhindorff - Post 47173http://boards.fool.com/Message.asp?id=1030013008689032&sort=postdate11/27/99 - I will move a stock to my anti-whipsaw port if it meets the following conditions:1. Timeliness is 1 (or 2 for my PEG port).2. Gain over the month.3. IBD RS = 98, 994. IBD A/D = A, B5. As an alternative to 3, 4, if there is space, it will be moved there if the stock is still on an RS screen.I don't know how this will work, and I am trying to define a more mechanical criteria, but this is the general idea. - FoolishlyFree - Post 47262http://boards.fool.com/Message.asp?id=1030013008761000&sort=postdate11/28/99 - Early Returns vs. Later Returns While on a trip this weekend while others drove, I took another look at Peter's data to see if early returns can predict future returns. Setup: Take the returns for 12 screens including Plow, Peg, Keystone, Spark, RSxx, use all 10 positions. Approximately 16,000 returns were considered. Sort the 16,000 by 1 month performance and segment into 10 decile ranges called 0% to 90%. Each range has 1,600 returns. The goal is to see if early performance can be used to predict future performance.One Month Returns as a Predictor of Future Return.Decile Avg 1 Avg 3 Avg 6 Avg 12 Improve 1:120% 0.81 0.85 0.92 0.99 22%10% 0.91 0.96 1.02 1.12 22%20% 0.95 0.98 1.04 1.14 20%30% 0.98 1.02 1.07 1.19 21%40% 1.01 1.06 1.13 1.25 24%50% 1.03 1.08 1.15 1.27 23%60% 1.06 1.11 1.17 1.31 24%70% 1.09 1.12 1.22 1.34 23%80% 1.14 1.20 1.26 1.43 26%90% 1.25 1.31 1.38 1.57 26%- klouche - Post 47376http://boards.fool.com/Message.asp?id=1030013008785000&sort=postdate12/3/99 - All that I can do is to provide my own particular strategy, for what it is worth. I look at the particular screen and its formula for each stock in my portfolio. If a stock was timeliness 1 when I purchased it and now it is timeliness 3 it might be a good idea to sell. If a stock had EPS 99 and RS 99 when purchased and now is 99/58 or perhaps 89/75, it would probably be a good candidate for selling. On the other hand my EMC, even if not on the screens, has timeliness 1 and 98/89 besides having many other factors going for it and as far as I am concerned it is a keeper.I purchase the stocks exclusively by the screens and I sell them on an individual basis, but still referring to the criteria for that screen. Hope this helps. - bobblaw - Post 48070http://boards.fool.com/Message.asp?id=1030013008882017&sort=postdate12/23/99 - The point of this exercise is to see if selling PEG stocks using a manual, trailing stop-loss point was better or worse than a time specific sell period. The rules I chose reflect my own personal trading style and may not be what anyone else is looking for. I don't really like using manual stops and I check my stock prices every morning anyway. For this exercise, I want a price dip to be a signal to watch the stock very closely, not to actually sell it. My sell rules are: . 1) sell the stock the next day if it closes at -20% of the buy price . 2) sell the stock the next day if it closes at -12% of its highest closeprice since it was bought . 2a) ... but only if that -12% price is above the buy price - ttribble - Post 51509http://boards.fool.com/Message.asp?id=1030013009418000&sort=id12/25/99 - What works for me may not work for you, but here is my approach.I look back at the last several years of stock prices for a particular stock. Then I calculate the percent that major retracements have lowered the stock price. Let's say four retracements have occured over the last five years of say 25-30%. I place my stop at the far end. If it blows through that percent, it is probably going to go further down. I usually make a point of staying away from common stop points (chart lines, moving averages, etc.), i.e. where funds or traders see other traders placing their stops, whether mental or actual. Distribution will occur to take out those stops, almost inevitably as strong hands attempt to push the stock through the traditional accumulation-distribution cycle over and over.- Sharoncrayne - Post 51844http://boards.fool.com/Message.asp?id=1030013009410007&sort=id12/24/99 - I look at each stock every morning (I use the MSN portfolio manager for this and look for large movements). I think an automatic selling system could work if you update the stops twice a week but I would prefer to check the stocks and place stops manually until I get really comfortable letting the system run by itself. For example, I have NO stops in place right now. After 1/1/00 (assuming civilization survives!), I will check my stocks everyday. If a stock closes under my stop point, I'll look to manually sell it in the next day or two. I've looked at enough charts to know that most stocks bounce up a little. If the stock plummets, the stop won't help you anyway (it'll leave your sell price behind!). - ttribble - Post 51697http://boards.fool.com/Message.asp?id=1030013009418006&sort=id12/26/99 - I want to propose a fairly mechaincal model of selling "the losers". If this has been presented before, I apologize in advance...perhaps I've missed its posting.For the purposes of illustration, I'll look at the returns of a straight-forward screen: take the Top-500 Market Cap stocks, sort by high 52 wk TR, and then select the Top-3 of these.The return for a 6 months hold is approximately 23%. This beats the overall market, but is paltry by MI Board standards.The question is: can you identify early on which stock(s) of the 3 will underperform by the 6 month rebalance time?If you sort the 3 picks by their 1 month performance, and then examine their subsequent performance over the next 5 months, here is what you see (the best performing stock at 1 month is labelled "1st"): gain in months 2-61st 10%2nd 12%3rd 4%It appears that the worst performing stock declares itself (in this screen) early...meaning it continues to be the underperformer.If, at 1 month, you sell the worst performing stock of the 3, and reinvest it in one of the other 2, you increase your annual gains from 23% to 32%.