No. of Recommendations: 68
May, 2006

What is MI? part 5 (of 6): Basics, Thinking

---- Basics
How much money do I need, in order to get started using MI?
Which broker should I use?
What are some MI screens or methods that are useful for the small portfolio?
What are some MI screens or methods that are well-suited to a taxable account?
What are some MI screens or methods that are well-scaled to the “medium”-sized portfolio?
What are some MI screens or methods that are well-scaled to the large portfolio?
What are some MI screens or methods that are well-scaled to the very large portfolio?
Anything else I should be aware of, when I start using MI?
What's an appropriate benchmark to compare my investment returns to?
Stuff to avoid?
---- Thinking
Is this investing?
So, does MI work?
How should I frame my expectations for my investment returns?
It's only money, right?
What is rich?
What do I do when I have enough to retire?
What if I die?
How do you stick to your screens?
Do you have to know all this stuff to follow the MI board?
Closing thoughts


“Don't just do something, sit there!”
Elan in FW#14101, 5/26/2000


How much money do I need, in order to get started using MI?

Here's one opinion:

It's All Relative, 9/14/1998
Spend the next 30 days rummaging around your house, your car, your office drawers, your closet shoe boxes, your penny jar, and anywhere else you can think of for loose change--pennies included. … I bet you can find $150 in loose change this way. I recently assigned my daughter to this exercise for a cut in the action, and she found $330. Next, go out to the garage and find a few old boxes of used books, and sell them. Sell those records out there too--CDs are in now. Through these kinds of collection efforts, you can raise $500… Next, take your $500 and invest it at an electronic brokerage in an IRA account where you pay commissions under $10 for trades. Buy the five stocks picked by my model, and switch them in a year. Do this for 25 years.

In general we like to say that you need to keep commissions & fees below 2% of your account. (Galeno used to insist that you need to keep your Total Expense Ratio below 1%.) So let's say your account is at Scottrade and you get $7 trades. If you want to trade annually, then that's 2 trades per position per year: buy & sell 1 stock, which gives $14 in commission per position per year. To keep commissions below 2% a year, you'd need a minimum of $700 per stock. Assuming a 12-stock portfolio, that means you'd need a total of $8,400.

What about monthly trading? No one recommends trading just one screen, 5 stocks: but let's say one part of your portfolio is a 5-stock monthly blend. Then figuring an account size to trade that portion of your blend would go like this:

Monthly trading = 24 trades per position per year (1 buy + 1 sell, * 12 months)
$7 trades * 24 trades = $168 commissions per position per year
To keep commissions below 2% requires $8,400 per stock.
So to trade 5 stocks monthly you'd need $42,000 allocated to that portion of your portfolio.

Moral: frequent trading costs money.

Elan used to have a rule of thumb formula he liked, to determine how many stocks to hold for a given account size:

Number of Stocks per X Amount of Money, FW # 17549 9/1/2000

I don't know if he would still advocate that formula. It gives the following:
    Account   #stocks
$10,000 5
$16,000 6
$26,000 7
$40,000 8
$64,000 9
$100,000 10
$159,000 11
$252,000 12
$399,000 13
$631,000 14
$1,000,000 15
$10,000,000 20

I personally have never been comfortable with this formula: way too few stocks as the account gets larger, for my taste. Notice how large each individual position is. With 12 stocks you're taking a 20 grand position in each. No small caps here, please! He had an earlier version, that doubled the number of positions at each level:

How many screens in a Port? FW # 6627 12/28/1999
Another neat formula might be a log scale. At $10K you buy 10 stocks. Every time your potrfolio gets 10 times bigger you multiply the number of stocks by two. The formula is N=10*log(P) where P is the portfolio size in thousands and N is the number of stocks.

What is interesting is the idea that your number of positions varies with the log of your account balance. It's a different way of thinking about stock portfolios, than you might usually see. This might be the way a mutual fund manager looks at stocks. Liquidity becomes very important; and you'll certainly want to use market orders rather than limit orders, when you start moving in and out of a $10,000 position.

I usually trade much smaller: I'm toward the opposite end of this spectrum. In fact I probably tend to trade too small, right up against the line where trading costs are an appropriate percentage of account size. Right now in one account I'm trading 24 stocks quarterly, with an account size around $57,000. The account is at Brown, with $5 commissions. With quarterly trading, that's 8 trades per position per year (1 buy + 1 sell, * 4 quarters), which gives $40 annual trading costs per position. Keeping trading costs at or below 2% per year means that each position has to be at least $2,000. My own average position size is about $2,375, so I'm just barely on the right side of the fees line. What happens if I have a bad year this year, and my account drops 16%? Suddenly my commission costs would be more than 2% of my account. Would I then have to cut down on my number of positions? It's a precarious spot. So I'm probably trading too many positions for my account size.
(It'd be easier for me if we didn't have so many good screens! Then I could make up my mind. ;-)

The right answer is probably somewhere in between. And I should tell you, Elan has much more practical investing experience than I have, so if we represent two ends of a spectrum then the sweet spot is likely more toward his end than mine.

This question about how much money you need is probably easier to answer in the context of a discussion about MI methods for different levels of assets, so see below for an extended discussion.

Which broker should I use?

There's a Discount Broker board here at TMF, where this question should probably be directed.

Discount Brokers:

The brokerage marketplace changes around fairly frequently. At one time Brown & Co was considered the obvious choice; more recently (as of early 2006) lots of MIers seem to have gravitated more toward Interactive Brokers or FolioFN. The points on which to evaluate a discount broker include (in random order):

• Financial strength & stability of company
• Fees & commissions
• Ease of using trading interface
• Customer service
• Availability of the securities that you want to trade

Surf that other board, and ask either here or there what the current thinking is.

What are some MI screens or methods that are particularly adaptable to the small portfolio?

How small is small? Shall we say that a “small” portfolio is one under $20,000? And then a “tiny” portfolio is from $0 up to say $5,000. A portfolio size of $0 would indicate that you're just starting, you've got basically nothing in your account, you're building it by making regular small contributions.

The Tiny Port

Remember we have purposely defined MI broadly enough, that a technique like dollar-cost-averaging into an index is properly “MI”. Say the portfolio you have with which to start with is just the $500 it takes to open an IRA somewhere; say further that the IRA is a Roth, and you can put $100 into it every month, or every-other-month. Start by buying $500 worth of PRF, the Powershares FTSE RAFI 1000 index ETF; and each month or two buy another $100 of that. If you're not sold on PRF, then try RSP, the Rydex S&P 500 Equal Weight ETF Why PRF or RSP, rather than SPY? Check out these 2 threads at MI # 174104 and #185679:

Most boring MI technique, MI # 174104 8/5/2005

RAFI 1000, MI # 185679 3/21/2006

(Everything beats marketcap. ;-)

This is Dollar Cost Averaging, and it is a simple yet powerful method of entering into the market. It's an old concept, well-known when Graham was writing The Intelligent Investor. If you're buying into the market with a new installment of the same amount of money every month, then in some months the market will be up and your purchasing power will net you fewer shares; and in some months the market will be down and your money will get you more shares. At the end of the day (decade), you'll have a solid block of shares, and you should have a decent cost-basis for those shares, because you bought “more” of them at lower prices and “less” of them at higher prices. This is a surprisingly powerful wealth-building tool.

Here are two TMF articles about DCA, although they emphasize using it to buy single stocks rather than a market ETF:

Blind Dollar Cost Averaging, 3/14/2000

Dollar Cost Averaging: Slow and steady wins the race, 5/23/2002

It might be useful to work thru an example. Suppose you started investing in early 1993: you have $0 in the account to start with, but you are able to kick in $100 a month. You're investing in the S&P 500: so on the 1st trading day of every month, you buy however many shares of SPY (the S&P 500 ETF) you can afford, and the leftover cash sits idle until you kick in $100 next month. Nowadays SPY is over $120 per share, so there will be some months where you won't actually make a buy, instead you just accumulate and make your purchase the next month. Back in 1993 SPY was around $37 a share, so you might buy two shares one month and three the next. So your first year would look like this:
Date       price  #shrs   Leftover   CumShrs
1-Apr-93 36.41 2 $27.18 2
3-May-93 35.82 3 $19.72 5
1-Jun-93 36.91 3 $8.99 8
1-Jul-93 36.56 2 $35.87 10
2-Aug-93 36.58 3 $26.13 13
1-Sep-93 37.83 3 $12.64 16
1-Oct-93 37.78 2 $37.08 18
1-Nov-93 38.45 3 $21.73 21
1-Dec-93 37.99 3 $7.76 24
3-Jan-94 38.27 2 $31.22 26
1-Feb-94 39.50 3 $12.72 29
1-Mar-94 38.40 2 $35.92 31

In April 93 you buy 2 shares of SPY at $36.41, which leaves you with 27 bucks left over. Next month you've got the new hundred bucks you just added, plus the 27 leftover, so you can buy 3 shares of SPY at $35.82, which leaves you with 19 bucks left over and a cumulative total of five shares. And so forth. At the end of the year you own 31 shares of SPY.

