PsyFi Search

Saturday 13 March 2010

Value in Mean Reversion?

Many shall be restored that are now fallen; and many shall fall that are now in honour (Horace, Ars Poetica)

From Horace to Graham and Doddsville

The quote above is now a couple of thousand years old but was used by Ben Graham and David Dodds in their seminal book on value investing, Security Analysis, a term invented by the book's title. At root it’s a simple plea for understanding that current market conditions, no matter how placid or tempestuous, will pass. The job of the conscientious investor is, at worst, to ignore the short-term forecasts or, at best, to take advantage of them.

Some psychological quirk means that a small subset of humanity latches onto this concept instantaneously and holds to it, like an investing life preserver, through thick and thin. The rest of us either learn the slow, hard and painful way or, more likely, continue to be storm-tossed. Occasionally someone gets washed up on a tropical paradise and is accounted a genius but mostly we drown, quietly, where no one can see us waving.

Mean Reversion is Mental

Value investing is the behavioural finance equivalent of Tantric sex; it takes a long time to work and requires considerable abstinence for long periods of time. But the end result is extremely satisfying. Well, so I’ve heard.

The point, the main point and possibly the only point about value investing is that it’s as much a psychological exercise as a technical one. Of course, we can’t hope to be great investors without hard work but most of us can be better than average merely by learning to control our emotions – which means building a financial platform that can withstand the worst excesses that the markets can throw at us.

The Thaler-De Bondt Hypothesis

In essence, mean reversion theories state that stocks that have previously done badly will tend to outperform and those that have done well will underperform. Much of the recent work on this topic originates with Richard Thaler and Werner De Bondt – see, for instance, Does The Stock Market Overreact? and Further Evidence On Investor Overreaction And Stock Market Seasonality:
“... the paper provides new evidence consistent with the simple behavioural view that investors overreact to short-term (i.e. a few years) earnings movements. Certainly, within the framework of the efficient markets hypothesis, it is distinctly puzzling that a dramatic fall (rise) in stock prices is predictive of a subsequent rise (fall) in company-specific earnings.”
Subsequent studies have confirmed this effect. In Death, Taxes and Reversion to the Mean, a report by Michael J. Mauboussin for Legg Mason Capital Management, we find:
"Various studies conducted over multiple decades document this reversion-to-the-mean pattern. We have reproduced the results here, using data from over 1000 non-financial companies from 1997 to 2006 ... We start by ranking companies into quintiles based on their 1997 ROIC. We then follow the median ROIC for the five cohorts through 2006. While all of the returns do not settle at the cost of capital (roughly eight percent) in 2006, they clearly migrate toward that level".
To summarise: mean-reversion is alive and well. The study's well worth reading for some ideas about predicting companies that don't mean revert. We'll look at that another day.

People Don’t Do Stats

If the idea behind mean reversion is so obvious then why doesn’t everyone eventually adopt it? Roughly we can point to a couple of issues. Firstly human psychology isn’t built around statistical concepts like mean reversion, we prefer to trend follow rather than go against the flow. Secondly, in the bubble of time in which we exist from moment to moment it’s all too easy to convince ourselves that the mean itself has changed: indeed this does happen from time to time – so what may have been an acceptable valuation criteria in the 1930’s or the 1990’s will prove to be far from the average in the 2010’s.

It’s this inherent uncertainty that makes value investing so difficult, along with the market’s tendency to stay irrationally high or low for extended periods of time. As social creatures we’re programmed to take our cues from other people and when these are sending strong signals about what we should be doing then it’s terribly hard to stay aloof and plough our own furrow.

Mean Regression Alone Won’t Make You a Good Investor

As we explored in Regression to the Mean: Of Nazis and Investment Analysts, mean regression won’t automatically make you a better investor – a below average analyst will get below average results. The idea that investing is somehow easy and that anyone can consistently make above trend results through some process that doesn’t involve thinking until the blood seeps out of the pores in your forehead is utter nonsense.

In essence performance depends on a combination of skill and luck. The rare combination of great skill and good fortune can lead an investor to amazing returns. Aggregating no skill and bad luck leads to horrendous ones. Most of us lie somewhere in the middle. However, luck has a habit of evening out over long enough periods: investor performance mean reverts as well as company performance.

Based on what we can see from studies of investor returns and mutual fund cashflows we can be pretty certain that the majority of people – and a significant majority at that – fail to even match the indices. Behaviourally induced overtrading, disaster myopia, lack of emotional control, poor use of gearing and a failure to manage the costs associated with dealing all contribute to terrible results, to the extent that most people would be better off using passive funds for investing purposes and spending their leisure time doing less self-harming stuff. You know, things like do-it-yourself body piercing with a rusty nail and trimming their fingernails with a chain-saw.