- Alevine - Post http://boards.fool.com/Message.asp?id=1030013009482000&sort=id12/27/99 - Just a couple o thoughts ref cutting losses and rebalancing1. Probably doesn't make a lot of sense for monthly (and probably quarterly) screens as the screen itself will mitigate losses soon enough. (Exception...substantive earnings/balance sheet disappointments.)2. The key item that needs to be discovered is what is an appropriate stop/loss point. O'Niel recommends 8%, others recommend 10%. But in today's volatile markets that happens with great frequency to great stocks. Does anyone even blink anymore when a tech stock tanks 10-15% or more? I'm thinking the appropriate point is somewhere in the 20s.3. Assuming you're not trying to time the market and can ride out the ugly times, the stop/loss should be compared with the overal market. For example, assuming your stop loss point is 20%, if the market is down 10% and a given stock is down 20%, the stop/loss would not be applied. (20-10 is only a 10% loss) The intent of the above being that you're mitigating losses in response to poor stock performance and not poor stock market performance.4. Another issue to figure out is what to replace the sold stock with. My thoughts are that if you're buying a screen for specific characteristics, then you should replace the sold stock with another stock in that screen. If not, you're not really buying the screen anymore. Return data by screen position would be particularly helpful if a cherry picking approach is taken. Position # 2 anyone? :-)5. Finally, another issue to deal with would be the susequent tanking of a stock after it reached a high later in the year. This issue is more difficult to deal with, certainly more difficult to track, and depending on where you were in the holding period maybe not even worth the effort. Ultimately I think the basic same rules noted above would apply, but whether or not to invoke them would be more of a judgement call.6. I believe a carefully crafted mechanical stop/loss regimen would be of most significance in enhancing returns for semi and annual holds. The screens are a rearward looking mirror not a forward looking window, and from time to time they'll select a lemon. In a semi or yearly screen you may get a lot more lemon juice than you care for. For the life of me I can't understand why anyone would have held OMI for almost a 50% loss in one of the greatest stock market years ever. There was a clear message there, and the pricing told it. Why ignore it other than you were chanting the mantra..."stick to the screen" over and over. Clearly any application of the above would involve more work on the part of the investor, is difficult to backtest and could easily lead one to emotional investing. The thing that keeps sticking in my mind is that just about every book I've read on momentum investing had a component of loss mitigation in it. Unfortunately the rules suggested were based on markets that clearly were not the same type we find ourselves in today.- KBGlenn - Post 51967http://boards.fool.com/Message.asp?id=1030013009482005&sort=id12/27/99 - I for one will read anything and everything published on the board about reasonable approaches to trimming losses....The key item that needs to be discovered is what is an appropriate stop/loss point. O'Niel recommends 8%, others recommend 10%.....The method I've suggested in the post entitled Interim Rebalancing avoids using a pre-ordained fall in stock price (or percentage of purchase price). To properly execute such a technique (sell when stock falls 20%, for example) really requires daily monitoring of your portfolio. Even with a limit order in place, you need to check to see if the sale took place in order to know when to re-use the resultant cash.The Interim Rebalance method picks one day (say, 1 month after original purchase) and sells the poorest performing stock (of a 3 stock screen....that's all I've looked at so far).And....although it may be anathema to some, this partial rebalancing technique often has you selling winners. Even if all three picks have positive gains at the end of 1 month, you sell the poorest performer.- Alevine - Post 51970http://boards.fool.com/Message.asp?id=1030013009482006&sort=id12/28/99 - During October and November, I was holding SRT when it lost about 50% of its value. At that point, I didn't have a rule for risk management in place, so I rode SRT down. :(Then I took my trading data from 1999 and examined all the stocks I held during the year that got at least 5% below their monthly opening price during the time I held them.What I found was that if I had sold each stock when it had a daily close 5% or more below its price at the beginning of the month, 5 times out of 6 I would have avoided a larger loss. One time in 6, I would have gotten myself out of a position that would've been a winner if I had hung on.So now I have a rule in place that says, "If a stock closes a day 5% or more below its price at the beginning of the month, I enter an order to sell at the market the next trading day."- tbarron - Post 52050http://boards.fool.com/Message.asp?id=1030013009482011&sort=id12/30/99 - The approach I took was to compare the returns at the end of the first quarter, and sell the worst performing stock and add the proceeds of the sale equally to the remaining four positions. At the end of the second quarter, sell the worst performing of the remaining stocks and add the proceeds to the remaining three positions. Then hold the remaining three stocks to year end. By doing this, the returns of the two screens improved dramatically. Spark 5 went from a return of 91.1 % to a return of 110.7% with interim rebalancing as shown, while PEG 5 changed from a return of 27.8% to a return of 40.0% with interim rebalancing. - dan2 - Post 52481http://boards.fool.com/Message.asp?id=1030013009482014&sort=id12/31/99 - The following tables show various Interim Rebalancing Periods for the Top-3 Spark picks, using monthly data from Jan-86 through Dec-98. Theses returns are not annualized.In this first table, the Top-3 Spark picks are re-evaluated at 1 month. The stocks are then ranked (1 is the best performer), and I show their average subsequent returns (geometric means)until the quarterly rebalance time (months 2--3), the semi-annual rebalance time (months 2--6), and the annual rebalance time (months 2--12).Using a 1-month hold, Interim Rebalancing appears to provide little (at best) advantage over a straight hold:1 month Interim Rebalance 2--3 2--6 2--121 2.6% 9.7% 18.8%2 3.5% 8.9% 20.5%3 3.8% 9.9% 16.9%Here is a similar table for a rebalance/re-evaluation after a 3 month hold:3 month Interim Rebalance 4--6 4--121 6.6% 18.3%2 6.4% 13.4%3 5.6% 13.4%Here is the data for a 6-month rebalance:6 month Interim Rebalance 7--121 8.6%2 8.1%3 8.5%It appears the the technique of Interim Rebalance is less successful in the Spark-screen than in pure RS/Momentum screens.- Alevine - Post 52567http://boards.fool.com/Message.asp?id=1030013009609000&sort=id12/26/99 - From time to time, the question of cutting your losses on the Mechanical Screens arises, using such methods as selling when a holding falls a certain percentage below the purchase price, etc.These ideas have not been terribly well received, because they are hard to backtest....so no one knows for sure if they've worked in the past, let alone might work in the future; they smack of market-timing; they deviate from the mechanical models which are time-tested; and they are fairly intensive...meaning they require nearly daily recalculation of the stock's performance.Also, the question of what to do with proceeds from a stock you sell before the appointed rebalance date needs to be addressed.I want to propose a fairly mechaincal model of selling "the losers". It appears that the worst performing stock declares itself (in this screen) early...meaning it continues to be the underperformer.If, at 1 month, you sell the worst performing stock of the 3, and reinvest it in one of the other 2, you increase your annual gains from 23% to 32%.