(For simplicity's sake I'm assuming zero transaction fees, which is not realistic but let's me illustrate the process. Note also the prices shown above are “Adjusted Prices” downloaded from Yahoo: they account for dividends.)

How would you have done, with that? If you had kept that up thru the beginning of March 2006, then you would own 209 shares of SPY, which at the March 1st price of 129.37 would give you $27,079.17 (that includes the $41 of left over cash you had after your March 1 purchase). There were 156 months in the period, so you kicked in $15,600 of your own money: the other 11 grand (73%) is investment return.

This is powerful stuff.

We know that looking at gross amounts can be confusing, when looking at investment returns: we want to figure out an annualized return. When looking at an investment with regular inflows of money, the XIRR() calculation is appropriate. XIRR() shows an 8.1% annual rate of return on the investment. So: you didn't break any records with this style of investing. In fact, if someone had taken half of the total amount you wound up investing, so $7800, if they had taken that and bought SPY on 1-Apr-93 and held it until March 2006, they would have $27,685.18 and a 10.3% annual return. This beats your annual return by a couple percentage points: because you only had small amounts of money in the early stages, you missed some of the runup to the great bull market of the late 90s.

But you didn't have $7,800 to start with! You had zero. What's amazing to me is that you caught up anyway. Disciplined investing with small amounts of money can create wealth over the long term. That's a very important principle to know.

So: Dollar Cost Averaging. You can, over the long haul, start to put together some real money. But it's a little simple for the Mechanical Investor, don't you think? We need to tweak, to fiddle, to switch, to use an algorithm to select from a pool of choices. SPY is so vanilla. Can we spice this up a little, while retaining the “sell never” feature that distinguishes Dollar Cost Averaging? Well, yeah.

Remember that Dollar Cost Averaging implies a holding period of “forever”: it's a very long term approach, even a “sell never” approach. Since there's no selling to take profit, the entry price becomes the most crucial thing. You need a low cost basis: you want to get your shares somewhat cheaply. Revisiting our DCA example above: suppose that instead of always buying SPY, you instead used the Arezi Ratio to determine what to buy. If the AR is below 1.1, then buy SPY; but if it's at 1.1 or over, instead buy shares of a bond ETF. The fund IEF, iShares Lehman 7-10 Year Treasury, is a nice choice: however, it doesn't have history going back far enough to use with our SPY history. I simulated it by using 7+-share blocks of PRULX, the T. Rowe Price U.S. Treasury Long-Term mutual fund. The simulation, using the Arezi Ratio to determine which of SPY or IEF to buy, gives this result: after the March 1st 2006 buys, you have 165 shares of SPY and 84 shares of the bond fund (along with $46+ in cash), for a portfolio value of $28,587.28. XIRR() gives the annual return at 9%. This beats buying just SPY by almost a full point in annual return pctg: potentially a big deal over a long investing life, and worth over 1500 bucks in the current example.

Drawdown is also interesting to compare between the two approaches. The SPY-only portfolio hit the top of the bull market in Sept 2000 with a value of $20,618, and dropped with the market for the next two years. By October 2002 it was down to $13,824, a drop of -33%. The actual percentage loss was worse than that, because this simulation continued to put $100 a month in during the downturn. Take out the extra $2500 kicked in, and you're looking at losses over 45%. That's big money. It's half your savings! It took almost a year-and-a-half to work back up to a portfolio value equal to the Sept 2000 peak of over 20 grand, and an extra $1600 of contributions.

The mixed stock and bond portfolio took a smaller hit: still a substantial hit, but smaller. It was at $19,759 in Sept 2000, and it also dropped for the next couple years. It bottomed in August 2002, at $16,364. That's still a loss of almost a third of our savings; but better than the SPY-only portfolio suffered. Notice that the bleeding stopped two months earlier for the mixed port: then it took 11 months (and an extra $1100 of contributions) to exceed its former high. This is of course characteristic of diversification; it's also representative of having a lower cost basis in SPY; and the bonds weren't horrible either.

You should be aware that the example is in one sense cherry-picked so that the bond switch really does help. Not every lookback period is going to include a stock market bubble as huge as this one, nor a big crash immediately following. If the maket had kept trending upward, never correcting, then the stocks-only portfolio would have looked better. But most multi-decade market periods do include some major swings up and down: a sound means of deciding where to park each installment of your money should boost returns and increase safety, over the long term.

Two additional details:

• We chose a fund of medium-term bonds, 7-10 years. Although we've discussed these purchases as “hold-forever/sell-never” approaches, it's worth remembering that 7-10 yrs down the road the shares of that bond fun you bought will be holding different bonds. I would consider selling those shares at some point during that 7-10yr window, probably at a point where bond yields were low relative to stock yields, and buying a stock ETF with the proceeds.

• I did this model of DCA using SPY because it was easy. In practice, I personally would not use SPY to invest this way. Rather than SPY, I would probably use something like RSP, the S&P Equal Weight index; or maybe PRF, the RAFI 1000 index. Or maybe IWM, the Russell 2000 index; or MDY, the S&P 400 Midcap index. More likely, knowing me, I'd do some kind of complex switch among various ETFs, because I just can't stay away from fiddling with a methodology.

This look at using an MI switch to decide between two ETFs only scratches the surface. As mentioned, there has been a huge proliferation of ETFs in recent years. What if, instead of looking at the S&P 500 as a whole, you looked at the various sectors or industries ETFs that comprise it? Some of them are going to be cheaper (in terms of PE) than others at any given time. Here's a cross-section of them:
Ticker   Category                     Fund Name 
IYM Sector - Basic Materials iShrs Dow Jones US Basic Materials
XLB Sector - Basic Materials Materials Select Sector SPDR
IYD Sector – Chemical iShrs Dow Jones US Chemical Sector Index
IYC Sector – Cyclical iShrs Dow Jones US Consumer Services
XLY Sector – Cyclical Consumer Discretionary SPDR
IYE Sector - Energy/Nat Resrcs iShrs Dow Jones US Energy
XLE Sector - Energy/Nat Resrcs Energy Select Sector SPDR
IYF Sector - Financial Svcs iShrs Dow Jones US Financial Sector
IYG Sector - Financial Svcs iShrs Dow Jones US Financial Svcs
XLF Sector - Financial Svcs Financial Select Sector SPDR
IYH Sector - Health Care iShrs Dow Jones U.S. Healthcare Sector
IYJ Sector – Industrial iShrs Dow Jones US Industrial
XLI Sector – Industrial Industrial Select Sector SPDR
IYV Sector – Internet iShrs Dow Jones U.S. Internet Sector
IYK Sector - Non-Cyclical iShrs Dow Jones US Cons Goods
XLP Sector - Non-Cyclical Consumer Staples Select Sector SPDR
XLV Sector - Non-Cyclical Health Care Select Sect SPDR
IYR Sector - Real Estate iShrs Dow Jones US Real Estate
VNQ Sector - Real Estate Vngrd REIT Index VIPERs
IYW Sector – Technology iShrs Dow Jones US Technology
XLK Sector – Technology Technology Select Sector SPDR
IYZ Sector – Telecomm iShrs Dow Jones US Telecom
IYT Sector – Transportation iShrs Dow Jones Transportation Average
IDU Sector – Utilities iShrs Dow Jones US Utilities
XLU Sector – Utilities Utilities Select Sector SPDR
MDY Index - Mid-Cap MidCap SPDRs
IJR Index - Small Cap iShrs S&P SmallCap 600 Index
IWM Index - Small Cap iShrs Russell 2000 Index

Yahoo quotes publishes a PE for most of these. What if every month you picked up a few shares of the cheapest (most out-of-favor) sector/industry, as sorted by PE? Over the course of a whole business cycle (say 10 years), you would eventually get exposure to a broad range of industries, almost as much as you have with SPY; but since each piece of the market would have been bought relatively cheaply, you'd stand to get a better return than with SPY. That sounds promising. Or maybe this wouldn't work well: maybe sector/industry PEs aren't directly comparable, maybe some industries always trade at lower PEs than others: for example, maybe cyclicals always trade at lower PEs than tech companies, so over the long haul you'd wind up getting no exposure to some of the more important parts of the economy. I don't know, could go either way. In the absence of a backtest for such a strategy (and many of these ETFs haven't been trading long enough to have the necessary history), it's tough to know which way it would work out. But maybe you can construct a good simulation, using various indices to proxy for the sector or industry ETFs, and then run it as a hypothetical portfolio and post your results. This is exactly the kind of work we love reading about on the MI board. It's probably worth looking into.

