PsyFi Search

Tuesday 6 September 2011

Are Traders Dumber than Drooling Dogs?

Salivating Spaniels

Given the ebb and flow of investor sentiment, often for no other apparent reason than the ring of the opening and closing bells of the market, it’s terribly tempting to compare traders to the salivating dogs from Pavlov’s famous experiment.  The dogs, of course, were trained to salivate at the ring of a bell in the expectation of a good meal.

Naturally, this would be demeaning, a gross insult to the intelligence of our canine friends.  Because it turns out that there’s quite a lot of good evidence that humans investors are, despite their superior intellectual capabilities, dumber than drooling dogs.

Reinforcement Learning

Ivan Petrovich Pavlov’s experiments were amongst the first in modern psychology, and involved the discovery that you could train dogs through reinforcement.  His dogs were taught to associate a ringing bell with a good meal, which eventually led them into salivating in the expectation of food at the sound of the bell in the absence of anything edible.  Psychology has long since abandoned a simple model of reinforcement learning, recognising that people are driven by a whole set of complex beliefs and behavioral biases rather than simple stimulus-response models.  

Still, there’s a suspicion that, when we’re faced with complex situations characterised by uncertainty, we fall back on simpler models of learning to create the beliefs that drive our behaviour.  Investing in the stockmarket certainly qualifies in this regard, uncertainty is pretty much everywhere, all the time: even if people choose not to recognise this for long periods.

Pride and Regret

Initial studies of investment behaviour identified standard behavioral biases as causes of irrational risk seeking and avoidance: overconfidence drives over-trading and a fear of regret drives an unwillingness to sell loser stocks, and so on.  This is the stock in trade of behavioral finance.

So, for example, Odean, et al. in Once Burned, Twice Shy find that the way investors handle stock repurchases is profoundly irrational.  They purchase stocks they previously sold for a profit at a lower price, but not those they sold at a loss.  Meanwhile they’ll purchase more shares of a stock they hold at a lower price, but not at a higher one.  The researchers attribute this behaviour to avoidance of regret – averaging down lowers the break-even price, while not buying back at a higher price allows the investor to avoid regret at their previous actions. None of this behaviour improves investor returns.  However, there’s a suspicion that these types of explanations may be more complex than required.  To understand this we need to revisit Pavlov style reinforcement learning.

Depression Babies

There’s considerable evidence that the experience of your upbringing influences the way that you invest.  Malmendier and Nagel in Depression Babies: Do Macroeconomic Experiences Affect  Risk-Taking? showed that people who experienced low stock market returns throughout their lives were less willing to take on risk and participate in markets.  In reverse people who’ve experienced low returns from bonds are equally unwilling to invest in fixed income.  The research also shows that actual experience matters much more than book learning: people are unable to abstract away from their own experiences, no matter how atypical or lucky.  Moreover, the researchers argue:
“We show that higher experienced stock returns are associated with more optimistic beliefs about future stock returns. This suggests that the experience effects could be the result of individuals’ attempts to learn from their experiences, albeit not by using all “available” historical data, as in standard rational and boundedly rational learning models, but by focusing on their life-time experiences.”
If so then apparently irrational behaviour isn’t caused by some invisible set of biases encoded into the brain, but is caused by reinforcement learning. When whole groups of people grow up under the same general conditions – a Great Depression or a Goldilocks Era – they all exhibit the same sorts of bias.

Power Law of Practice

Now if investors are drooling dogs and not biased bozos there are a few clues we can look for.  High among these would be evidence of the Power Law of Practice: an overly complex way of saying that people should become better at investing the more they do it. They should improve as they learn what works and what doesn’t and the people who aren’t any good should drop out, leading to a survivorship effect where the only people left standing are either good investors or completely and utterly deluded (see Don’t Lose Money in the Stupid Corner).

These effects are difficult to discern because we have new entrants to the market all the time but research by Choi and colleagues suggests that they’re observable: basically, if you’re skilful or lucky enough to get high returns or low volatility in your portfolio you’ll continue to invest.  They explain:
“The observed patterns are explained by a naïve reinforcement learning heuristic: assets in which one has personally experienced success are expected to be successful in the future.  Consistent with reinforcement learning, we also find evidence for the Power Law of Practice: return chasing and volatility avoidance decline with age as a large stock of experience is accumulated, though they remain present throughout the lifecycle.”
Learning Reinforcement Bias

As you might expect, given the vast range of biases we’ve covered in the past, all of this rampant drooling, tail-wagging behaviour has led to the proposal for yet another one: learning reinforcement biasDe, et al also propose that the biggest driver of future investor behaviour is their past experiences of losses or gains, but it’s not the size of these that matter but whether they’re profitable or not: so, success increases an investor’s wealth, which increases their appetite for risk taking, showing up in greater trading volumes, and so on.