- Alevine - Post 51913http://boards.fool.com/Message.asp?id=1030013009482000&sort=postdate12/31/99 - In post number 52481, I made a mistake in the opening prices for the screens. When I looked at the monthly prices at the Yahoo site, I assumed that the opening prices shown were for the beginning of the month. In fact, they were the opening prices for the last day of the month. The corrected opening prices are shown below. This changes the returns as follows. Spark 5 went from a return of 91.1 % to a return of 110.7% with interim rebalancing as shown, while PEG 5 changed from a return of 27.8% to a return of 40.0% with interim rebalancing. My apologies for the confusion. Dan Note: See post for performance table.- dan2 - Post 52530http://boards.fool.com/Message.asp?id=103001300948201512/31/99 -Here are the tables that show the effects of Interim Rebalancing on the PEG-screen.As before, I looked at various holding periods, and then re-evaluated (ranked) the stocks by their performance during that period. The 1st table takes the Top-3 PEG picks, and ranks them by return after a 1 month hold. It then shows their subsequent return until the quarterly rebalancing period (2--3 months), semi-annual rebalancing period (2--6 months), and annual rebalancing period (2--12 months). The data is derived from monthly VL files (Jan-86 through Dec-98), and are the geometric means of the returns. This data is not annualized:1 month Interim Rebalance 2--3 2--6 2--121 3.5% 12.5% 19.6%2 6.2% 12.2% 16.7%3 1.7% 6.5% 11.3%3 month Interim Rebalance 4--6 4--121 5.0% 11.9%2 6.9% 12.3%3 8.1% 8.7%6 month Interim Rebalance 7--121 6.1%2 4.1%3 0.3%It appears that the technique of Interim Rebalancing is at least moderately useful when investing in the PEG screen.- Alevine - Post 52569http://boards.fool.com/Message.asp?id=1030013009610000&sort=id1/2/00 - I use a -20% (decline from the highest position that I owned the stock or what I payed for it), stop at the end of closing day and a -7% drop for a 4wk period.- SeriousMac - Post 52853http://boards.fool.com/Message.asp?id=1030013009644009&sort=id3/31/00 - But what I can do is to look at monthly cycles, and arrange to drop a stock that has a negative return for the first month following purchase, even if it is still on the screen and would ordinarily be held another month, to see what impact this may have on the long term CAGR, which will at least tell us whether there is any point pursuing the idea any further or not. So I have included a table below with some results for a small set of representative screens.To interpret these results, you need to know exactly what is being backtested. It turns out that I can only get useful results to compare to the Hold-Till-Drop case (which tends to hold stocks longer), because the tester, at least as currently written, could end up "buying back" any dropped stock which still happens to be ranked in the top 5 of the screen. But in the HTD scenario, stocks bumped from the bottom half of the screen (and at least some of those that may still be in the top 5) will remain bumped. So to gauge the results, we need to see runs for which a significant number of losers have been dropped, and then look for some resulting improvement in the return. Bumped stocks are always replaced (immediately) with the highest ranked unheld stock in the top 5 of the current monthly screen. (Hope I explained that right.) The table shows several screen in three variations: "Standard" 1-5, Hold-Till-Drop, and Hold-Till-Drop with stop loss. These are monthlies, positions 1-5: … see post for table ….Remember, this post is not about HTD: we just need to use the HTDresults as the reference, since this is a modified HTD run which drops"losers". So what do we find?Well, not very much, I'm afraid. First, the Peg results are prettymeaningless, because there just aren't enough "losers" being dropped. Evenwith PegRSW, we only found 10 losers to drop in the 14-year test run. And theshort-term Peg (Peg-13) simply turns over too quickly for losers to beidentified by the stop loss: typically they are already gone from the screenanyway. (We would probably see similar results for short term RS screens). So let's look at the rest:With RSW, we replaced 51 losers, and it looks like we gained 2 points in theCAGR, but this is not at all significant given the high GSD. The result forRS26 is similar, gaining 3 points (out of 57), but again the GSD is so highthat the "improvement" is surely just random chance.Keystone and Plow are normally run as semi's or annuals, but it is interestingto see what happens to them on these monthly runs. It turns out that for bothscreens, we actually hurt the CAGR by dropping the losers, but again weare only talking about a couple of points, probably not significant.This is only a preliminary quick look, but I have to say that these resultsare less than exciting, and they fail to show any significant benefit to beachieved with a stop-loss strategy. I believe this is consistent with theconclusions drawn from earlier looks at the same issue, although I have notactually seen those previous studies. But given the statistic I saw quotedthe other day, that something like 85% of our screen stocks will drop belowtheir purchase price at some time after buying them, I am not surprised.I have also tried applying the stop-loss only to more severe losers (say,20-30% loss). Results were no better.- FoolislyAl - Post 64780http://boards.fool.com/Message.asp?id=1030013011619000&sort=postdate4/8/00 - This Recent Unpleasantness is exactly why risk management must be your number one priority. You can't predict or foretell when the market will take a short, sharp dump. No one can. All the handwaving and handwringing in the world won't change this fact. So stop with trying to fight the last war. Whatever you come up with won't work for the next time.What counts is that you structure your portfolio so that even a 3-sigma event doesn't take you out of the game. For us, this essentially means don't use too much margin and don't put too much into options.Forget about trying to protect your profit. Focus on protecting your ass so that it doesn't get fried. Like all those poor sods who got the margin calls on Monday -- and got sold out on Tuesday.- Rayvt - Post 65762http://boards.fool.com/Message.asp?id=1030013011777016&sort=postdate4/8/00 - This is really an excellent point. I use a 12% stop strategy with monthly PEG stocks. I found this to be very effective in "culling the losers" so that I can redirect my money to the winners.But...on January 3, 2000, I found myself stopped out of all of my positions. That was not so bad in itself. What was bad was that I did not have a strategy to apply to the situation. Yeah....I had kind of a half-baked idea of what I would do, but it was not specific enough to keep me from panicking, and I made some really bad decisions. Moral of the story: A stop loss strategy that culls the losers demands a firm, well thought out strategy for the selloffs that take you out of all of your positions. They are two different types of events, and you need a strategy for both.