Extraordinarily average

Regression to the mean is a well documented phenomenon in investing, as it is in many other areas of life. 

It refers to the basic tendency of a statistical series to move toward average over time. Just as trees don’t grow to the sky and average IQ’s don’t double in a generation, returns on equity investments tend to move toward average, too.  

As I described in my January blog, this phenomenon can be used to profitable advantage, and now seems to be one of those times.  

Why do I think that? Because group performance of some stocks seem to be far out of line from long term history, and if those relationships are restored–as they always seem to be–then money can be made investing in a way that supports a regression to the mean.

To flesh this out, let me describe one way of looking at historic returns: by assessing them relative to company size and valuation. For example, you can group large companies together and small companies together, or look at companies with high price to earnings (growth) or low price to earnings (value).  

One popular way to look at this, used commercially by Morningstar and described academically in a famous study by Fama and French, is to break companies by size and valuation in to four groups: 1) small value, 2) small growth, 3) large value and 4) large growth.

I like to look at this data over 5 year periods to observe whether one group is moving further from average and may snap back.

Over the very long term, from 1927 to 2011, small value has been the best performing group, followed by large value, large growth and small growth. In fact, over 84 years of data, small value did 3.7% better than average, large value did 0.5% better than average, large growth did -1.7% worse than average, and small growth did -2.4% worse than average.  

If you’ve ever heard someone say buy small companies or value companies, they were referring to this long term data set.

But, there is a problem with just buying small value without further thought, because your particular returns depend on how far from average the data is at the time you purchase. If small value has done much better than average when you buy, it won’t generate historically average results for you. And, if small value usually does best but has been worst recently, you’ll probably do much better than historic average.

Knowing the long term average and recent performance can lead to profitable discoveries, in other words.

With this in mind, how has recent performance looked? Instead of small value leading the pack over the last 5 years, small growth has. The 84 year historic worst group–small growth–has been best over the last 5 years (+4.9% better than historic average). In my opinion, this seems like a terrible time to buy small growth.

Large growth, usually the second worst group, has been the second best group with 3% better returns than usual. Small value, usually the best group, has been the second worst, -2.5% worse than historic average. Large value, historically the second best group, has under-performed by -5.4%!

Does this mean something fundamental has changed, or will regression to the mean bring things back to average. Both analysis and experience leads me to believe that regression to the mean will occur like it always has, making small growth a poor place to invest and large growth the belle of the ball over the next 5 years.

Will this smooth transition begin the minute I publish this blog? Almost certainly not. And, that’s the rub: regression to the mean happens over the long term, not the short term. We don’t know when things will regress, only that they almost certainly will.

Will large value greatly out-perform small growth? I don’t know how soon or by how much, but I am very comfortable betting it will over time. As always, I’ve put my money where my mouth (blog?) is.

Nothing in this blog should be considered investment, financial, tax, or legal advice. The opinions, estimates and projections contained herein are subject to change without notice. Information throughout this blog has been obtained from sources believed to be accurate and reliable, but such accuracy cannot be guaranteed.

Extraordinarily average