However, this comes at a cost: the investors in their research collectively passed $2.31 billion over to the securities industry during the sample period.  They suggest the reason for this is that: 
“Individual investors who are relatively less influenced by the reinforcement learning bias are more successful than others. If institutional investors are rational, or at least less biassed than individual investors, then reinforcement learning bias could provide an explanation for the wealth transfer observed in our case.”
Which helps explains why the securities industry is hugely profitable and traders, generally, aren't. The bell, and most of the experience, is on the side of the professionals and most of the drooling is on the side of the traders.

Overtrading is not Irrational

This idea, that traders are driven by a combination of their own skill, or lack of it, and the random luck of their experiences is expounded upon in some fascinating research by Juhani Linnainmaa: Why Do (Some) Households Trade So Much?  The answer proposed is a bit surprising: people don’t trade too much because they’re irrational and overconfident but because they’re rational and trying to figure out whether they’re any good as investors or not.  It's learning reinforcement and not behavioral bias.  As he explains:
“Investors learn from their own experiences, and the resulting evolution of beliefs is an important driver of investor behavior.”
Which adds to the evidence that the longer people trade the better they should become, as they build up greater stocks of expertise.  Which is exactly the result uncovered  by Seru, Shumway and Stoffman who showed that greater experience  reduces the propensity of investors to fall for the disposition effect (see Disposed to Lose Money) – selling winners and keeping losers; to boost pride and reduce regret – and that trading style changes with age, moving towards larger stocks and a greater emphasis on diversification.  Encouragingly for all those investors suffering today:
“Learning is particularly strong among specific groups of investors, including unsophisticated investors, investors who start out earning consistently poor returns, and females. Investors also learn faster during general market downturns.” (my bold)
Essentially losing money when you start out gives you a better grounding than making it: you learn the proper lessons.  And, as we’ve seen before, being female gives you a head start (see Investors, Embrace Your Feminine Side).

All of which suggests that although we’re mostly drooling dogs when it comes to learning about investing, we’d be better off finding out a bit more from books and a bit less from experience.  The trouble with experience is that it’s an expensive way of learning: although you won’t find many people in the securities industry telling you that, as they happily ring the bells and rake in your dollars. 

Pavlov's Dogs and Schrödinger's Cat: Scenes from the living laboratory (Popular Science)Wealth of Experience: Real Investors on What Works and What Doesn'tBehavioural Finance: Insights into Irrational Minds and Markets


  1. a quote from your post.. "They purchase stocks they previously sold for a profit at a lower price, but not those they sold at a loss."

    that seems rational.. you buy when you think the market is mispricing the share. with your winner, you buy believing the new lower price is an undervaluation, ie diverging from true value. but with your loser you agree that there's a continuing correction of the market's [and your own] previous overvaluation, ie converging on true value. the market is an environment of opinions, thoughts, prejudices, data etc among which you move as an investor: what i think you're suggesting is that it's irrational to attribute a correct reading of it to your own skill, and an incorrect one to your own ignorance. which would be true only if the opinions, data etc were in a state of random churn.. so how far is that true?

    and another one: "Meanwhile they’ll purchase more shares of a stock they hold at a lower price, but not at a higher one. The researchers attribute this behaviour to avoidance of regret – averaging down lowers the break-even price, while not buying back at a higher price allows the investor to avoid regret at their previous actions."

    again, you buy when you think the market is wrong. not averaging up is perfectly rational if you think a firm's NAV growth isn't keeping pace with the share price growth.. say an investor buys XXX at 100p, believing the NAV/share is 200p, 50% of true value; if the share price has fallen /1.5 to 66.7p, 66% underpriced, he rationally buys more; but if risen *1.5 to 150p, only 25% underpriced, he doesn't, having less to gain from averaging up. but your quote seems to assume the investor had no idea in his mind of true value. & in that case i don't know what the point of the quote is. what i want is to understand the mental flaws of a reasonably engaged investor, not of a bingo player.
    but maybe i misunderstood you..

    a bit off topic for your essay, but i'd say that even a moderately stupid pi is more intelligent than a managed-fund investor. i just read the annual brochure of a big fund manager, in which the main holdings of their various high-growth, high-income, childrens funds etc etc were laid out: they were all almost identical, and the childrens fund heavy on tobacco.

  2. Another excellent article from your superb blog. Some real food for thought there.

    As with some of your previous articles, I will post a link to this item on my linkedin group 'trader, trading, and risk psychology', I think I may also poat it on the 'Behavioural Finance - theory and practice' group, where it may illicit some interesting debate.