- jhindorff - Post 65783http://boards.fool.com/Message.asp?id=1030013011777019&sort=postdate4/3/00 - It is more of a change of trend analysis than a stop loss. Basically in an uptrend a stock will make a series of higher highs and higher lows. If you look at a 1 year chart of VISX you'll see that it hit a high of a little over 100 at the beginning of August. It had essentially been in an uptrend until then and I'm guessing was still on many people's screens. It then dropped to below 80. I wouldn't have sold yet because this was still a higher low than the previos pullback. However, this becomes my stop point. You'll see that VISX tried to rally to its old highs twice and failed. This is a bad sign. It then dropped below the previous pullback. This is when I would have sold and been out long before it reached its terrible demise. VISX may have fallen off the screens at this point which is good for people if they follow monthly screens.With QCOM it pulled back from its highest high to 135 which became my stop point in early january. It rallied and failed twice at 165--bad sign since it didn't make a new high. I sold when it broke 135 in the middle of January. QCOM never sold off like I thought it might after that, however it did flounder for a couple of months as it gained strenght. I've been out of it until this past friday and in the meantime had at least a month of gains in another stock that was performing alot better during QCOM's floundering. Many people still had QCOM in their screen during that time. My stop loss is based on support points. Basically a correction low becomes my stop.So basically my strategy will get me out of a potential situation that might leave others in just because they follow the screen. Holding onto QCOM in this case wouldn't have been a bad thing but I made more money being out of it in that time. You'll notice that alot of the top stocks in the screens are showing some of the signs I described above. - eboller - Post 65061http://boards.fool.com/Message.asp?id=1030013011601020&sort=id4/4/00 - RayVT and kuperman indicated during talks at the conference that they will often sell when a stock goes down 30% below cost. Ray further reiterated his approach to closing out a position when his total gains have been cut in half. - StanelyManley - Post 65161http://boards.fool.com/Message.asp?id=1030013011601025&sort=id4/9/00 - I would sell if it went below a recent correction low. I would also not buy a stock that appeared in the screen if its current price was below a recent correction low. If you look at QQQ (nasdaq100) you can see the corrections I'm describing. They occurred on (dates approximate) Dec 1st, Jan 5, Jan 29, March 15th, april 2nd. As you can see all of these points except the last one have been higher than the last. Earlier this week I would have sold out when the stock undercut the March15th low. So I basically set a stop as the stock is going up under each successive- eboller - Post 65900http://boards.fool.com/Message.asp?id=1030013011601036&sort=id4/9/00 - EXCEPTIONAL Just because your stocks went down is no reason to alter a strategy.Just as getting lucky after the fact doesn't justify a bad decision, getting unlucky after the fact doesn't justify abandoning a good decision.For every story about a successful stop-loss, there is at least one story about getting whip-sawed, or getting taking advantage of by the market makers in a panic. I have a feeling that, like limit orders on fast moving stocks, a stop-loss might save you a little bit of money most of the time, but will cause you to lose a butt-load of money enough of the time to make it not worth your while. The classic example is ORCL in March '99, which dropped 25% on an earnings warning, but is up 400%+ since then.A proper exit strategy should be that your stocks no longer fit some fundamental criteria, probably the criteria you used to select them. Up or down doesn't mean anything. I might prematurely sell an MI pick if there was news of accounting irregularies or criminal activity, for example, but not because it is going down. As it turns out, that is exactly what our screens do. At the predesignated evaluation time, if the stock no longer fits the criteria it is sold. The more we see stop-loss strategies offer little or no benefit, the more I am convinced our current approach is the right way. Even if there is a stop-loss strategy that offers some incremental improvement, I for one would hesitate to use it because it would mean that I would have to constantly monitor 12-15 stocks, make an emergency sell, and then decide what to do with the money. Last Thurs-Tues, I was on vacation playing with my daughter. The increase in returns for a stop-loss strategy would have to be substantial to make me want to sacrifice my vacation to deal with a market-meltdown non-event.A fundamental rule of investing/gambling/probabilistic endeavor is that when you have an edge, a positive expectation, the more money you invest and the longer you keep it invested, the more money you will make. Any expert will tell you to ignore short-term fluxuations and stick to your plan, since if you have an edge you aren't trying to take advantage of short-term volatility, you are instead trying to defeat it. This of course has to be combined with RayVT's warnings about risk; you can't win if you've been knocked out of the game, so don't take a risk that will take you out of the game.Really, it's that simple. It's anecdotal, but last week, who came out ahead: those with stop-losses, or those that ignored their investments? In August-September '98, who came out ahead, those with stop-losses or those who ignored their investments?Stop worrying wring every last drop of gain out of an investment (because in the end it will cost you), and concentrate on the best risk-reward ratio, put as much money as you can into this good decision, and let time work for you.- GrandPoohbah - Post 65932http://boards.fool.com/Message.asp?id=1030013011795011&sort=postdate4/15/00 - I have been thinking the same thing myself, so did a little investigation. I pulled up the backtest data for the following screens: RS-26, RS-13, RS-O, PEG-13 monthly. Then I looked at various strategies of using cash (or treasuries) to balance out drawdowns.Three models:A) Always hold 20% of assets in cash.B) hold 20% in cash until a crash, then hold 0% cash for the next 12 months.