For anybody interested in this approach, you'd go long tonight at close or tomorrow at open.The strategy: Be long the Russell 2000 (I tested using futures) for the"superior six" days around the ends of the 8 months per year October through May inclusive and cash the rest of the year.So, you're long only 48 days a year which is about 19% of the time.The superior six are defined here as the last 4 calendar days of one month and the first two of the next.So, you'd buy at close on the fifth-last trading day of the month (today)and sell at close on the second trading day of the following month (next Friday).The end October period is historically about the best of the bunch.To the extent that there is any statistical support at all, you couldeven go another couple of days into November and do well.It works about as well in bear markets as it does in bull markets.The last post on the subjecthttp://boards.fool.com/how-robust-is-it-when-nh-nl-is-negati...Jim
Just when the NH-NL signal is going bearish. * Last 5 days NH-NL per Stockcharts has been negative. * EMA(9) of it went negative on Tuesday.I moved my timing funds from long (15% or port) to short (10% of port)yesterday and I'll be on vacation next week (no access to the market) while undoubtedly there will be a big Hot48 bounce for the next 6 days.Sigh!StevnFool
Mungo, thanks for raising the topic. I have been researching this for last few weeks. Here is the academic paper that examined the end-of-month effect in US equity markets.http://papers.ssrn.com/sol3/papers.cfm?abstract_id=917884The charts on page 31 (period 1926-1986) and page 32 (period 1987-2005) are all you need to look at.Conclusion:for 1926-1986 period: last 2 and first 5 trading days of the month were good.for 1987-2005 period: last 5 and first 5 trading days were good for equal weighted (EW) strategies and last 2 and first 2 are good for value weighted (VW) strategies.Quoting from paper:Thus, over the period 1987-2005, the turn-of-the-month effect ispronounced and, as we will show, highly statistically significant. Additionally, as is apparent, especially with VW returns, virtually all of the excess market return over this 19-year interval accrued during the four-day turn-of-the-month period such that investors received little or no reward for bearing market risk over the other 16 trading days of the month.FWIW, i have tested this effect in one other international market and the effect holds.
Just when the NH-NL signal is going bearish.Ah, if only this stuff were simple and consistent. And always right, too!For further complication the seasonal buy signal will be soon, basicallyon the first small rise in the market.FWIW, one strategy I've looked at is to stay long all the time,but add hedging whenever *BOTH* NH-NL and Hot48 are bearish.You can think of this as hedging whenever NH-NL is bearish, but takingoff the hedges for brief intervals during the Hot48 bullish days.In short, the month end effect seems to work just fine during bear markets.It's the middles of the months (and summers) that get a lot worse.Ultimately the NH-NL signals are probably better.They react to what's going on in the market today, not merely the average over many years of history.Still, a few days of bounce would be unsurprising.I have another very short term purely technical signal that is bullish at the moment.It averages about one state change per week, bullish about 78% of the time,so an average cycle is long for 7.4 days then bearish for 2.1 days.Invented in spring 2010, CAGR bullish has been +24%/yr and CAGR bearish has been -20%/yr since then, similar to the original backtest figures.Jim
Seems like an interesting system to try out on a Leveraged ETF (TNA) perhaps?
Seems like an interesting system to try out on a Leveraged ETF (TNA) perhaps? Hmmm, perhaps...But first, check to see how it worked during the six days end of May start of June this year.Big difference between "working on average" and "working".Jim
So with the market closed for possibly days, when are the superior six days THIS month?
So with the market closed for possibly days, when are the superior six days THIS month? I have historically done all my day-of-month analysis based on scheduled trading days, whether they took place or not.In this case it's easy, since the buy signal has already happened andwill last into the start of next month. Had the shutdown been the firstcouple of days of November it would have cast into doubt the meaning of "stay long the first two trading days of the month", but for now it's unlikely to change the calendar dates for the trades based on the definitions "Nth last trading day of the month" and "Mth trading day of the next".The hold period will just have fewer trading days than normal.I'd interpret the usual sell signal to be unchanged, close Friday Nov 2.FWIWMy testing suggests that historically for the end-October period youwant to hold more than just 2 days into November. 3 days more seems optimalfor the Russell 2000 which I use, assuming that history is any guide.That would put the "best" sell signal at close Wed Nov 7.I use the Russell because it moves more, and still has futures.The typical R2k month-end six-day return is about .35% better than for the S&P.(that's index-only returns from Sept '87, not cyclically linked futures chains)Jim
My testing suggests that historically for the end-October period youwant to hold more than just 2 days into November. 3 days more seems optimalfor the Russell 2000 which I use, assuming that history is any guide.My testing suggests the same. The problem is that this year makes future backtesting less than reliable. I for one don't qualify past days as scheduled or not.
