Thursday, July 05, 2007

Overconfidence and Underconfidence in Trading: Biases in Processing Emotions

I'm reading an older, but excellent summary of research called Well-Being: The Foundations of Hedonic Psychology, edited by Kahneman, Diener, and Schwartz. The large volume covers what we know about positive and negative emotion from personality, social, and biological perspectives. The gist of many of the chapters is that our processing of emotions tends to be biased, and these biases affect our perception, judgment, and behavior. Many of these biases raise particular challenges for traders and investors in financial markets.

Kahneman's chapter on "Objective Happiness" introduces a bias consistently supported in the research literature. Let's say we track a person over time and every so often ask this individual to report her emotional experience at that moment. This experience sampling method is one way that we can track emotional ups and downs through the day and connect those to thoughts, events, and behaviors occurring at that time. We might even try to derive a measure of objective happiness by averaging ratings of emotions over that period of time.

A different method for studying emotional experience is simple self-report. We can ask the individual to look back on a particular day and rate her emotional experience. Self-report is the method most often employed by psychologists, no doubt because it is far easier to obtain the desired information. Unfortunately, however, self-report turns out to be quite biased.

Our self-reported emotion for a given period of time, Kahneman reports, is relatively insensitive to the duration of the period being rated. For instance, let's say I am rating the pain from my recent bout of appendicitis. The amount of time that I was in pain does not significantly affect my retrospective self-report. Nor is my self-report an average of the pain ratings I might have made on, say, an hourly basis over a two-day period. Rather, my self-reported pain is influenced by only two variables:

1) The peak amount of pain that I experienced during the episode; and

2) The final pain that I experienced.

This "peak-end rule" appears to apply to a variety of emotional experiences. Our self-reported emotional experience is greatly biased by moments of intense emotion and by our most recent emotions.

Here's a simple application to trading:

On Monday I placed a particularly poor trade. Despite the consistent strength in the NYSE TICK and despite the low volume of the pre-holiday trade, I tried to fade the S&P 500 Index. The position sat and did little for a considerable period of time, owing to the light volume. I muttered to myself that it wasn't worth trading this market. Just at that time, a program trade hit the tape and helped lift the market over a full S&P point. Now I'm particularly frustrated, having just given up the gains from a prior trade during a period when I shouldn't have been in the market at all. Grimly, I held the position because I saw no upside follow-through to the program trade. Over the course of the next half hour, the market retraced and, as soon as I was able to take a small profit, I did so. That mercifully ended my trading day.

Later that day, I caught myself feeling good about my recent trading--and then wondering why the hell I was so pleased. Yes, I finished the day at a new equity curve peak for the last several years and that's always nice. But I really had traded Monday's market poorly. During the trade on Monday, I was miserable. I knew I was in a bad trade. Because my size was small, however, and because the market didn't explode in the wake of the program trade, I never hit a peak level of extreme pain. Moreover, I ended the trade green for the day and so didn't experience an ending level of discomfort. My subsequent emotional processing of the day was far more positive than warranted--a situation that could easily have caused me to enter Tuesday's trading in an overconfident state.

Now let's imagine the opposite scenario. Suppose I place three good trades--each with a historical edge in my favor--and size those positions prudently. Each of the three goes against me and I stop myself out at my chosen level to avoid catastrophic drawdowns. If I trade once a day and average 60% winning trades, I will be likely to encounter such strings of three consecutive losers eight or more times during a year simply as a matter of chance. But what will my emotional experience be over those three days? Instead of feeling positive about how I'm trading, my peak negative emotion and my final emotional state are apt to be one and the same. Frustrated over not making money despite doing the best I can, I will report more negative emotion than an average of momentary experiences over the three days would warrant. Out of that negative emotion, I could enter the next day in an underconfident state, reducing my trading size just when I'm likely to get my trading back on track.

The peak-end rule bias reported by Kahneman helps to explain why traders can oscillate between periods of overconfidence and underconfidence. We base our confidence, not so much on the process of our trading or even on the average of our emotional experiences, but on peak and recent events. A recent big winner can lead us to over-rate our positive emotion and make us overoptimistic in subsequent trading. Similarly, a recent losing streak can lead us to conclude that we're in a "slump" and no longer able to trade well.

Is there a way out of the dilemma? It turns out that simple self-awareness might be all that is needed. In one study, experimenters asked subjects to rate their life happiness. Subjects who were asked the question on a rainy day rated their life happiness lower than those asked on a sunny day. It's a clear example of how our emotional judgments are biased by our most recent experience. When, however, the experimenters first raised the issue of weather with a simple comment, such as "It's a nice day today", the weather no longer biased the subjects' responses.

Similarly, by keeping peak and final emotional experiences in mind and reminding ourselves of our entire emotional experience (e.g., reminding myself how miserable I felt when I was in a bad trade even as I'm feeling self-congratulatory), we can minimize biasing effects. In a sense, we want to fade our own emotional responses: our processing biases suggests that things may not be as bad as we're feeling when we're really down or as good as we're feeling when we're euphoric.

Moreover, by sizing positions appropriately for our account sizes and by adhering to prudent rules that keep risk and reward aligned (including stop-loss levels), we ensure that we never hit peak levels of pain that can traumatize us and greatly bias our subsequent judgment. Outsized emotional experience--positive or negative--is more likely to sway our perception and behavior than moderate levels of feeling. Sound money management is perhaps the most powerful form of psychological prevention among traders.


The Most Important Question to Ask When You're in a Slump

Why It's So Easy to Lose Money in the Markets

Ten Lessons I've Learned From Traders