(Basically invest more after a crash)C) hold 0% cash until a crash, then go to 20% margin for the next 12 months.The idea is that after a crash, it makes more sense to invest in the market. In all cases, I did not move the additional cash into the market until 1 month after the crash.Results: inconclusive. Through many variations of holding periods, numbers of stocks, screens, crash limits, and %cash/margin I did not find any outstanding discoveries. Most is as you would expect:Holding constant 20% cash always (Plan A) reduces variability. (I rebalance the cash position monthly)It also reduces gain.Moving to margin after a crash (Plan B) does increase CAGR gain slightly, and reduces GSD variability slightly, but it was not consistent over all screens. More effect on RS-13, less on PEG-13. Only a few percentage points. Probably luck.Bottom line: keeping a constant percentage cash does help reduce volatility. Throwing all into the market after a crash may not be the smartest move. After the next rebalance period, I will shift some cash into the market so that I am at a constant percentage.- DNeuman - Post 66614http://boards.fool.com/Message.asp?id=1030013012011003&sort=postdate4/18/00 - With volatile stocks like those selected by some of the MI screens, there's a pretty good chance that a sharp drop will trigger the stop-loss order, after which the price will rebound at least to a level that wouldn't have -- and later the stock will rose above its previous level. MI practitioners in general disapprove of anything that interferes with following the screen, and following the screen includes following the holding period.- TchrP - Post 66861http://boards.fool.com/Message.asp?id=1030013012057014&sort=postdate4/18/00 - I've read (in 3 or 4 different articles and books) that stops accumulate at certain levels and that market makers will manipulate the stock to get it to the trigger point. Having been a floor trader earlier in life, I can tell you that a lot of this goes on, much more than the public knows or realizes. Some of it is legit. If a floor trader wants to risk his/her own capital trying to push the market down to trigger stops, then that's fair, if unwelcome. But if the broker/market maker is also trading and either: (1) discloses those stops to traders who touch them off, profit, and kick back $$ to the broker/market maker or (2) touch them off themselves as they trade in front of them, then that is criminal and should be punished by fine or expulsion. As you might imagine, detecting and enforcing this activity is very difficult, especially since regulators don't generally know as much as traders. So this indeed is one of the risks of using stops.- arezi - Post 66880http://boards.fool.com/Message.asp?id=1030013012057020&sort=postdate4/26/00 - At the end of last month, just for curiousity, I had a list made of the five stocks with sales over $30 million per annum having the highest ratio of Market Cap/Sales.The result was this list.SYMBL...RATIO...HIGH...3/30...4/2...MKTCAPINSP.......435.......138.5.....65.1.....65.....16.0BJNPR.......290.......312.9.....267.....192.5.....47.7BARBA......270.......183.0.....110.....66.9.....24.2BMFNX.....268.......57.9.....45.3.....30.2.....27.3BVRSN......235.......258.5.....152.....111.....19.5B...Note that all five had already made their highs for the year: INSP,ARBA and VRSN in early March, and JNPR and MFNX in late March. Four of the five are lower now than a month ago by fairly significant amounts.In retrospect, i.e., with 20/20 hindsight, it seems obvious that these stocks, and even their less egregiously overextended bretheren, were vulnerable to a significant decline. All five have dropped a lot more than the overall market has. One might have concluded on the basis on these highfliers' market cap/sales ratios that they were vulnerable to a signifcant decline. But I can find no other financial variable that would have predicted, or did predict, that the time had arrived. The only indicator was failed momentum in the prices of all five stocks. - arezi - Post http://boards.fool.com/Message.asp?id=10300130121800004/26/00 - The Troublesome Trifecta is an attempt at this, and is based on rules suggested by William O'Neill, David Ryan, some TA people, and others. Here's how it works: I used the Java charting tool at www.equis.com/java to produce charts on the five stocks you mentioned following at the end of March. They were INSP, JNPR, ARBA, MFNX, VRSN. You type in the name of the symbol there atop the left side of the charting engine, after waiting a minute for the app to load. Then, you can pull down indicators over to the right, from a drop-down box which houses indicators such as RS, Volume, and Momentum.The statement of The Troublesome Trifecta reads like this, with the rules applied in this exact order, at the end of each month: Troublesome Trifecta Rules1) Momentum has dipped below 75 (75 is at the low end of the range of Momentum that www.equis.com/java measures for its Momentum indicator);2) RS has dipped below 40, as measured by www.equis.com/java (40 is at the bottom end of the range of RS that is measured by the RS indicator in the charting engine); 3) Volume is at least double what it was at the beginning of the month.The final Troublesome Trifecta rule is that if 50 percent or more of the stocks in a momentum investor's portfolio have triggered the Trifecta, then 6/3 put options are bought on the stocks that did the triggering, with the puts bought at 10 percent of the underlying stock's value. The above rules are designed not to pick when to sell every momentum stock, but to call out three troublesome trends that--when seen working in conjunction--could be a warning sign. The rules were not arbitrarily chosen, and are designed to be triggered only at extremes. They don't call for selling stocks, only for sometimes adding puts. They are designed as possible occasional enhancements to MI, not as frequently applied TA Rules- sparfarkle - Post 67420http://boards.fool.com/Message.asp?id=1030013012180001&sort=postdate4/30/00 - As one who has used stop losses and talked about it on this board, I can confirm that they increased losses during the "Recent Unpleasantness". I have also recently figured out how to backtest the strategy I was using (at least back to 11-98). That strategy used 12% stops with monthly PEGs. Believe me, you don't want to use this strategy.- jhindorff - Post 67741http://boards.fool.com/Message.asp?id=1030013012221004&sort=postdate12/23/99 - The point of this exercise is to see if selling PEG stocks using a manual, trailing stop-loss point was better or worse than a time specific sell period. The rules I chose reflect my own personal trading style and may not be what anyone else is looking for. I don't really like using manual stops and I check my stock prices every morning anyway. For this exercise, I want a price dip to be a signal to watch the stock very closely, not to actually sell it. My sell rules are: . 1) sell the stock the next day if it closes at -20% of the buy price . 2) sell the stock the next day if it closes at -12% of its highest closeprice since it was bought . 