Anyway: putting small, regular sums into the market, using some reasonable decision-mechanism for determining where to put it, will eventually grow your portfolio out of the “tiny” range, and into a range where you start to have more alternatives open to you for profitable trading.

The Small Port

A small port is very vulnerable to transaction fees when trading MI strategies. You certainly need more than $5,000 to absorb the costs of any real activity in the account, and probably much more than that. Any investment strategy that works for the tiny port will work just fine for the small port: simple Dollar Cost Averaging can take you a long way. But in this 5 to 20 grand range, you open up the possibility of selling a purchase to take a profit and then buying something else: that is to say, the possibility of trading.

To keep transaction costs small, you'll probably want to gravitate toward rebalancing annually. There's more detail on annual screens below, in response to the next question, about taxable accounts. I think those recommendations would work well for a small port.

I personally am a fan of another method, which I hesitate to recommend because there isn't really a backtest to support it. I like using RRS to trade the ETFs. The backbone of this goes back to some work Klouche did:

Using Relative Strength of Sector Funds, 4/3/2000

Actually, it goes back further, to some chapters in Schwager's Market Wizards books.

Sector fund RS, MI # 96843 3/23/2001
Gil Blake, profiled in The New Market Wizards. Schwager calls him "The Master of Consistency". In the twelve years leading up to that interview, Blake had averaged a 45% CAGR, with his worst year being a +24% return, and only 5 losing months out of 139 (96% non-losing months). He was (is?) a mutual fund timer. He traded the Fidelity sector funds!

For a more recent treatment, check out some of DreadPotato's posts, for example:

ETF - RRSXG Screen Ranks, FW # 32439 4/19/2006
(scroll to the notes at the bottom)

Another method: conceivably, at a low-cost broker, you could trade 2 or 3 screens quarterly (3 stocks each) starting at around 10 grand. There's more discussion on quarterlies below, in response to the question about medium-sized portfolios.

What are some MI screens or methods that are well-suited to a taxable account?

It is in general not a great idea to let tax considerations be the overriding determining factor in making investment decisions. You don't want to avoid everything that can generate taxes: you need to retain some sense of scale. Is it better to pay 35% taxes on a portfolio's 20% annual gains, or is it better to have the portfolio gain 5%? After taxes, the 20% gain becomes about a 13% gain, which is a terrible drag on performance: on the other hand, it's obviously better than a 5% gain. Yeah, taxes suck, but trying to avoid highly-taxed investments can lead us into bad decisions.

In the earlier discussion on friction, I included some links about tax angles in MI. Here they are again:

Gains and losses, MI # 67041 4/20/2000

Monthlies + margin + dump = Disaster, MI # 72247 6/21/2000

When to Go to Monthly in a Taxable Account? MI # 45249 11/12/1999

Ok. So generally when discussing taxable accounts our default assumption is that we would rather be taxed at the capital gains rate than at the income tax rate. This means holding stocks for over a year. So you want to look at what screens work well with annual holds, across start months. Here's a recent thread on that exact topic:

Annual blend - work-in-progress, MI # 180889 1/4/2006

That thread discusses some blends of VL screens, and constitutes an excellent recommendation. Here are some other ideas you should look at, from the SIP dataset:

Optiman: Dividend Growth Screen, MI # 95695 3/7/2001
3/4/2006: The annual OptiMan screen now has five years of post-discovery data. The CAGR for a five-stock annual has been 20% with a GSD of 14. This stomps all over the S&P 500 with a CAGR of negative 1% and a higher GSD of 20.

Shrinkage: The case for shrinking stock, MI # 141725 2/22/2003
2/5/2006: Another year of post-discovery data for the annual Shrinkage screen. The average return since discovery has been 20%.

Halloween and the Dows, MI # 163794 7/30/04
(This one involves semi rather than annual trading.)

Another Possible Annual Screen, MI # 182243 1/23/2006

You could do a lot worse than a blend of YieldEarnYear, HI_INC_Cash, Optiman and Shrinkage, 1-3 annual Dozens. You might want to read around on the board for any recent developments with annual screens (search the last six months for “annual” on the Datahelper search page).

In addition, I think the Benchmark Investing method is a good tactic for annual holds, and a nice diversification from the other MI screens:

Benchmark Investing:

Likewise the BMW Method is another tactic for using MI with annual or ever longer holds:

The BMW Method:

I also am very impressed with the Hidden Gems screen developed by mklein9, using the SIP dataset. This screen is intended for annual holds. Mostly smaller-cap growth companies: this should blend well with the Benchmark or BMW picks which are typically larger companies. Mklein9 did an actual MI-style backtest, with pretty nice results. Check here:

I personally would only trade annuals using a Dozens approach, for what it's worth. So if you're trading three screens 4-deep, so holding 12 stocks, then I would buy 1 stock from screen1 in January, 1 from screen2 in Feb, 1 from screen3 in March, then back to screen1 for April and so forth. Some rotation like that. That's me personally: other people buy their annuals in a basket at the start of the year and forget about them til next year, and they report nice results. One nice thing about a Dozens rotation is it lets you add money thruout the year, and you always have a place to put it. Since many taxable accounts represent continuous savings, where you are adding money on a somewhat regular basis, that can be a nice feature.

Remember that you have to hold your stocks for a year and a day, to get taxed at the capital gains rate rather than the income tax rate. Don't close a winning trade early. One thing I'm considering for my annuals is using a 13-month hold, just so as not to have to worry about it. Oh, and remember also that if you have a loser there's probably some tax advantage to selling it early, like after 11 months or 11-1/2 months.
The short-term loss nets against your income, so it can help you more on the tax front than a loss at a capital gains rate. Check with a tax professional for your specific situation

What are some MI screens or methods that are well-scaled to the “medium”-sized portfolio?

I'm not sure what to consider “medium”. Shall we agree that a port size over $20,000 and less than about $100,000 is “medium”?

First of all, you can do annual trading forever. There is nothing wrong with it. It works for any account size. It's a great way to trade. The portfolio doesn't need constant babysitting: it is unlikely to crater on you if you go to the Grand Canyon on vacation for two weeks. There is no need to bump up to a more frequent trading schedule. However, it is a fact that the sexiest returns here are associated with more-frequent trading.

I am somewhat in the minority here, in that I like to trade quarterly screens. I find them a nice compromise between the high returns of the go-go monthly screens and the lower costs associated with less-frequent trading. (Of course I use a Dozens rotation. ;-)

Zeelotes recently posted a look at 2-screen blends with quarterly rebalancing:

Small Port: Quarterly Rebalance, MI # 184345 2/23/2006

In addition, I recently recommended another 2-screen (6-stock) quarterly blend to someone; however, he had explicitly stipulated holding around 5 stocks:

Re: Gentle Screamers questions, MI # 184319 2/22/2006

The above links deal only with VL screens. There are some SIP screens that seem to work ok as quarterlies, over the short backtest available:

GSMungo, MI # 178541 11/27/2005

HighRelativeValue:, MI # 162979 6/24/2004

Up_5%, MI # 159290 3/18/2004

Small_Value, MI # 111479 11/20/2001

Eastwood: No Name Screen, MI # 159437 3/22/2004

That's not a comprehensive overview, there are probably more.