Jim,My testing suggests that historically for the end-October period youwant to hold more than just 2 days into November. 3 days more seems optimalfor the Russell 2000 which I use, assuming that history is any guide.That would put the "best" sell signal at close Wed Nov 7.I use the Russell because it moves more, and still has futuresI have been using IWM as a Russell 2000 proxy.I am uncertain about the futures contracts, are yourefering to these:The Russell 2000® Index is the recognized benchmark measuring the performance of the small-cap segment of the U.S. equity universe. The Russell 2000® Index is a subset of the Russell 3000® Index representing approximately 10% of the total market capitalization of that index. It includes 2000 of the smallest securities based on a combination of their market cap and current index membership. Mini Russell 2000® Index futures are available for trading on the ICE electronic trading platform for 22 hours a day...Symbol Mini-size contract = TF; Block Trade Symbol = TS... Tick Size .10 = $10If so, the tick size is $ .10 = $10.What is the expected AMPLITUDE of the ticks? In other words,what is the "number" of the Russell 2000?I studied for about 6 months before trading ES minis and feellike I understand (and trade) them with confidence.I am interested in getting a grasp on the Russell futures incase I elect to trade them in the future.I have been using ticker IWM for the Hot48.Thank you,kcanant
What is the expected AMPLITUDE of the ticks? In other words, what is the "number" of the Russell 2000?The Russell 200 is at 813.25 at the moment.http://finance.yahoo.com/q?s=^rut&ql=1The most important thing to know before doing any future contract isfinding out the multiple for that contract.For the E-mini Russell 2000 contract the multiplier is $100.Therefore one contract corresponds to 813.25*$100= $81325 at the moment, quite a chunk of change.For amounts less than that you'd have to use IWM.The multiplier for the E-mini S&P contract is $50, for the Nasdaq 100 it's $20.The value of stock controlled by each contract is relatively similar in each case.$53k for the Nasdaq, $70.5k for the S&P, $81k for the Russell 2000.If you have been trading ES before, the Russell futures are the same deal,just a different index and a different multiplier. They are all cashsettled, so you get an unrealized mark-to-market gain or loss whileyou hold the positions, and a cash settlement on expiration day if you don't sell first.For anybody who's going to try the Hot48, I recommend committingin advance to trying it for at least a year before trying it at all.The month by month results can be depressingly erratic even thoughthe rolling-two-year returns can be startlingly smooth.It's quite likely that someone starting out will have a couple ofbad months and quit, which isn't a very good outcome.As an aside:One nice thing about the Russell 2000 is that it isn't dominatedby just a hand full of gigantic firms: the 10 biggest in the indexaren't much bigger than the next 10. Thus, you're not unduly exposed to the noisy news from a single firm as with Apple in the Nasdaq.For this and other reasons I find it is better behaved and more predictable.Jim
TD Ameritrade has a trading platform called "Thinkorswim" in which you can trade the Rus2000 futures (/TF).1-Does any one have had any experience trading (/TF) on this discount broker platform? What is your observation?2-They also offer Euro/USD futures, i) E-min,(/E7) Euro FX futures Qty x tick 0.0001=$6.25ii)E-micro,(/M6E) Euro/USD future Qty x tick 0.0001=$1.25How about these two?
Hi Jim,Any thoughts on using the S&P500 instead of the Russell? I tried it last month (ok, not Hot48, just a superior 6 and trial run) and again this month (placed the bet before seeing your post). I'm not sure if the Russell is available in the most convenient form for me, so I might be restricted to the Dow/Nas/S&P.SA
Any thoughts on using the S&P500 instead of the Russell?It's fine.The problems are that it doesn't move as much short term, and it'sdominated by a relatively small number of ultragigacap companies whosemovements could cause a big shift that isn't a market-wide shift.Apple, or financials, or whatever. But it works fine too.In fact, it works better than the Russell on a few month ends.There is a modest small cap/ large cap seasonal cycle.I actually use whichever has worked best at any given month end, but that's usually the Russell.This is very unlikely to be a big enough effect to be worthwhile, but I'm a compulsive overtuner.A simpler rule is to use the Russell because it worked a fair bit better on average in the past.A quarter or a third of a percent extra per month-end adds up.Jim
Best Of |
Favorites & Replies |
Start a New Board |
My Fool |