2a) ... but only if that -12% price is above the buy price NOTE: The -12% and -20% numbers are probably not optimal but they are relatively easy tobacktest with and do not seem to be overly volital. I'd like to play with the trailing percentage but have not yet.NOTE 2: I started with '95 because 5 year charts are much clearer to read than 1 decade charts. (I usedBigcharts.com for all charting)NOTE 3: I used PEG-3 because PEG-5 took too long to backtest. I also did a PEG-4 test from '96 and the results were similar. Also, some stock's charts were non-existant so I used the next pick on thePEG list. And, I would not buy more of a stock if I already owned it, I used the next pick on the list.NOTE 4: I only did buys on the 1st of the month. So, if a stock was sold on 9/5/95, the next buy date would be 10/1/95. NOTE 5: On months where the buy dates for two or more stocks are the same, rebalancing was done. - tribble - Post http://boards.fool.com/Message.asp?id=1030013009418000&sort=username5/10/00 - I posted the backtest results for '95-'98 in December (#51509) and I finally got around to doing '99. The point of this exercise is to see if selling PEG stocks using a manual stop-loss point was better or worse than a time specific sell.My sell rules are: 1) sell the next day if the stock closes at -20% of the buy price 2) sell the next day if the stock closes at -12% of its highest close price since it was bought 2a) ... but only if that -12% price is above the buy price Can we draw any conclusions from this limited backtest? Only one: that a manually implemented stoploss-sell system appears to have enough credibility to warrent further investigationAvg gain/loss = 49.8%, # winners = 7, # losers = 2'99 Start up cash = $316,720 with End cash = $972,647 This is a gain factor of 3.071. '99 CAGR = 307.1% This tacked on to the '95-'98 data the compound '95-'99 CAGR = 200.8% - ttrible - Post 68764http://boards.fool.com/Message.asp?id=1030013012426000&sort=username6/25/00 - A "save gain" sell strategy I then watched as SDLI rocketed from my 133 (split adjusted) price to 221 by March 3rd. All of a sudden I was sitting on a 60%+ gain. And not being the good-dooby mechanical investor, sat on it for more than the month, and watched it go right back down to virtually my purchase price (luckily, I again un-loyally did not sell, and watched it go back up to 267 – SOLD!). I digress. The point is, in the entire 14-year back-test of '86-'99, there had never been a month with a 60% return! Although I didn't know it at the time, I was in “three-sigma” territory. As the price went down, I muttered the familiar “Drats, I should have sold back there!” story to myself, and decided to do a little research to investigate when I might actually sell mid-month on a similar price spike in the future.The object would be to not sell too soon, but ensure a healthy gain once achieved. I decided somewhat arbitrarily that if the screen made it into plus (+) two-sigma territory, I should make sure to reap that reward as a minimum.My simple calculations of two sigma (95.45% of a 166-sample size) showed that less than 7.55 occurrences would be in that range. For one-month gains, the “seven occurrences” cut-off ended up being >43%, and for rolling two-month gain it was >63%.The strategy would go something like this:Once the screen rises past a two-sigma gain, say 5% above, you would put in a "save gain" sell order at no lower than the “two-sigma” point.My theory is the two-sigma gains will happen significantly more often during the course of any month than the end-of-month price shows. For example, although there are only 7 occurrences of 27% gainers in a month on my sample PEG-O, there would be substantially more times that the screen was up more than 27% at some time during that month).I can see that one would definitely lose out on some winners that pull back to the "sell point", but which would have gone back up before the end of the month. But you wouldn't necessarily miss all the three-sigma gains either, because you don't automatically sell the screen once it hits two-sigma levels. If the theory is correct, you will keep more “BIG” winners than you will lose out on “HUGE” winners, having a net positive effect on the overall long-term CAGR.- DragonTales - Post 72707http://boards.fool.com/Message.asp?id=1030013013158000&sort=postdate
Avg. # Add. 6 Month Arith. Weighted # of Stop No Stop Days Add. Average of AverageStop% Stops Ret. Average to Stop Returns Combined Stops 1 755 0.90% 0.00% 1 0.12% 1.02% 1.02%2 755 1.30% 0.00% 3 0.28% 1.58% 1.58%3 755 1.58% 0.00% 5 0.49% 2.07% 2.07%4 755 1.89% 0.00% 7 0.74% 2.63% 2.63%5 755 2.36% 0.00% 11 1.08% 3.44% 3.44%6 755 2.34% 0.00% 14 1.41% 3.75% 3.75%7 754 2.70% 44.82% 19 1.89% 4.59% 4.64%8 753 2.79% 22.95% 23 2.30% 5.09% 5.14%9 753 2.88% 22.95% 28 2.73% 5.61% 5.66%10 750 3.47% 49.13% 34 3.32% 6.79% 7.07%
Monthly=======RSW rank stnd. buy ascending, sell #5 desc.======== ===== =========================== 1 32.5% 32.2% 2 40.4% 26.0% 3 18.5% 33.2% 4 4.1% 41.4%
Decile Avg 1 Avg 3 Avg 6 Avg 12 Improve 1:120% 0.81 0.85 0.92 0.99 22%10% 0.91 0.96 1.02 1.12 22%20% 0.95 0.98 1.04 1.14 20%30% 0.98 1.02 1.07 1.19 21%40% 1.01 1.06 1.13 1.25 24%50% 1.03 1.08 1.15 1.27 23%60% 1.06 1.11 1.17 1.31 24%70% 1.09 1.12 1.22 1.34 23%80% 1.14 1.20 1.26 1.43 26%90% 1.25 1.31 1.38 1.57 26%
gain in months 2-61st 10%2nd 12%3rd 4%
2--3 2--6 2--121 2.6% 9.7% 18.8%2 3.5% 8.9% 20.5%3 3.8% 9.9% 16.9%
4--6 4--121 6.6% 18.3%2 6.4% 13.4%3 5.6% 13.4%
7--121 8.6%2 8.1%3 8.5%
1 month Interim Rebalance 2--3 2--6 2--121 3.5% 12.5% 19.6%2 6.2% 12.2% 16.7%3 1.7% 6.5% 11.3%3 month Interim Rebalance 4--6 4--121 5.0% 11.9%2 6.9% 12.3%3 8.1% 8.7%6 month Interim Rebalance 7--121 6.1%2 4.1%3 0.3%
I need to figure out a way to get that Excel graph of period returns by positions on my board. You can see how the cumulative gains and loses mount for each period. Maybe you'll see something I don't--I'll admit I'm new to stop-loss strategies.I'm not certain by what you mean be "graph of period returns by positions" but I took the liberty of posting your cumulative RS-IBD graphical charts on Manlobbi's interactive site listed below. Its under - Primary Screens->RS (Relative Strength)->Graphic Summary.Just the Monthly and Quarterly are up, the Semi and Annual are not yet. I took the liberty of adding them to the site but if you wish they can be removed as you are the creator/"owner" of the charts.http://digitalscores.com/mi/Best Regards,- Mike