Six is a small number of stocks; and two is a small number of screens. I personally would look to use 3-5 screens, holding a total of 9-15 stocks, as a starting MI quarterly blend. Play around with the backtesters to see what blends work well.

Somewhere around this port size is where you can start considering the switch to monthly trading. In fact, given ultra-low-cost brokerages like FolioFN and InteractiveBrokers, you could theoretically start monthly trading at a much smaller port size than we would ever have considered before.

What are some MI screens or methods that are well-scaled to the large portfolio?

Assuming that a “large” portfolio is between say $100,000 and $1 million?

Everything. All of them. All of the MI screens and methods are geared specifically to this account size. This is the basic MI approach. The vast majority of what is written about and backtested here is tradable at this account size. All the great blends trading monthly, all the sexy momentum stuff, everything. Look at the posts about Zeelotes “Gold” screens, look at the various optimized blends, it's all in your reach. Go get it.

There are several users on the MI board with port sizes in this range. The general comments seem to be:

• diversify well over screen types
• keep an eye out for volatility: blend to reduce it
• watch out for liquidity issues in small-cap stocks. Use limit orders.

Have fun. I expect (hope) to be joining you in 2 years.

What are some MI screens or methods that are well-scaled to the very large portfolio, like over $1 million?

I have no idea. If you've been able to scrape together a mil, I doubt you need my advice on how to manage your money.

At this size I imagine a primary consideration is allocation among different types of assets. Not just equities, but bonds too, and REITs, and commodities; not just financial instruments, but real assets like investment properties. Maybe some hedging against currency risks. If you take a port over a mil, and divide it into asset classes, then maybe the stock part falls into the lower $100k-$1mil category and is merely a “large” portfolio, and the same strategies useful there would be appropriate.

The poster Sparfarkle on this board used to muse that there were some interesting possibilities with bonds and options, for large account sizes. For example, you could invest a large amount in the “risk-free” investment (US treasuries), and use the yield to buy options on stocks or indices. What makes this interesting is that it is theoretically a risk-free investment. If you completely blow up in your options trading in any given year, all you've lost is that year's yield: the capital is untouched. If you hit homeruns with your options one year, then you've made some money. There might some other useful ideas on the board about how to handle large money.

Maybe we'll all have this problem in a decade or two. ;-) Wouldn't that be nice?

Anything else I should be aware of, when I start using MI?

I think the first thing to be aware of is that, if you've used other more restrictive forms of stock-picking in the past, you're probably going to see more losers in your portfolio that you're used to seeing. You will probably see more losing trades in your port than you've ever seen before, especially if you're doing any kind of monthly trading. Get used to it. Remember that you're trading baskets rather than individual stocks. You will never stop having losing trades: you'll just accumulate more and more losers over time. You will accumulate a lot of losing trades: but you should also accumulate a lot of winning trades. And if you've chosen a sensible blend, over time the trades should balance out in a way you won't mind. But it will be nervewracking for a while, until you become used to the process.

Taking the Plunge, FW # 20678 11/3/2000

The second thing I would suggest is that you TRACK YOUR RESULTS! Set up a spreadsheet, use Excel's XIRR() function, or some similar tool; keep track of the dates and amounts when you add money. Also take snapshots of your portfolio value at somewhat regular intervals, like monthly or so. And TRACK YOUR RESULTS! Know how well or poorly you've done since you've started using MI. Be able to compare it to a benchmark, like the S&P 500 or the Russell 2000 or something. If you're underperforming and would be better off just buying the index, you need to know that. TRACK YOUR RESULTS! If you're kicking butt, doing much better than the index, you'll want to know that too. TRACK YOUR RESULTS! MI is all about being well-informed when it comes to investing. Being well-informed starts at home: TRACK YOUR RESULTS! Note that if you keep a log of monthly “snapshots” of your portfolio value, that will help you later if you want to calculate the volatility of your investments or a Sharpe Ratio of your performance or whatever. Keep track of your results.

Another admonition: “Plan to change.” Our state-of-the-art gets better and better, and we should reserve the right to know more next year than we knew last year. If you have a schedule under which you will re-think your screen allocations for the following period (maybe every year), it can reduce jumpiness or doubt.

What's an appropriate benchmark to compare my investment returns to?

Good question!

Probably not SPY, the S&P 500 index, even though that's an oft-quoted standard. The market is much wider than just the mega-caps included in SPY.

One good candidate is RUT, the Russell 2000 index. There's an ETF which tracks that, IWM.

Another good candidate is MID, the S&P 400 Midcap index. There's an ETF which tracks that, MDY.

Another interesting possibility is RSJ, the S&P Equal Weight index. There's an ETF which tracks that, RSP.

An even more interesting possibility is the RAFI 1000 index. There's an ETF which tracks that, PRF.

Remember we mentioned RSP and PRF before:

Most boring MI technique, MI # 174104 8/5/2005

RAFI 1000, MI # 185679 3/21/2006

Probably the thing to do is use all of those: SPY, MDY, IYM, RSP and PRF. More information is better, right? See how your results compare to those, over a period of years. If your MI returns come in 3rd or 4th (or 5th!) when added to that list, it might be an indication that you'd be better off in a good ETF.

It's also not a bad idea to keep track of how Warren Buffett is doing with BRK.A.

Some other threads on benchmarks:

MI Benchmark, MI # 184854 3.4.2006

The temptation to apply “common sense”, MI # 187420 4/26/2006

Benchmark selection by market cap, MI # 187520 4/28/2006

Benchmark Correlations, MI # 187789 5/3/2006

Stuff to avoid?

Won't do THAT again, MI # 184528 2/26/2006


Is this investing?

Or is it gambling, speculating? That's a fair question.

In its purest form, a mechanical investor using stock screens might not even know the name of the companies he just bought! Just the ticker symbols. And in this, I think participants on the MI board have long felt like exiles here at TMF, the redheaded stepchildren of the Investing/Strategies discussion boards. Most of the investing discussions at TMF, and also the official tack taken by the flagship newsletters and portfolios at TMF, focus heavily on a Graham-and-Dodd-influenced style of security analysis, where you look carefully at the business, pore over 10k's and such, estimate the discounted cash flows, interview the CEO, try out the products, look at the company's competitors, etc etc etc. From this exhaustive process the securities analyst is supposed to develop a sense of which business are outstanding, and worth tying your fortunes too. You can see how, surrounded by writers extolling the virtues of that approach, the freewheeling MI trader might feel out of place.

There's more. Not only may we not know the name of a company we just bought stock in: but that stock might be pure junk in the eyes of a careful securities analyst, particularly if the stock is one of our high-flying momentum screens. The analyst is looking for value, trying to buy a company that will be worth more in the future than the price at which the stock is trading today. If we have a momentum screen in our blend, then we might but a stock for no better reason than that it's gone up 25% in the last three months.
(And note that the chances are very good that we do have a momentum screen in our blend, because those are the ones with the most exaggerated, sexy returns. It can be hard to resist the temptation to throw ALL your money in momentum screens, GSD and Sharpe Ratio be damned. We're as vulnerable to irrational exuberance as any other market participants.)

The analyst blanches. We very possibly just bought some overpriced garbage stock, that is emphatically not likely to be worth more a couple years down the road than its current price today, because the business fundamentals aren't there to support the stock valuation. What's funny is that we don't necessarily disagree with that assessment: the stock we bought might just be overpriced garbage. But since we plan to flip it next month, hopefully for a nice little 4 or 5% gain after transaction fees, we don't care.

You can see how there can be a tension, between what we do here and what is advocated elsewhere at TMF. And I think that when most people think about what “investing” is, their intuitive picture is more along the lines of that Graham-and-Dodd-ish analysis than the flipping and trading we do. If that's investing, then if you look at MI (particularly with momentum screens), you almost have to conclude “That's not investing!”

Why I Abandoned Mechanical Investing, 1/8/2001
I believed that by limiting MI to perhaps 10-15% of my portfolio, I would find the risk acceptable. I did not. I happen to like all of my money and don't find it easy to potentially part forever with 1/8th of it. … At a psychological level, I found MI profoundly unsatisfying. I like owning stocks, reading annual reports, and watching them hopefully grow and prosper. With MI, you are just renting the flavors of the month. It's quite tedious and objectionable, really.

