A recent research study reported in the New York Times (January 20, 2004; Subconsciously, Athletes May Play Like Statisticians by David Leonhardt
offered intriguing insights into how athletes are able to make complex decisions in a split-second fashion.
www.nytimes.com/2004/01/20/health/20TENN.html
Consider a couple of examples that I have assembled:
- A major league baseball batter faces a pitch thrown at over 80 miles per hour. The pitch is headed directly for the batter's head. If it is a curve ball, the pitch may break down and away for a strike. In that case, the batter will want to stand his ground and take a swing. If it is a straight fastball, the pitch may strike the batter and end his career. In a couple of moments, the batter must assess numerous variables to determine whether to stay in the batter's box and swing or duck and protect himself. The speed and rotation of the ball, the movement of the pitcher's hand, knowledge of the pitcher's repertoire-all to into an equation that is solved faster than it can be verbalized.
- Col. John Boyd studied the behavior of “top gun” fighter pilots of military aircraft and found that, during dogfights with enemy planes, they make sophisticated decisions in a matter of seconds. He described the decision-making process of Observe, Orient, Decide, and Act (OODA), and became convinced that the key to training was to accelerate the OODA loops so that pilots could respond more quickly and accurately than the enemy. A top pilot himself, Boyd was famous for his challenge that he could start from a position of disadvantage in the air and get behind the tail of any pilot who challenged him within 13 seconds-a challenge he apparently never lost.
What makes these examples worthy of consideration is that the expert performer is facing a high degree of uncertainty. There is no way of predicting in advance whether a pitcher will throw one pitch or another, and there is no way of anticipating one's next adversary in the air. Once the situation arises, there is also no opportunity for calm, rational, explicit reasoning through the situation. The subconscious mind must assess the relevant variables and make the optimum decision before any explicit reasoning occurs.
How is this possible?
Here's a snippet from the New York Times article that describes the recent research:
Each participant in the experiment sat down and placed a hand on a tabletop. A projection of a computer screen blocked their view of the hand. The goal was to guide a cursor, which followed the movement of the hand, from one side of the screen to a target on the other side.
Adding to the uncertainty, the cursor usually appeared slightly to the right of the hand, and the participants caught at most a quick glimpse of it when it was halfway across the screen. Sometimes, the cursor appeared as a discrete point; other times, it was an ill-defined cloud.
The researchers found that when no cursor flashed, people relied on what they had learned during 1,000 practice runs before the experiment: namely that the cursor was, on average, one centimeter to the right of the hand. When a cloud flashed, they considered it, but only somewhat, in a pattern that followed what Bayes's formula predicted. When a distinct cursor flashed, they relied on it and not past experience.
"Most decisions in our lives are done in the presence of uncertainty," Dr. Körding said. "In all these cases, the prior knowledge we have can be very helpful. If the brain works in the Bayesian way, it would optimally use the prior knowledge."
There are several important pieces of information from the research:
- To make their decisions, subjects needed a prolonged learning period. It took 1000 practice runs to enable subjects to make accurate predictions of the location of the cursor. This is consistent with the “implicit learning” research, which has found that subjects who are exposed to complex patterns for thousands of trials eventually learn those patterns and can make predictions based on those patterns at levels far better than would be expected by chance. Amazingly, however, they cannot verbalize those patterns or how they are making their decisions. Their learning is truly implicit-processed entirely by the subconscious mind.
- When subjects encountered new, uncertain information that contradicted the patterns they had learned for 1000 trials (the cloud), they weighted the new information in a complex manner that was consistent with Bayesian mathematics. Stated in practical terms, the subjects gave increasing weight to the new information over time, revising their estimates of the cursor's location in a way that integrated the new, uncertain information with what they had learned previously. This also is consistent with the implicit learning research literature. People can process information in complex, mathematical ways, even when they are not familiar with complex mathematics!
- When subjects encountered new, certain information that contradicted the patterns they had learned for 1000 trials, they abandoned their previous learning and went entirely with the new data. That is, they were able to put aside their prior expectations and quickly create a new basis for decision-making. This is not as easy as it may seem. People have a natural “confirmation bias” that leads us to seek out information that supports our expectations and ignore data that contradict what we believe. Here's a link to an academic article on the subject of confirmation bias: www.peel.pitt.edu/esa2003/papers/wolfe_confirmationbias.pdf What is particularly interesting in this article is that the authors find that the confirmation bias prevents people from making decisions in a Bayesian fashion. That is, people do not accurately integrate new, uncertain information with old experience-the way experts do-when they stay anchored to their old expectations. This makes them too slow to react to new data. Experts are thus not only able to make lightning, subconscious decisions; they can also quickly revise those decisions as new data arrive. Fast and flexible are hallmarks of expertise.
There are very important implications for short-term trading in this research. One key conclusion is that the distinction between intuitive, discretionary trading and research-based, quantitative trading may be smaller than we think. Here is the outstanding finding of the research:
Our subconscious mind is a quant!
When expert traders have been exposed to thousands of examples of intraday price and indicator patterns, their subconscious minds are able to extract patterns from the noise in a manner that they themselves cannot verbalize. This is implicit learning. As new information from the market arrives second by second, the subconscious mind integrates it with the old knowledge in a complex, Bayesian fashion. One or two pieces of discrepant information-one or a few ticks against one's swing position, for example-will not necessarily trigger a decision to abandon the trade. As the discrepant information accumulates, however, it changes the expert trader's estimates of the likelihood that the trade will succeed. The trade starts to feel wrong before the trader can put his finger on what's not right. That gut impression comes from an important place-and is not just a subjective intuition. It is the result of a rigorous (subconscious) mathematical analysis!
