Thursday, November 06, 2008

Research With Practical Relevance: Learning by Trading

I recently posted on the topic of learning how to trade. Thanks to an alert and very helpful reader for passing along this link to a recently published paper on the topic. The paper is entitled "Learning by Trading", and it draws upon an extensive database of trading results from Finland to arrive at several fascinating conclusions about how market participants learn to trade.

One of the first observations from the study is that traders learn trading from experience, but also learn from experience that they lack trading ability and thus stop trading. Attrition is common in trading, and failure to take it into account makes it look easier than it is to learn trading. There is a built-in survivorship bias any time we observe traders: we're more likely to see those who have learned how to trade than those who have learned that they are not able traders.

A second fascinating observation from the study is that traders are unusually susceptible to the disposition effect. By tracking the buying and selling behavior of the traders in the database, the authors discovered that traders are 2.8 times as likely to sell their stock after it has risen than after it has fallen. Moreover, the correlation between a trader's disposition effect in one year and in the year previous is .36, suggesting that this bias tends to persist. Most interesting of all, when traders with the lowest disposition tendency bought stock, the stock averaged a gain of 46 basis points (bp; or .46%) over the following 30 days. When traders with the highest disposition tendency (i.e., those most likely to sell winners, not losers) bought stock, their shares averaged a loss of 54bp over the next 30 days. It's a dramatic illustration of how the bias to hold losers and take quick profits damages profitability over time.

Finally, the authors come to a very interesting conclusion. They note that a subset of skilled traders does appear to "learn by doing". What could look like overtrading to an outside observer might actually be rational, as developing traders recognize that they learn from experience. Significantly, trading skill was a function of the number of trades placed, not the number of years spent trading. This has meaningful implications for the training of intraday prop traders (who trade high frequency) versus longer term hedge fund portfolio managers and investors.

There are a number of fascinating studies conducted by academic departments of finance and financial engineering that have significant practical relevance to the trading community. Unfortunately, as in other fields such as my own discipline of psychology, there tends to be a gap between what is known through research and what is applied via practice. I'm pleased to announce that I am joining the board of advisers for Kent State's Master of Science Program in Financial Engineering. It's an excellent program, and I hope my involvement will bring me closer to the world of cutting edge research--and its application to real-world trading.