I got tired of [holding stuff] like LTR, PKX, KEP, and AET. Yuck! I mean, supposedly I'll get rewarded for owning dreck like that which is why I was using the system -- but, yuck! … overall, I didn't lose too many dollars. I'm pretty sure that someone who used MI carefully and religiously probably beat the typical actively managed aggressive growth fund, too. But it's not for me. When the system tells you to own things that subjectively you perceive to be garbage, it's very difficult to hold on to them just because some tests indicate garbage did well from 1986 to 1999. It also (from my point of view) turns an enjoyable hobby into something as stimulating as a slot machine. I'm still kind of under the impression that if you had a tax-deferred blind trust run by a monkey who put you into MI stocks no matter what for the next 20 years, and appropriate diversification was practiced, that you would probably beat most mutual funds, and probably by a wide margin. But the practical implementation, and the human factor, and all that make it no good for me.

Rather than trying to make myself into the sort of person who can do MI reasonably successfully, it is far easier and more likely to be successful for me to make myself into a person who can do Buffettesque value investing reasonably successfully.

And related thread:

How to give up on MI, MI # 89699 1/9/2001
MI just isn't for everybody. And, the single most important concept in developing an investment plan is to choose a method that suits your personality.

That intuitive picture of investing, it would be nice to refine that a little. What aspects of our image of investing are necessary to the activity, and what parts are extras? Do you have to try the product, and interview the CEO? Is there a definition of investing, that we can use?

As it happens, Mr Graham wrote about the difference between investing and speculating back in 1949:

“What do we mean by 'investor'? …The term will be used in contradistinction to 'speculator'. … An investment operation is one which, upon thorough analysis, promises safety of principal and an adequate return. Operations not meeting these requirements are speculative.”
– Benjamin Graham, The Intelligent Investor, 1949

Promises safety of principal? If that's the standard, we can throw in the towel right here. No MI method promises safety of principal. In fact, there's no promise of safety of principal with any method of stock investing. Graham seems to concede that, writing further:

“We trust that the reader of this book will gain a reasonably clear idea of the risks that are inherent in common-stock commitments – risks that are inseparable from the opportunities of profit that they offer, and both of which must be allowed for in the investor's calculations. What we have just said indicates that there may no longer be such a thing as a simon-pure investment policy comprising representative common stocks … that involves no risk of a market or 'quotational' loss large enough to be disquieting. In most periods the investor must recognize the existence of a speculative factor in his common-stock holdings. It is his task to keep this component within minor limits, and to be prepared financially and psychologically for adverse results that may be of short or long duration.
– Benjamin Graham, The Intelligent Investor, 1949

So there is a speculative factor in stock investing; there's no stock investing method that involves no risk of a market (or 'quotational') loss large enough to notice. The promise of safety of principal seems more like a best-effort statement of intent. We might justifiably restate Graham's definition along these lines: “An investment operation is one which, upon thorough analysis, tries to ensure safety of principal while providing an adequate return. Operations not meeting these requirements are speculative.”

So does MI screening try to ensure safety of principal, while providing an adequate return?

Well, I think that if you look at the backtested returns of some of our go-go monthly momentum screens, it's pretty clear that ensuring the safety of principal is not a big priority. RS26 monthly 1-5 shows a CAGR of 25 and GSD of 42, from 1969 to 2005. Nothing with a GSD over 40 is ensuring the safety of anything! Even holding 10 stocks from this screen leaves GSD at 33. The momentum screens are volatile: they go up, and they go down. And if you look at the individual stocks selected by our momentum screens, these stocks would certainly be described as “speculative” rather than as “safe” investments. So there are MI activities which definitely qualify as “speculating” rather than “investing”, per Graham's definition.

“Two [points] should be added about stock speculation per se, as distinguished from the speculative component now inherent in most representative common stocks. Outright speculation is neither illegal, immoral, nor (for most people) fattening to the pocketbook. More than that, some speculation is necessary and unavoidable, for in many common-stock situations there are substantial possibilities of both profit and loss… There is intelligent speculation as there is intelligent investing. But there are many ways in which speculation may be unintelligent. Of these the foremost are: (1) speculating when you think you are investing; (2) speculating seriously instead of as a pastime, when you lack proper knowledge and skill for it; and (3) risking more money in speculation than you can afford to lose.”
– Benjamin Graham, The Intelligent Investor, 1949

(If I ever write a book about MI, my working title is going to be “The Disciplined Speculator”.)

HOWEVER. The question was not whether we can speculate using MI techniques. We certainly can. But the question was whether MI as a whole can qualify as an “investment” operation, rather than a speculative operation. Can we construct a blend of current MI screens, that would meet Mr Graham's definition in terms of protecting principal while providing an adequate return, thus constituting more of an investment operation than a speculation operation?

How about we pick 12 stocks for an annual hold, using these 3 value screens:

Benchmark Investing (SP100)
S&P 100 stocks
Sort by low Price /Downside price

Timeliness 1&2 stocks
Current Dividend > 0
Debt < 65% of Capital
sort by high (Net Income + Cash)/MarketCap

The 35 stocks in the VL dataset with the highest value of
(DividendYield * EarningsYield)
Sort by 1-yr return

Screen 1 gives us mega-large companies at a relatively cheap price; screen 2 gives us “timely” stocks which pay a dividend and have low-debt, and are selling for a low multiple of their cash-on-hand and net income; screen 3 gives us high-yield / low-PE stocks which have been appreciating in price.

From 1991 to thru the end of 2005, if you had bought the top
four stocks from these screens in January and held for a year, your returns
would have been:
Year    Blend   S&P
1991 58 31
1992 36 7
1993 32 10
1994 -2 2
1995 50 39
1996 27 23
1997 33 27
1998 11 33
1999 2 20
2000 22 -11
2001 24 -9
2002 -1 -22
2003 56 29
2004 15 10
2005 28 8

CAGR 25 12
GSD 19 15
Sharpe 1.12 0.59

Doubled the return of the S&P 500 and nearly doubled the Sharpe Ratio; lost to the S&P only 3 times in the 15-year period. Risk? The volatility is only a few points higher, and I would argue that the blend seems less risky than the S&P 500 over the backtest period: only two losing years, and the largest loss was -2%.

(Results from )

This is widows & orphans stuff. You don't think this meets the criteria of “investing” rather than “speculating”? I'm sorry, this seems like a rock-solid investment operation. Mr Graham would totally approve.

(Caveats: We only have the Benchmark Investing picks in Jamie's backtester from 1991 on: and we only have it for the January start. So this posted result might be artificially inflated from what a longer backtest or an all-months test might show.)

So we can construct an investment composed entirely of MI screens, that does a pretty nice job of keeping principal safe while providing an adequate, or more-than-adequate, return. What you notice about the blend we chose is that we specifically went all-value, leaving momentum stuff entirely out of it. This let us use an annual hold, which keeps trading costs low; and it's also in line with Mr Graham's original stock
screen recommendation, “The Relatively Unpopular Large Company” mentioned in the intro to this FAQ: low PE Dow Jones stocks. If we define “investing” vs “speculating” along the same lines as the definitions Mr Graham proposed some 50 years ago, we see that MI can qualify as “investing”

You might still wonder: does this mean that momentum screens are strictly speculative, and any use of momentum in your portfolio moves your operation
out of the “investing realm” and more into the “speculation” realm? I don't think so. Suppose we construct a blend of 3 screens, including our momentum buddy RS26. The blend will trade 15 stocks total, and rebalancequarterly. For the other 2 screens, let's use YieldEarnYear, which we've mentioned a few times before, and another screen:

Timeliness 1-4 stocks
Long-Term Debt = 0
10% with lowest PEs
Sort by 1-yr return

This screen runs on the ValueLine dataset. It excludes those stocks which VL gives its lowest-rating; and it selects low-PE stocks with zero long-term debt, which have been appreciating in price.

