Have you ever read The Millionaire Next Door? It’s an insightful book that looks for common traits among people who have been particularly successful at accumulating wealth.
For example, one trait that’s common among “prodigious accumulators of wealth” is that they’re frequently self-employed. From this, we can conclude that self-employment is likely to increase one’s wealth. Right?
Well, no, we can’t.
What if self-employment simply increases the likelihood of extreme outcomes at both ends of the spectrum? That is, what if self-employment increases not only the likelihood of becoming very wealthy but also the likelihood of going bankrupt? If we only look at the success stories, we have no way to know whether or not that’s the case.
In order to determine whether or not self-employment, on average, increases one’s wealth, we need to do a survey of self-employed people–and not just those who are wealthy.
Same Thing Goes for Picking Mutual Funds.
One approach many investors take to picking mutual funds is to find several funds that have been successful and see what they have in common. For example (and I’m completely making this up), if the top 5 international stock funds over the last 5 years all had the following characteristics at the beginning of that period:
- Expense ratios between 1% and 2%,
- Fund managers with 3-7 years of experience, and
- Less than $1 billion in assets.
…then it would make sense to look for funds that have those characteristics today, right?
Again, no, not necessarily. Because that’s only half the picture.
Going Back in Time
To test whether or not such a fund selection strategy might be successful, we have to go back in time. We have to go back to the beginning of the period in question, look to see what other funds also had the same characteristics, and evaluate their performance as well.*
If all (or nearly all) of the funds with those characteristics performed well, then we might be on to something. (Though even then, it requires a great leap of faith–one I’m personally not comfortable making–to assume that the same pattern will hold true in the future.)
But if half of the funds with those characteristics performed very well and half performed very poorly, then this probably isn’t a great strategy.
In other words, even if all of the top-performing mutual funds share a few characteristics, that doesn’t necessarily mean a darned thing. We also need to check to make sure that those characteristics are underrepresented among poorly-performing funds in order to conclude that they might be useful as predictors of success.
*It’s probably worth pointing out that most of us individual investors don’t even have the resources to do this kind of research. In fact, I’m only aware of two products that make such research possible: CRSP’s Survivorship-Bias-Free US Mutual Fund Database and Morningstar Direct, neither of which is exactly intended for individual investor use.