Sux,I'm coming in late on this thread, having been over in CANSLIM land for a few days.I would like to bring the following points to the discussion.1) You might want to consider (if it's possible) to qualify the stop loss with a time window. IE, if the stock loses X%, and stays there for a number of days - sell it.2) The stoploss used by CANSLIMMers is traditionally no more than 7 or 8 percent. The reasoning is that anything more than that can result in too much risk to your capital. If it triggers accidently, so what. You'll get 'em next month.3) I would humbly suggest that comparing the stock with an index is exactly what you DON'T want to do. This too comes from CANSLIM. Look at it this way: If you owned a big fat NASDAQ stock in April of this year, and it started crapping out, would it make you feel better to know it wasn't crapping out any worse than QQQ? No Way Jose. You want to preserve what's left of your money and get the heck out.4) The other thing you might want to think about is adding a bit of a qualifier on the buy side. If a stock is more than X% above where it was last month, you might just want to let it go. It's probably due for a rest. THat's the closest thing I can think of to CANSLIMs rules regarding only buying stocks as they come out of solid bases. It's weak, but it might help.5) Another bit of buy side qualification might be to see if the appropriate index is gaining or losing. If you are going to buy a NAS stock, and QQQ is down this month, you might just be buying some trouble. One of the greatest truths of CANSLIM land is that you can't argue with the market. If things are weak and rotten, just wait till next month.I know these ideas represent heresy and a lot of work to implement, but who knows? THey might help....Z1
Sux wrote:I'll put it on the block for testing. Remember I have all the data, but it takes a little bit of time to set up the portfolio you want. What do you want to use? 1 stock? 3 stocks? 5 stocks?How about that 30% gain/25% stop? Cumulative over the life of owning the stock (you could have owned BBY and DELL for over a year in the 5-stock version in '97 and '98, respectively) or just monthly reset every "purchase" day?Sux,I propose the 1 stock model. On my Occ'l O'lap screen, the 2 sigma threshold was 43% for one month, and 63% rolling two month. Don't know if the RS-IBD has the price swings mine did, so maybe a 40% gain/35% stop with monthly reset. If your program allows, maybe add in the rolling two-month stop at 60% gain/55% stop. The rolling two-month may be the same stock held over, or it may be two different stocks' compounded gain. Threshold would apply either way.The Occ'l O'lap I specifically addressed had the following data:URL: OOv86990101rqrss12rqpss23pqpss23pqpq22'86-'99CAGR 155GSD 49Sharpe 1.49Plus two sigma monthly: 43%Plus 2 sigma rolling two-month: 63%suggestion was:monthly 48% gain/43% stoprolling 2-month 68% gain/63% stopTIA,Tailsemailed and posted
I wonder how MGBloomfield's ATR stops have developed?What say you Mike B.?MSN article on ATR stops 11-10-99:http://moneycentral.msn.com/articles/invest/models/4858.aspmrb
<<What would you recommend? I was going off of what you mentioned at MICon about your 50/50 rule. I just used the +10% threshold as an "activation point" for the rule. I believed that would be an acceptable point in wanting to perserve gains. Do you think using that type of methodology on a short-term screen is not good, or is it more that my bound is way too tight? Should I be looking at cumulative loses instead of perserving gains?>>Alas, this is the problem with trying to make formal rules out of a wishy-washy handwaving rule-of-thumb. ;-)Yes, I think the 10% threshold is much too tight. Maybe it's okay for a day-trader (or a week-trader), but I think it's too tight for mere mortals with a life.The 50/50 rule ("half the profit belongs to the market and half belongs to me") is a way to help you avoid ulcers in deciding what to do with your stock after a huge runup---50% or more. And to keep you from selling a winner just because it has a normal slight negative fluctuation. It is not applicable to a mere 10% profit.Ray
Do you mean a '72 to '82 flat market? The DJIA roughly doubled between '72 & '86, not including dividends (which were very significant). And the fear you're concerned about distills to the biggest 1-day drop in history, followed by the worst 2-year market ('72-'74) in the last 70 years. Kinda double counting. Frankly if these kind of things concern you (and it is not unreasonable that they do), you should think twice about the more aggressive MI screens. --Terry Didn't have the exact data in front of me at the time, but looking it upnow I see the S&P Composite's flat period was 14 years from 1968 to 1982.Extended flat periods are not that uncommon. Over the period from 1871 tothe present, you had 31% chance of not breaking even at the end of 5 years.After 10 years the percentage is 21% and after 15 years it's 13%.I couldn't determine the exact statistics on losses, but by one simplecalculation if at any time during that period you incurred an instantaneous15% loss, you had a >40% chance of not recovering that lost buying power(CPI) during the next 10 years.MI is a process for investing unemotionally using established rules.The optimal screen selection criteria varies from person to person.Some young persons with long term investing periods ahead of them cantolerate extended dry periods so high CAGRs are their criteria. Forothers near or in retirement, minimal Drawdowns and moderate CAGRS would be the appropriate evaluation criteria. While seeing 3 digitCAGRs sets everyones heart racing, some of us must cover our eyes andjust peak at the numbers. I hope this board does not become obsessed with CAGR but continues toprobe GSD, negative sigmas, Sharpe, drawdown, recovery period, etc.i.e. RISK related parameters. DGM
Oh, what the heck.I am definitely a minor leaguer here, but I'll toss this in anyway.I have been trying to work out a stop loss method just for the heck of it, but I don't have the data or spreadsheet knowhow to really test it. The following is purely hypothetical.Base your sellout on a percentage of GSD (50%, 75%? I don't know how to test it). For instance, A Keystone screen I'm looking at right now has a GSD of 20. If a stock in the screen drops say, 15% over the holding period, drop it. A PEG screen here has a GSD of 49. Hold on a little longer, because you know it's more volatile, and has a better chance of swinging up again. Hold until it drops 37%. These examples are for 75% of GSD, but I hope you get the idea. Just another cook throwing spice into the broth.Bill.
If the owner sells out before the monthly update then cash is held until the next update.That right there could be the cause of your underperformance. The screen is holding cash while something else (conceivably the stock you were just stopped out of) is going up. In a soaring market the opportunity cost of being in cash is high.One way to approach this might be to examine a stock's price at every rebalancing period. If it is X% below the purchase price/last rebalance price/recent high, ditch the stock regardless of its screen ranking and buy the next one down the list.someone that only invested in the Position #1 stock for that period had a 163,100% returnMy abacus tells me that over a 15 2/3-year period this works out to a CAGR of 60.3%. Is that what everyone else gets?