You can see what we did in creating this blend. We've got three “families” represented: momentum, yield, and value/contrarian. If you look at Emintz' chart breaking out screens into groups based on correlation, we've chosen one from the top, one from the group in the middle, and one from the bottom of the chart. Here's the backtest from 1986 on, averaging across starting months:
S&P avg Jan Feb Mar

1986 24 29.3 29 30 29
1987 5 -0.3 7 -3 -5
1988 11 18.3 7 21 27
1989 36 40.3 63 26 32
1990 -5 28.3 6 48 31
1991 31 64.3 77 66 50
1992 7 32.7 37 34 27
1993 10 22.3 23 25 19
1994 2 -1.0 -2 1 -2
1995 39 38.0 36 48 30
1996 23 45.7 57 34 46
1997 27 25.7 28 21 28
1998 33 63.0 91 51 47
1999 20 90.7 116 62 94
2000 -11 30.7 28 30 34
2001 -9 28.7 15 40 31
2002 -22 4.3 11 1 1
2003 29 59.7 54 52 73
2004 10 19.0 19 15 23
2005 8 31.0 18 48 27

Blend CAGR GSD Sharpe

Jan 33 21 1.35
Feb 31 21 1.30
Mar 30 21 1.24
avg 31.3 21.0 1.30

S&P 12 17 0.53

(Results from )

I don't know if Mr Graham would approve quite as whole-heartedly here as he would of the other bend: this port has more-frequent trading, and it's not tilted so completely in the direction of value. But this really doesn't look any more speculative than the all-value annual blend we looked at above. The GSD is a couple percentage points higher, but not much: and the CAGR is noticeably higher along with the Sharpe Ratio. The -3Sigma return (from Jamie's backtester) from both blends is about the same, -25% for one and -26% for the other. The worst year shown in any of the start-month cycles for the quarterly blend was 1987 with a March start: -5%. The blend beat the S&P in 16 out of the 20 years; and it handled the tech downturn 2000-2002 beautifully. What's not to like? To use something like this, you'd have to have some faith that the backtest results are indicative of future returns, and you'd have to be willing to do some frequent trading, and you'd have to have the appetite for a GSD over 20: this isn't necessarily widows & orphans stuff. But if you meet those criteria, if you want to be an active, “enterprising” investor, what's not to like?

By the way, to me this illustrates the key of using momentum screens in an MI portfolio: you keep the momentum in there for the juice it provides, but you blend it with more defensive stuff (value, yield) in order to stay alive during the inevitable corrections that hit momentum stocks.

(By the way #2, note that all three of the screens in this blend – RS26, YieldEarnYear, LowPE_ZLTD – utilize a momentum final sort. RS26 uses 26-week return, the other two both sort on 1-year total return. You could take this as further proof that MI really “Momentum Investing”: even the value screens use momentum in some way! Lots of people require that they have no trace of momentum in their non-momentum screens: so if they were using RS26 in a blend, their other screens must have no momentum sort. This is a reasonable position. My own take is that YieldEarnYear and LowPE_ZLTD cannot possibly be momentum screens, regardless of the final sort. Before the final sort is applied, the LowPE_ZLTD screen restricts itself to ~160 stocks in the VL db that have the lowest PE; the only stocks in the VL dataset that could have PEs lower than these stocks, are for companies which VL rates Timeliness 5 (its worst rating), or stocks with long term debt. These are all value/contrarian stocks, no matter how you sort them. YieldEarnYear restricts itself to the 2% of stocks in the VL dataset with the highest combined earnings yield time dividend yield. 2%! These are all value/yield stocks, and they stay that way no matter how you sort them.)

So. Is MI “investing”?

I think that when we take the care to construct balanced portfolios, using screen blends with different styles represented, with an eye toward creating a decent chance of weathering downturns, MI definitely qualifies as investing. It's an operation which, upon thorough analysis, looks like it will keep largely protect principal while providing a good return. We certainly analyze the hell out of it. We may make mistakes, we may draw faulty conclusions, we may place too much confidence in the predictive value of backtesting: but I don't think anyone can claim we don't analyze this stuff thoroughly. When we go chasing after CAGR by taking on all-momentum blends, we are probably moving more toward speculating.

Is MI gambling? FW # 20100 10/23/2000

So, does MI work?

I had hoped in writing this to make it clear that this question is not answerable in any simple way. We don't really know. We really don't know. And there are many formidable obstacles in place, to our ever being able to know in an unambiguous way: Survivorship Bias, Multiple Hypothesis problem, and so on. We have a number of people on this board who've made nice money over the years using MI. But does that prove MI works? How many people have tried MI and wiped out, never to post here again?
Hell, what would even constitute “proof” that MI works? What does it even mean, to say that MI “works”? Higher CAGR than some market benchmark? Better Sharpe Ratio? Than which benchmark? Over what time frame? For all MI users, or just some who do it “the right way”, or what?

Here's an interesting inversion of the question. What would constitute proof that MI didn't work? Some of our screens show very high GSD: so high that even in the normal course of events we would expect to see large losses. CAPRS 1-5 with monthly rebalancing showed in its 1969-1999 backtest a CAGR of 37 and a GSD of 38. Huge volatility: that backtest showed a -2sigma return of -28% and a -3 sigma return of -48%. Given a normal distribution (which we know is a way overoptimistic assumption), you could expect to see that -48% return about once every 20 years (5% of the time). Every 20 years you lose half your money! In 2001, about a year post-discovery, CAPRS cratered: -42% return, with a -28% return the following year. If you bought in at the beginning of 2001, two years later you were down to about 42% of your starting value.

Is that proof CAPRS doesn't work? Or is that exactly what we'd expect from the backtest? One -48% return in 33 years; two other drops of more than -20%.

How do you grapple with something, where even a loss of more than half of your capital does not constitute “proof” that the investment method is not working as expected??? It reminds me of the difference between science and religion. You know how they say that science is made up of falsifiable statements; that what distinguishes religion from science is that you can't disprove religion? Well what the heck is MI then? Even if you make a lot of money that doesn't prove it works. And if you lose almost 60% of your money in 2 years that's no proof it doesn't work!


“Does MI work?” The question just isn't answerable.

Here's one fabulous post with a sober assessment of MI:

Worth it? My 5 Yr MI Port Results, MI # 155313 1/10/2004
The average CAGR of my Individual Stock Portfolio from 1999 to 2003 was approx. 4% versus (2%) for the SP 500 and 3% for the DOW. … My evaluation of MI is still up in the air. I could have bought the DOW index and achieved almost the same results with much less volatility, taxes, and effort. I am using MI again this year, but have put more of my portfolio into broad index funds… The recent final EOTG Contest Scorecard for 2003 (#155278) is interesting. My guess is the average participant in the contest is extremely smart, put considerable thought into their real portfolio selection, and is above average in MI and investment knowledge. The average gain of their 11 reported real money portfolios was approx. 25%. Their average return was no better than a DOW or SP 500 index fund for 2003 and was achieved with higher volatility! Their non-market beating return is additionally discomforting to me as MI is generally thought to work better in up markets (like 2003?) than down markets.

Here's another perspective:

Backtesting, MI # 60025 2/29/2000
Many of the people following these screens have been investing for a lot of years. Eventually they came to recognize that the stock market is a good place to invest at least some of their money. Most have been burnt with the market timing stuff. And have not had any long term success picking their own stocks. So the next step becomes buying the index fund and let it ride. These screens sure look like a good alternative to the index funds. And I sure don't have a reason to expect a combination of these screens to perform worse than the index fund over a period three to five years. So where is the risk in following them? But I can sure see the potential benefit. And rather than wait for another 15 years for the satisfactory out-of-sample data to evolve, I think I will try to be part of the experiment.

I like that phrase, “part of the experiment.” Another precinct heard from:

Bonferroni-correction, MI # 187792 5/3/2006
In the 60-cycle all-start-days tests by pnyberg, 44% of the published screens outperformed the market using the S&P 500 equal weight total return index (my personal fave benchmark), and over 72 cycles only 18% of screens outperformed the market (this does not include any friction costs, which would of course make things worse for MI). The overall market returns for this period weren't hugely atypical; 8-9.3%/year.
Some folks thing a smaller-cap index is a fairer test. Using the Russell 2000 total return index, the figures are 21% of screens outperforming the market over 72 cycles, and 35% outperforming over 60 cycles. The overall market returns were 7.3%-9.3%/year over this period. Again, this is not including any cost of friction.
Any test under 5 years is pretty iffy, especially since it has been mostly bull market recently, so I consider these figures to be the best answer currently available to the question “does MI work”. The answer so far might be summarized “no sign of it yet”. … As a bottom line, one could make a case that almost all the published results from backtests are wildly optimistic (largely because so many tests have been done, and the bad stuff ignored)

MI Vs. Mutual Funds, MI # 188114 5/9/2006
Those high quality managed funds have kicked my MI's @ss every year since 1999. Roughly half of that was due to my allocation decision in 03&04 (out of MI in May-October). Every fund I've picked in my 401k (4) has beaten my MI blend since 1/1/03. Majority of evidence leads me to believe that bailing on MI where you are now, Ox, would not hurt you in the least. Exceptions abound, of course. The funds are HAINX, FLPSX, FPURX and WFIVX.