    I hope this is acceptable to you. Please let me know if for any reason should this not be so.

    Once again thank you for your excellent work on this blog.



  3. Hi Gooner70 - No problem, anything for a fellow Gunner :)

    On the previous anonymous comment it's a point I've some sympathy with: after all, the definition of rationality in economics is a bit odd by normal standards. However, the underlying research seems quite sound: investors are reluctant to repurchase stocks they previously sold for a loss regardless of what happens afterwards, but they'll repurchase stocks that they previously sold for a profit if they've dropped. I probably didn't make this sufficiently clear in the article.

    Now, of course, it's possible to construct a case to say that this is logical in some cases. However, this is a general finding, that investors generally behave this way en-masse and it yields no financial benefit. It's difficult, I think, to construct an argument that this behavior is always correct and rational in these circumstances and a behavioral explanation - e.g. investors are acting to avoid feelings of regret by avoiding stocks they've previously lost on - seems a more parsimonious explanation in the circumstances.

  4. bit late at night for rational thought, but here goes..
    i read a paper you quote, 'Once Burned, Twice Shy'. here's a quote: They find, for example, that investors are more likely to buy stocks that are in the news, simply because they think about those stocks. Investors are also more likely to think about stocks that they’ve owned in the past year than about other stocks. Thus, it is not
    surprising that investors tend to repurchase stocks they previously owned and sold (as well as stocks that they still own, but have not sold) at much higher rates than they repurchase stocks that they have never owned
    from which i conclud that your researchers aren't as smart as they could be.
    is the economist's definition of rationality just Profitability?

    best wishes, i learn a lot from your blog..

    but, to continue. i think there's a wish to, kind of, merge with the zeitgeist when rebuying at a lower price. on the 1 hand you're applying your previously amassed knowledge of the company, but also it's a momentum trade... you [I] think that you're, at last, at one with the market's random fluctuations. ie i took a real bath the 2nd time i bought petroceltic. but james fisher turned out well. so there.

  5. Hi Timarr

    Thanks for your reply, glad to hear you're a fellow Gooner.

    I'd like to respond on some of the points 'x' brings up. Before I do I would like to stress that my personal history is as a trader and not an investor or portfolio manager. As such my experience is mostly based on taking leveraged positions in FX and Fixed Income, and not running any sort of portfolio.

    With regard to averaging in at a lower price, this is something I avoided doing, though did sometimes get caught doing it, particularly in my earlier years. I have known many people do this, most unfortunately have come a cropper eventually. - I say eventually, because sometimes it works. But when it does not work it can be devastating, partly because the trader in adding to their position at lower prices tended to display greater confidence in their view, which would then morph into overconfidence which would lead to a dangerous personal attachment to their view.

    What I am saying, is that there are times when it does and will work, however it is a dangerous practice that is fraught with risk. The trader/investor who practices this (and there will be people who practice this, for whom it works) needs to display the utmost rigour and discipline. From experience I would always advise against it, my most successful trades have involved averaging-in at better prices after the move has started to head in the right direction when it is gaining momentum.

    With regard to trading familiar stocks, I think 'x' is spot on with this point. One knows the territory and character of these particular stocks, having followed them, a person may intuitively (as a result of experience and familiarity) know how and when they offer superior value, over and above the face value analysis available on all and every stock.

    I would also like to take issue with the article as to what it defines as rational and irrational. I often feel these are totally misued terms that economists (traditional and behavioural)apply. - Every trader and investor acts to increase their wealth and value of the their portfolio, whatever they do. That is the most rational behaviour possible. If they acted deliberately and with concious intent not do so, then that would be irrational. - Further to this, I think behavioural economists should be replacing the terms with words such as optimal and sub-optimal. For exaample, I happen to believe that the behaviour of averaging in at lower prices described above is sub-optimal, but certainly not irrational.

  6. Hi Gooner70

    I'll leave the debate on averaging down for now. On familiarity there's absolutely no doubt that this is an issue. Consider, for instance: Puke: Don't Invest in the Familar and Fooled by Fluency.

    As regards the definition of economic rationality, I've written extensively about how this doesn't make sense in the real world. See, for example, Unpredictably Rational.

    However, in the article the use of "rational" and "irrational" is correct in the context. Redefining these to mean something different, however logical, would cause even more confusion. The alternative is the Humpty Dumpty effect, which wouldn't be much use to anyone: "When I use a word," Humpty Dumpty said in rather a scornful tone, "it means just what I choose it to mean -- neither more nor less."

    However, I do disagree that the motivation for all investors is simply to increase their wealth. See, Love Your Kids, Not Your Stocks. People are far more interesting than economists would like to think. And therein lies the joy of life :)