May

23

 I attended a presentation by Andrew Lo on 'The Psychology of Trading,' on May 21 2007. Here is a brief summary of his remarks:

On the one hand the Efficient Markets Hypothesis is at the foundation of Finance (for example all the work by Black-Scholes assumes that the market for options is efficient) on the other hand many people nowadays find it hard to believe that EMH is literally true. This has led to the development of Behavioral Finance, which studies biases that may hinder financial decision making. BF has acceptance problems of its own: it brings up so many possible biases that it is hard to believe (if all these biases are true) that anyone is ever able to make a correct decision. Many economists ask if the behavioral biases even exist.

To try to advance beyond the EMH/BF debate, Andrew Lo has been working on his own framework, which he calls the Adaptive Market Hypothesis, and has been investigating the role of emotion in trading by reading the neuropsycholgy literature and conducting experiments, some of which will be described below.

Do perceptual biases really exist

The first experiment involved the audience. They were invited to watch a video showing college students, some in white T shirts and some in black T-shirts throwing basketballs to each other. Lo told the audience to concentrate on the white T-shirt players and count the number of times they passed the ball to each other. The exercise was made more difficult by the fact that the black-T shirt players intermingle with the white shirt players and that Lo kept talking throughout the video to try to confuse the audience.
After the video was over Lo asked "how many people saw the gorilla?". More than half the people in the audience had not seen any gorilla. [I personally did not see the gorilla, even though I knew that Andy Lo is famous at MIT for showing a video in which a gorilla appears !]. Lo replayed the tape, and sure enough a man dressed as a black gorilla walks through the scene halfway into the tape. Lo explained that people who are concentrating on white figures will often miss black objects; in some sense the human perceptual system is filtering out the black objects.

In conclusion, said Lo, in any debate between economists and psychologists as to whether perceptual biases really exist is going to be won by the psychologists, who have demonstrated these phenomena beyond doubt through careful experiments.

The neuropsychology literature

The book "Descartes Error" by A. Damasio has changed how we view rationality. The classical philosophers believed emotion and rationality were polar opposites. Damasio investigated people who have suffered serious brain injuries and found that people who do not perceive emotions correctly will act irrationally. Emotion is necessary for rational behavior, Damasio says. Emotions allow you to choose quickly and easily among the many choices constantly available to you, saving you time and allowing you to zero in on correct solutions to problems.

The Triune Model of the Brain was proposed by Paul McLean. The human brain is made up of three parts: -the brain stem, which controls basic functions such as breathing and wakefulness is the oldest part of the brain, philogenically speaking. It exists in reptiles as well as in higher life forms. -the midbrain is involved in emotions (such as fear and greed, sexual preference and so on). It exists in mammals. -the neocortex controls higher functions, is the seat of thinking, language, etc. and exists only in hominids. There is a definite order of priority among these three subsystems; a painful stimulus for example will disrupt the processing functions of the neocortex for several hours according to experiments in which blood flow to the brain is measured via MRI scans. When a lower level is activated it disrupts (or takes priority over) the higher level mental functions.

From a financial point of view it is clear that risk-preferences and decisions under risk arise from interactions between the midbrain and the neocortex. Rational decision involves a balance and/or cooperation between the emotional and calculating parts of the brain.

Experiments in neuropsychology and finance

(1) Studying professional traders as they go about their job. Lo attached sensors to traders to measure emotional responses. (2) Lo also interviewed 80 neophyte traders who were learning to trade in a class given by LBR and reviewed their trading

Conclusions

a. emotion is definitely involved in trading decision making, even in the case of experienced decision makers (i.e. it is not solely the beginners who experience these emotions). However, the emotions are somewhat more controlled among the more experienced or more able decision makers. b. traders who experience little emotion during trading have a lower P&L, however traders who experience a great deal of emotion during trading also have a lower P&L. It appears that there is an optimum level of emotion somewhere in the middle. c. people who excessively internalize the outcomes (i.e. attribute everything that happens to their own doing) have a lower P&L, however people who attribute everything to luck also have a lower P&L. Again there appears to be a proper balance, i.e an attitude that events are partly due to ability and partly to luck.

The Adaptive Markets Hypothesis

The AMH takes a biological/evolutionary view of markets, whereas the EMH took a physical/engineering view.

The AMH postulates that financial decision makers:
1. act in their own self-interest
2. make mistakes
3. learn and adapt (through heuristics, not through optimization)
4. competition drives adaptation and innovation
5. natural selection drives the ecology of the markets
6. evolution drives market dynamics

With regard to point 3. Lo has a high regard for Herbert Simon and his idea of "safisficing" (not optimizing), and of making decisions through simplified (and non-optimal) heuristics (since an optimal decision is computationally infeasible). A question that Simon could never answer is "where do heuristics come from", but Lo thinks the answer is that "evolution determines heuristics" (point 5). He did have not elaborate on this. Lo expressed the view that Simon's work is even more important than the Theory of Rational Expectations, even though it has received less attention in economics.

The AMH implies that anomalies can appear, disappear and then reappear again as the market ecology changes. For example the profits to Statarb have waxed and waned over the last 15 years. It is true that the profits have dropped sharply after the Summer of 2002, but this does not mean that hey have been permanently arbitraged away. Statarb profits may not be gone forever, they may come back at some time under different market conditions than what we have now.

So far the AMH is incomplete. Lo is working on extending it and convincing others.

MAIN CONCLUSION

We need emotions to be rational. We need both, it is not either/or. In trading there is a right level of emotion. There is a "right zone". The Zen of Trading.

Jean Paul Schmetz writes:

It is almost impossible to see the gorilla [see video] if you concentrate on the players. I have tried this video with 100+ people in the room and very rarely did more than 10% see it. It usually helps to mention beforehand that males or females are better than the other at keeping score (you do not tell which and so people concentrate even more). 

Chris Hammond adds: 

There is an article in Scientific American this month that discusses a game called the "Trader's Dilemma," which is a variant of an older game called "Prisoner's Dilemma." Experiments involving this game address some of the issues mentioned earlier. There is, of course, some extraneous background story, but essentially, the game works as follows: two people pick a dollar amount between 2 and 100. The smaller of these two amounts will be awarded to both players, except the person who chooses the smaller amount will receive an additional $2 and the person who chooses the larger amount will receive $2 less than the lower of the two amounts.

If the players pick the same amount, then there is no penalty or reward. If you assume that both players are working solely for their own self-interest and make rational decisions, then both will pick $2. However, in an experiment where the range was between 80 and 200 cents, and the penalty/reward varied between 5 and 80 cents, the player's average choice was never the Nash equilibrium of 80 cents. For the 5-cent penalty, it was 180, and when the penalty was 20 cents, it was 120. This particular study used economics students. A similar study used game theorists, and the results were similar.

More interesting is what happens when people play the game repeatedly. Apparently, for "large" rewards, the amount that is picked as people play many times tends towards the low number, 80. For "small" rewards, the amount moved towards the high, 200. The article is not more specific than that.

Here, a system evolves as participants learn. It's also interesting that there is a bifurcation at some particular reward amount, and that the system evolves completely differently on either side of that value. Also worth noting is that not even game theorists think like game theorists.

Yishen Kuik writes:

I have been reading E.O. Wilson's Consilience, which has a nice chapter on what we know about the brain.

He also has a good chapter on what it takes to be a productive scientist engaged in the business of discovery, some of which seemed to me to be uncannily applicable to describing the business of uncovering statistical edges to trading. 


Comments

Name

Email

Website

Speak your mind

Archives

Resources & Links

Search