MrToast weighs in with this:

Yet Another MI Poll, MI # 187355 4/25/2006
I think MI works, but the evidence is lacking. We'll know later, when we're rich or poor.

Yeah, the evidence is lacking. We'll know better in a decade or two.

Is that all you're gonna say???

I guess if you actually slogged thru the ~45,000 words in this FAQ, and clicked thru to read some of the links I selected, you might feel that you're entitled to know what I think. For the most part I have tried to steer clear of personal opinions, and stick with supportable statements.


For one thing, MI matches my temperament. I'm an algorithm guy. I'm in software by trade, and some of the pastimes I enjoy include fantasy sports: using stats to rank players, etc. I like, no really I love developing algorithms, evaluating them and implementing them. I'm not convinced that I can spot a good investment unaided: that I would know a developing market trend or an undervalued business if it bit me on the ankle. But comparing algorithms, I can do that. So you can see how MI would appeal to me: I like technique, maybe more than I like businesses. Also I gotta admit I like the action. I like trading: not necessarily every week, in fact I can go days without looking at my account; but I do like to keep the portfolio moving a little. I've structured my quarterly MI screens using a Dozens rotation, so I'm trading something every month. So, if MI were only as good as any other reasonable investing technique, my predilections would probably lead me to this method of investing over LTBH on select companies.

It's clear to me personally that there is money to be made with these techniques. Looking just at the results I've seen in my own account, MI picks have produced many stocks that go up 10% or 15% or 20% or even more over a short holding period of just several weeks to a few months. I have fond memories of AMD giving me a +91% in under 3 months, from 10/16/2001 to 1/7/2002. Nice! There's just no way you can't make at least some money using MI techniques.

Yet it's equally clear that you can blow up, too. Looking again at my own account, there are plenty of -15% and -20% losers. Stupid GNSS went down -69% for me, from 1/14/2002 to 4/4/2002; stupid GSIC repeated the favor in the very next holding period. Those gross losses are exceeded on an annualized basis by WTSLA, which needed only one month to drop -36%, 6/15/2001 to 7/13/2001. Ouch. Those losses aren't just a thing of the past, either: stupid ADAM is down -31% for me in just the two months from 3/8/2006.

So you can win a lot, and you can lose a lot, with MI techniques.

Personally? I think this is the place to be. The emphasis on demonstrating that particular methods have worked well in the past, and on risk management, I think these are things that can give a sustainable advantage over the long run. Also there's a phenomenal group of very experienced investors here: the casual wisdom these guys let drop from time to time is very educational. I get exposed to a ton of ideas here, on all aspects of investing; and also to the habit of mind which should help weed out the good ideas from the weak ones. I think the education alone would be valuable, even if I were a 100% discretionary trader. Add to that the chance that the screens will make money – I personally think this is the place to be.

Certain screens and blends show CAGRs over 30% and even 40% in the backtest results. How good or bad is that? What should I shoot for? How should I frame my expectations for my investment returns?

Warren Buffett's annual letter to Berkshire Hathaway shareholders came out recently. His annual letters are archived at the Berkshire web site:

Here's the specific one I'm talking about:

If you haven't read any of Buffett's letters, stop reading my blather and go learn something about investing. Anyway, in the 2005 letter Mr Buffett reports his average annual gain from 1965 thru 2005. That figure is 21.5%. (Over the same period the S&P 500 returned 10.3% per year.) This man is widely acknowledged as one of the finest investors of our time, if not the finest investor, and he clocks in at 22%. So, from this I gather that these 30% and 40% CAGRs that show up in the backtests are nice to look at, but perhaps not what we're actually going to get for ourselves.

If you can get 20% per year returns from your investments over the next few decades, it's likely you will have done superbly for yourself. That kind of performance may be enough to make you one of the best-performing stock investors in the world. I guess we're all trying to shoot for more than that: it's tough to see those backtest numbers and not get greedy. I've put together a blend that I'm hoping gets me over 30%. But to a certain extent that's fantasy. From the beginning of 1950 to the beginning of 2006, the S&P 500 grew from an adjusted price of $16.98 to an adjusted price of $1285.48: that's an average gain of 8.03% per year. I think the range defined by the long-term S&P return and by Mr Buffett's returns ought to help give us a sense of what's possible: 8% to 22%. An MI investor is likely to blow that away in some given year; but then is also likely to give some of it back in some other year. As a practical matter, if you can beat the S&P 500 by a few percentage points over the long term, like decades, then you will have done extremely well.

I think a nice reference point is to look at how long it takes to double your money, given different CAGRs. Consider this table:
CAGR   doubles in
3% 24 years
4% 18
5% 14
6% 12
7.2% 10
8% 9
9% 8
10% 7.3
11% 7
13% 6
15% 5
19% 4
26% 3
42% 2

Mr Buffet has been among the last few rows of this table, since 1965. If you're doubling your money every 3-5 years, over the long term you'll have no cause for complaint.

The MI Appendix, MI # 113184 12/18/2001
MI is not about predicting the overall return of the market. Well, at least some of us think so and we get into periodic flame wars with those who think otherwise. ;-)
MI is about finding inefficiencies in the market that allow us to select a handful of stocks that will beat the market. Whether the market is flat, or grows by 7% or 20%, my goal is to beat it by 5% annually. Anything above that is icing on the cake.

Expectations, MI # 187061 4/19/2006
I plan based on getting something close to the backtested GSD, and returns no better than the market. I expect to beat the market long term by a few percentage points. I hope to do better.

It's only money, right?

This is probably my favorite post here, and one of my favorite threads ever. A little meditation on what matters. I couldn't come up with an actual question that this addresses, but perhaps you won't mind reading it anyway:

The Time Value Of Money, MI # 68176 5/4/2000
People actually do pass assets down across 200 year time frames. What's your plan?

One of the things I personally have been very surprised to discover, and very pleased too, is the way that thinking about investing and thinking about the future helps reconnect you with your values. Investing well requires things like prudence, patience, discipline, consistency: and thus trying to invest well helps grow qualities like prudence, patience, discipline, consistency. Maybe I only notice it because those qualities do not come naturally to me, and are very painfully learnt: but there's a way in which investing-by-method is like training certain character traits. (Or more like boot camp.) It can make you a better person, in some ways. Maybe not for Scrooge: but it seems that way for me, flighty & mercurial as I can sometimes be. That post above is an example of what I mean.

What is rich?

What is Rich? MI # 78911 9/1/2000
Last month I seized an opportunity, and sent my son with a group of local college kids to Nicaragua. They lived with local families in a tiny village near the Honduran border, ate their food, drank their rum, learn a lot of Spanish, and worked all day every day at a camp for people made homeless by Hurricane Mitch. It was a success: my son returned home far richer than when he left.

What do I do when I have enough to retire?

Interesting question, along with the companion question “How much is enough to retire?” Here are some discussions:

Galeno's Mechanical Retirement Strategy, MI # 33394 8/3/1999

Safe Withdrawal Rate, MI # 140499 2/4/2003

Sustainable income from investments, MI # 183988 2/17/2006

Temporary milestone, MI # 187034 4/18/2006

SWR & MI -safe withdrawal rate & MI, MI # 187074 4/19/2006

There's a board at TMF devoted to this, and it has a rep of
having solid info:

Retire Early Home Page


And an off-board site:

What if I die?


So the thing about this type of investing is, this isn't like buying Microsoft & Berkshire and holding for decades. You can't just leave it there until your grieving spouse gets around to hiring a new financial advisor. (Or marrying one.) You have stuff in your MI account that really should be closed out pretty soon. Some of that frothy momentum stuff (I knew you wouldn't be able to resist that stuff!) is going to come tumbling down when it runs out of steam; and god forbid you have a few options in the account, that need to be sold before the time decay really starts to eat away at it. Not that it's likely to happen to you; but what happens to your account, to your family's account if, you know…

The Backup Plan, MI # 177432 11/2/2005
You have just been hit by a bus. Does your significant other know enough about your investing strategy to be able to continue it in your absence?