When new information arrives to the expert trader that obviously contradicts his or her expectations (such as a sudden market reaction to a piece of economic news), the prior learning goes out the window. The expert trader is quick to exit the trade and revise expectations, rather than wallow in a confirmation bias that now fights the tape.
What keeps traders from becoming expert performers? How can we become good Bayesians? If Colonel Boyd can train a top gun to analyze an enemy fighter pilot's actions and outmaneuver than enemy in 13 seconds, perhaps we can learn to accurately respond to short-term market patterns.
In Part One of this two part series, I pointed out research that suggests that the subconscious processing of athletic performers displays a Bayesian quality, weighting recent events against past ones to respond to shifting patterns of circumstances. When a market shifts volatility, for example, the expert trader must integrate the new data with his or her large database of experience to anticipate the next tick or the next directional move. How much weight to give the new experience, and how much to alter that weighting with each fresh experience, is the task of the Bayesian. It appears that the subconscious mind is capable of performing sophisticated integrations of new and old data to anticipate future events. This allows for peak performance under fast, stressful conditions that do not permit conscious, explicit processing, such as those faced by fighter pilots, professional boxers, and scalpers.
Can this subconscious basis for expertise be cultivated? In an earlier article drawn from the training of elite Army Rangers, I looked at one possible model for facilitating superior performance. Perhaps we can also learn from the training of elite athletes and develop models for improving the performance of traders.
Research in Sports Psychology
What do we know about the psychological factors that contribute to success among athletes? A comprehensive research review reveals several important ingredients of superior performance:
- Goal-Setting - Over 500 studies find that setting goals facilitates performance across a variety of short-term tasks and long-term objectives. Goals that are highly specific, achievable but challenging, and performance rather than outcome-focused tend to be most effective. There is also evidence that blending short- and long-term goals, as well as group and individual goals, can enhance effectiveness in achievement-oriented settings.
- Practice - A wide range of studies of athletes, as well as experts in other fields, finds that cumulative practice is closely related to the development of expertise. This is especially the case where there is implicit learning-situations requiring the acquisition of performance skills that cannot be verbalized (such as learning to hit a tennis ball). Large numbers of practice trials with immediate feedback are particularly helpful to skill development.
- Arousal - Research suggests that moderate degrees of stress aid performance, while high distress can produce catastrophic declines in performance. Recent findings indicate that it is how anxiety is processed-how each person views their own stress-that is more important to performance than the absolute degree of physiological arousal.
- Self-Efficacy - The athlete's beliefs about his or her own capacity to perform is significantly related to actual performance. The belief in one's own capacity to succeed is the single most important personality factor associated with athletic success. Studies find that interventions that improve self-efficacy tend to improve athletic performance, even when they involve mental practice (imagery) rather than actual athletic performance.
- Coaching - Recent research finds that the self-efficacy beliefs of coaches are significantly and positively correlated with improvements in the performance of their athletes. Differences in coaching style also meaningfully affect athletic performance, primarily by influencing the motivation of the athlete.
How are these different findings related to the research on subconscious learning?
A plausible explanation is that goal-oriented intensive practice with rapid feedback not only builds skills, but also increases the self-efficacy of athletes. This allows the elite athlete to use arousal to facilitate rather than interfere with performance, as stress is viewed non-threateningly. The role of the coach is less to teach skills-since the skills are implicit-than to create the conditions under which skill acquisition can be maximized. These include motivational conditions, but also conditions related to the effectiveness of practice sessions.
It is worth noting that much of this elite training takes place in team environments, including sports teams, military units, and business settings. The team setting assists with motivation by creating a supportive environment for demanding training, but also allows participants to learn from each other through observation and peer coaching. Friendly competition between teams also serves as a preparation for actual competition, honing skills under realistic performance conditions.
Summary
The success with which a trader can learn to read ever-changing market patterns may be related to the quality and extent of training he or she undertakes. Every major study of implicit learning finds that it takes thousands of concentrated learning trials before an individual develops mastery of a task. Most traders fail, I would suggest, because they never develop the critical mass of intensive learning trials needed for subconscious expertise. Without a team environment to model skills, motivate skill development, and provide goal-oriented coaching, the majority of individual traders may never develop the self-efficacy needed to weather inevitable periods of loss and flat performance. This is particularly problematic for the part-time trader, who may lack the intensive practice needed for mastery simply as a function of the reduced amount of time available to internalize market patterns, rehearse execution skills, and utilize feedback.
While much market writing focuses on the development of new indicators or methods for analyzing market data, it may well be the case that the most promising avenues for improving trading performance are tools and technologies that supercharge learning and accelerate the learning curve.
Brett N. Steenbarger, Ph.D. is Associate Professor of Psychiatry and Behavioral Sciences at SUNY Upstate Medical University in Syracuse, NY. He is also an active trader and writes occasional feature articles on market psychology for MSN's Money site (www.moneycentral.com). The author of The Psychology of Trading (Wiley; January, 2003), Dr. Steenbarger has published over 50 peer-reviewed articles and book chapters on short-term approaches to behavioral change. His new, co-edited book The Art and Science of Brief Therapy (American Psychiatric Press) is due for publication during the first half of 2004. Many of Dr. Steenbarger's articles and trading strategies are archived on his website, www.brettsteenbarger.com.