dmiddlebrook said: MI is a process for investing unemotionally using established rules.The optimal screen selection criteria varies from person to person. Some young persons with long term investing periods ahead of them can tolerate extended dry periods so high CAGRs are their criteria. For others near or in retirement, minimal Drawdowns and moderate CAGRS would be the appropriate evaluation criteria. While seeing 3 digit CAGRs sets everyones heart racing, some of us must cover our eyes and just peak at the numbers. I hope this board does not become obsessed with CAGR but continues to probe GSD, negative sigmas, Sharpe, drawdown, recovery period, etc. i.e. RISK related parameters. *******************************************************I really agree with these observations. In reviewing the extensive posts on stops, it seems evident that much of the problem is that the variety of schemes that have been imagined have not really been tested with anything approaching the rigor of MI. But more fundamentally, what stops are "best" may indeed not be measurable. The reason is that while when it comes to CARG we all agree that more is better, our appetite for and understanding of risk varies widely. So aside from the factors that DGM mentioned above, we all vary in our simple tolerance for risk.This is also why I agree with the wish or hope that MI not become too focused on CARG. It seems clear at this juncture that MI can achieve high returns over time. What is less clear is that MI has an adequate handle on risk. Even more basic, we seem to lack a uniform and comprehensive format for expressing the risk of various screens. CARG alone seems insufficient. It occurred to me in reading LAPropDoc's excellent guide to MI that it would be useful to have a more comprehensive set of statistics for each screen. I am thinking of simple stuff like: percent winners/losers, average winner/loser and average trade, maximum drawdown, etc. From such simple statistics it would then be easier to compare not just CARG's of screens, but expected values, etc. Being the rational and emotional creatures we are, some of us much prefer techniques with higher percentages of winning trades, while others find low drawdowns much more palatable. Others just want high returns, risk be damned. We need more information and more data to make those choices.arezi
Sux: Sorry.I was refering to a trailing stop in the conventional sense,namely re-setting your stop after "n" periods of time,ie after each daily closing,or after the friday close each week etc. If you use a -30% trail stop,you would start at your purchase price and then after each "n" period, if the close was higher than the previous close,re-set the stop,if it wasn't keep the same stop,etc. I hope this clarifies,if not please advise.Your work is more than greatly appreciated by us non-quants.Wayne
If you use a -30% trail stop,you would start at your purchase price and then after each "n" period, if the close was higher than the previous close,re-set the stop,if it wasn't keep the same stop,etc. I hope this clarifies,if not please advise.OOoohhhh. Assume that I be clarified.Sux
Wow! Amazing work, Sux. One more to add to the "to-try" list... consider a multiple (say, 2.5) of Average True Range for the previous month. This gives you a self-adjusting trailing stop which depends on the volatility of the underlying stock for the previous month. That should help with the stopping-out-of-winners-because-of-volatility problem, while keeping most of the protection. More volatile securities get wider stops.Of course, it may well be that in a positive expectation system, introduction of stops _must_ reduce the performance of the system (at least I think I once read that somewhere...). The win from stops is that is may also reduce the volatility/risk exposure.-gelasmus
Of course, it may well be that in a positive expectation system, introduction of stops _must_ reduce the performance of the system (at least I think I once read that somewhere...).Interesting observation. "Positive Expectation" means that on average we expect our stocks to gain money over time going forward. Even if we lost a bundle, from here we expect it to go up. So holding the stock always has better CAGR than holding cash.A couple of implications. If there is a drop so large that the stock becomes a shorting candidate, it would no longer be "positive expectation." But we are mostly buying huge RS stocks, so a big drop mostly puts our stocks as "large RS" instead of "huge RS." We see this in our screens with some regularity, where stocks with recent noteworthy drops are still ranked very high in RS because the monstrous runnup earlier in the period.Second implication - switching is more promising than stopping. "Better than holding cash" is not the same as "better than holding an alternative stock." This is starting to sound like "Hold until Drop."
"Positive Expectation" means that on average we expect our stocks to gain money over timeThe market itself (as represented by unmanaged indices) is said to have a CAGR of roughly 10%, hence the "positive expectation".
it would be useful to have a more comprehensive set of statistics for each screen. I am thinking of simple stuff like: percent winners/losers, average winner/loser and average trade, maximum drawdown, etc. From such simple statistics it would then be easier to compare not just CAGRs of screens, but expected values, etc.Have a look at http://members.aol.com/wwwebspace/Note that the studies examine returns relative to the market, not absolute returns.
This is in ref to Elan's msg # 77500...First thanks to Sux!! You just saved me a year of hand tracking a small test port that would have provided anecdotal evidence. This is an area I've spent a lot of time thinking about and discussing with others on the board...Elan has summed up pretty well what I believe the informal consensus for the best way to proceed is.First, test a simple stop loss based on the purchase price. Assume that the money from the sale is kept in cash until the next rebalance. Try selling at -5% from the purchase price, -10%, -15%, etc., maybe down to -40%. See if you can identify a sweet spot range. Let's say you find it between -20% and -30%. If you see such a range, try to zoom in at 1% increments.The lower your stop, the less stocks will be sold. I suspect that the number sold at -30%, let's say, in a monthly holding screen will be so small as to be statistically meaningless. Try to test the same idea on a quarterly, semi, and annual screen.Next step, like Hank suggested, might be to decide based on a drop relative to the market index. Try the S&P as the most obvious benchmark. The stops would probably be much tighter. Dropping 10% more than the index might be enough to trigger the sale. I would add just two teaks to what Elan has suggested.1) Once sold, buy the next stock down the list immediately or on rebalance and don't go back to the sold stock. (What would be really cool is if somehow you could backtest by keeping it off the buy list until it moved up in rank again.) Your initial example demonstrated why this is important. In addition, if possible run the test using weekly (Friday/Monday) vice daily data points. Do any of us really want to follow every stock in every screen we have every day? I know I don't. While this may not be the optimal thing to do, I think it is definitely more practical if we are going to implement a strategy based on these results.2) Use the Naz instead of the S&P for a relative stop... particularly if the data is late 80's forward. My personal guess is that a relative stop/loss will be the most effective (if any are). This assumes one wants to stay fully invested. O'Niel of course felt being stopped out of everything was a clear indicator to hit the sidelines for awhile. In hand tracking a relative stop over the past two months, my sense is that you should start at 10% as suggested. But like the straight stop...you'll probably want to test various levels. I initially thought a higher relative level would be better, but now I have my doubts. Any higher and it's looking like (for a monthly anyway) that you'll rarely be stopped out before rebalance time.Thanks again Sux...I can harldy wait for the results. I feel like a kid waiting for Christmas morning.
Thanks again Sux...I can harldy wait for the results. I feel like a kid waiting for Christmas morning.I'm not going to be performing those stop-loss backtests. There are too many requests and I'm too tired to do backtests. I provided the 16 year daily data for RS-IBD monthly on my website. I am beginning to accumulate the Formula 90 data and I'll post that in about 2 weeks. I'm done with testing. I'll assemble data but I'm not doing the stop loss stuff for folks. You can have the daily data for free if you report the results of your research.Sux
<<I'll assemble data but I'm not doing the stop loss stuff for folks. >>Oh darn!!I bet you're gonna bow out of the Quest For The Holy Grail, too!!!!And the Philosopher's Stone, too.Wimp ;-)
Wimp ;-):p - ppppppppppzzzzzttt!!Hey, you have to be hard sometimes.
Sounds like an ad for Viagra.:-)(Getting punchy waiting for VL to update)
Very interesting thread! For the RS-IBD monthly top 1 screen, one might consider using the 20-day Bollinger band instead of an arbitrary stop-loss %.If the closing price breaks below the band, shift to cash until the next month's purchase. This should protect against major downfalls beyond the normal short-term volatility of the stock.Just a pebble to toss in your pond.
Best Of |
Favorites & Replies |
Start a New Board |
My Fool |