A backup plan, MI # 185875 3/27/2006
I had a vision the other day of my (hypothetical) wife, posting something on this board some year, along the lines of: “Hi, I'm JimZipCode's widow. There's a note among his stuff that says our IRA is in something called 'SOS YY / TValue, 1-8 Dozens 4-month'. Can anyone tell me what that means, or what I'm supposed to do?? Any help would be appreciated.”

How do you stick to your screens?

This can be harder than you might think.

How to stick to your screens, MI # 72835 6/27/2000
I wanted to try to distill some basic ways to make it emotionally easier to stick to your screens, without impacting the bottom line or deviating from the basic course of “follow the screen”. These are loosely ranked based on what has worked (and not worked) for me. I think most of these ideas are good even if you don't have trouble sticking to the screen.

Do you have to know all this stuff to follow the MI board?

The mechanical investing board itself has a lot of traffic on it. God do we yammer on: sort of like this FAQ. Warrl recently addressed this very well, classifying the general traffic:

The case against small caps, MI #184358 2/23/2006
[the board is]:
• A school (mostly, but not entirely, in/through the FAQ, but occasionally we have to give directions to a new student)
• A research lab - or perhaps several of them
• A stock-picking newsletter - actually several of them
All on the same board, with no clear differentiation between them. I can see how a [new] person could be confused.

Greg Trocchia adds:

The case against small caps, MI # 184340 2/23/2006
We may be giving a highly misleading impression about what is required to do MI investing. One of the things that attracted me to MI, before we were even calling it MI, is the inherent simplicity of this method of investing. You select your screens and holding periods, get the weekly posting of the screens at the appointed time, sell whatever stocks fall off the screens, buy whatever comes on the screens, and hold whatever remains on the screens (rinse and repeat each time a holding period expires). It is way simpler than Rule Makers, Rule Breakers, Hidden Gems, or any emotional (judgment based) investing technique out there. Once you have decided on the screens and holding periods, the rest is just the details of execution. So why does it seem so complex? I'm going to let you in on some of MI's dirty little secrets: The first dirty little secret is that you can safely ignore 95-99% (or more) of the stuff posted here and still operate an MI port successfully. Presidential cycle Monte Carlo simulations, timing signal based leveraged fund switching, 6/3 Options, inverted yield curves, NONE of these things do you need to know the slightest thing about to do MI well.

Why are so many of the posts devoted to stuff which is not at the core of MI? That is the next dirty little secret: We here at the MI board are incorrigible tinkerers. The analogy that comes to me (I love analogizing) is to the tuner culture in the performance car community. For the price of an average car, you can get a car that, out of the box, can humble the supercars of decades past. For most people that would be enough, more than enough, but some buyers get bitten by the horsepower bug and can't resist coming up with “mods” to get yet more extreme levels of performance from their cars. We are the investing counterparts of those folks. We cannot resist the urge to see if we can tune things to get more CAGR here, lower GSD there, or decreased friction over there.
This brings me to the final dirty little secret I will disclose here: If it wasn't for this investing “tuner culture” stuff we wouldn't have much to talk about (much that is on topic, that is). Consider – a couple of my screen stocks are currently Apple and GM. Am I going to cruise Thinksecret, trying to guess the next product Steve Jobs is going to unveil to post here? No. Am I going to debate the merits of GM's recovery plans here? No. Discussing such stuff might lead me to second guess my screens which is a bad thing, so far as MI is concerned. So if you are not going to talk about that sort of stuff what is left for the collection of talent assembled here to discuss? How to “juice” our screens.

Point being: there really doesn't need to be a lot to talk about, just doing MI month to month. Ho hum, I bought some stock I never heard of last month; it's up 2%; I'll be selling it on rebalance day this Friday. What else are you going to say? Greg adds this toward the end of his post:

This is the core of MI: how to do a backtest, how not to mislead yourself with backtests, how to put together a blend that is well suited to your circumstances and personality. Understand these things and you have what you need to jump in and invest using MI. All the rest is stuff at the periphery that you can pay attention to or not as you see fit.

There's an exchange in Jack Schwager's book Stock Market Wizards, in the interview with trader Steve Lescarbeau:

Any advice for novice traders?
Don't confuse activity with accomplishment. … I hardly spend any time trading. Over 99 percent of my time is spent on the computer, doing research.

But we natter away regardless. We tend to generate a lot of traffic about stuff we might do or stuff we want to research or approaches that may be found useful or an interesting article we just read. Don't feel bad if you just skim it, or find that 65% of it is not of interest to you. I personally try to glance at every post; but in very many cases I just skim. And then often a week later there'll be a riveting new post in the thread, and I'll go “Wait, what?” And then go into whole thread mode and read carefully. There will be lots of posts on stuff that you're not doing at all. But then in a couple years you might start doing something new: trading monthly, or buying options, or whatever. At that point you might want to go back and catch up. Don't worry; there's time. Focus on what helps; come back later for the rest when you want it.

Ha. On the other hand, maybe you should dwell on every word here:

In Defense Of Mechanical Investing, MI # 122226 4/10/2002


Closing thoughts:

Elan on efficient markets, MI # 172682 5/27/2005
I like the pseudo scientific aspect [of MI], which says that in an almost efficient market the only investment strategies even worth considering are those that can be shown to have worked in the past

Jim mungofitch on emotional investing, MI # 184675 3/1/2006
Human brains have circuitry (see: behavioral finance) which is not well constructed for the purpose of dispassionate investing. As a result, any investment decision (purchase or sale) which is made mechanically is likely to outperform, being merely random with respect to the price demons, rather than a losing proposition on average. …
It is my belief that the bulk of “excess” returns from mechanical investing, to the extent there are any, come from this insight. This also applies to the mechanical application of a trading system; it's a truism that the best systems in terms of reward and risk are the most emotionally draining to actually implement. This is not a perverse coincidence: it's the reason why they work best, since the victims of the price demons, unable to overcome their pattern detection hardware, are on the other sides of the trades making the investing decisions that seem satisfying. Other things being equal, the more emotionally draining a trade is to make, the more likely it is to be profitable

MindsEye's collection of Foolish words for “times like this”, MI # 66170 4/12/2000
These are some of the Foolish words of wisdom I have read here over the past 2 1/2 years that I think many of which are relevant to remember on down days like today.

Sparfarkle on the psychological challenge of investing, MI# 71758 6/15/2000
One distinct advantage to being a participant on this board, of course, is that you get a sounding board populated by people doing essentially what you're doing in the market. Many of us underestimate the value of that, I think, because investors tend to underestimate the psychological challenge of investing well for long periods--the challenge of having a plan and sticking to it. … I won't be surprised in 15 years to hear at least a few people say that they benefitted enormously from the invention and backtesting of the screens they used, but that they benefitted exponentially more from having someone to pick them up each time they fell down.

Ray on how a man's got to know his limitations, MI # 58638
If you know the limits of your ability, you can make plenty enough money. The further out your limits are, the more you can make. But if you go beyond your limits, you risk losing big time.

Sparfarkle on keeping your head on straight, MI 133863 9/29/2002
Be rational. Don't do any one thing – be diversified instead. Remember that American businesses will very probably go up in value over long periods of time. Don't try to trade or time anything if you don't understand it; sometimes doing nothing is the smartest thing. Don't listen too much to cocktail-party talk. Remember that the talk that goes on on this board has to do with potentially innovative strategies, not necessarily your core strategies. Be proud if you have held on to your most important things during this generationally awful market.

Print the post  


What was Your Dumbest Investment?
Share it with us -- and learn from others' stories of flubs.
When Life Gives You Lemons
We all have had hardships and made poor decisions. The important thing is how we respond and grow. Read the story of a Fool who started from nothing, and looks to gain everything.
Community Home
Speak Your Mind, Start Your Blog, Rate Your Stocks

Community Team Fools - who are those TMF's?
Contact Us
Contact Customer Service and other Fool departments here.
Work for Fools?
Winner of the Washingtonian great places to work, and Glassdoor #1 Company to Work For 2015! Have access to all of TMF's online and email products for FREE, and be paid for your contributions to TMF! Click the link and start your Fool career.