Mean Reversion Makes Momentum Trading Successful

Yet, as ever in the murky psychological waters of finance, you can't find a commonly understood trend without finding someone betting against it, often accidentally. The observation that momentum trading, buying stocks that are winners or shorting those that are losers, suceeds over short-time scales of up to a year or so has led some researchers to wonder whether this is also a facet of mean reversion and human psychology.

Kevin Wang, in Mean-Reversion and Momentum, looked at this, the idea being that the wide-spread knowledge of mean-reversion effects may lead investors to sell winners, expecting them to fall back, and buy losers for opposite reasons. Intuitively, he points out, this seems to be wrong - companies which have released good results might reasonably be expected to carry on benefiting from whatever trend they're riding for a while: the invisible hand may be inevitable, but it's not instantaneous.

This is exactly what he found when he tested this simple behavioral theory. Note also that this has nothing to do with the probabilities of actual returns - it's simply investors reacting to changes in stock prices, not the underlying earnings and fundamentals. It's emotion and behavioral bias, not reason and dispassionate analysis: people appear to be reacting to a popular idea, a so-called meme rather than logically assessing a corporation's real outlook.

Emotional Management is Value Investing

All of which suggests that investors can lose out long-term and short-term by not appreciating mean-reversion effects and just goes to show that we can no more separate our emotions from our investing than we can separate our minds from our bodies. However, once our emotions are out of control so are our investments, no matter what returns we’re making and what popular ideas we're following. There’s skill – in terms of the technical business of flushing out undervalued stocks – and there’s art – in terms of learning to control ourselves by insisting that we Get an Emotional Margin of Safety.

Mostly investment gurus major on the skill aspect of investing – which is fair enough, because it’s the bit you can teach. Value investing based around an understanding of mean reversion is one of the few ways of making money through active investing – for the private investor it may be the only way. However, the art to being successful is to make sure you're in control of yourself because when things go wrong that’s just about the only thing that matters.

Related Articles: A Sideways Look At ... Behavioral Bias, A Sideways Look At ... The Randomness of Markets, Buy and Hold, the Least Worst Option?


  1. mean regression won’t automatically make you a better investor – a below average analyst will get below average results.

    This is not an issue if you invest in index funds.

    Combining index-fund investing with Reversion-to-the-Mean investing is the best of all worlds. There's no need to do any analysis whatsoever -- everything is built into the price of an index with the exception of overvaluation or undervaluation. And, with a Reversion-to-the-Mean filter added, you've got the overvaluation and undervaluation problems covered too.

    This must give you better than average returns over the long run. As you note in the article, the majority of investors are buying and selling at the wrong times. If most are getting worse than average returns, those who are keeping their risk level steady (by lowering or increasing their stock allocation as needed in response to price changes) must get better than average returns.

    The only way this would ever change would be is all investors got with the program. But in that event price volatility would be greatly diminished and we would have a far more stable and productive economy. So we would all get better returns that way.

    This appears to me to be the proverbial free lunch. With dessert!


  2. Hi Rob

    Agreed. All I'd add to this is that dollar-cost averaging using indexers will get you a lot of this benefit without the dangers of behavioral reversion.

    Although I, like you, believe it's possible to outperform this benchmark I think it's difficult to do without unusual emotional focus. From personal experience I'm not convinced that knowledge is enough to reach an investing Nirvana :)

  3. I'm not convinced that knowledge is enough to reach an investing Nirvana :)

    We are in 100 percent agreement that knowledge alone will not do the trick, Tim.

    My view is that we need to supplement knowledge of what works with emotional encouragement to do what works. We do that by promoting what works in articles and books and calculators and discussions boards and blogs and all that sort of thing.

    People don't see today how much better investing could be if we only started doing what we have never done before. After we do it, we will look back and wonder why it took us so long to decide to do so!

    People didn't appreciate all that computers could do for them until they started using them. You've got to take the first step to have any hope of realizing the benefits of the 50th step.


  4. To save your auto loan deal from going haywire it is important you take care of it at its root. The root is at the financing and the insurance office.

    Payday loans

  5. It took me 10 years to learn to control my emotions in investing, but one I did, I became a reasonably good value investor. It is not for everyone, though, and I like it that way.

  6. I really don't think that by combining the Combining index-fund investing with Reversion-to-the-Mean investing is the best of all worlds. i think the m,market is pretty v volatile and nothing can be predicted.