Return to August Contents
TRADING PSYCHOLOGY
Statistics And Trading Behavior
by Ari Kiev and Ken Grant
Use statistics to profile patterns of trading behavior as part of an ongoing training program to help traders maximize their performance.
Traders have habits and beliefs that have significant impact on their trading, whether they realize it or not. Introducing a transaction-level database, though designed primarily for risk management, makes it possible to identify and examine specific and measurable patterns of trading behavior. The database, which we use in our weekly trading seminars, has enabled us to identify characteristics of those traders who are successful and those who are not. It has also led us to a greater understanding of the behavioral components of winning strategies and how to help traders improve their performance.
A variety of data is routinely collected in a relational database of historical information. The data includes such information as the security traded, the buy/sell indicator, the position initiation, liquidation indicator, price, quality and so forth. Other information such as asset class, trading instrument (cash security, option, future, forward), market sector, instrument volatility, beta or other benchmark correlation statistics and average daily number can be superimposed on the database.
From this data, we can infer profit and loss on individual transactions, average transaction sizes, average holding periods, percent of transactions liquidating profitably, average gains on winning transactions, average losses on unprofitable transactions, average profit by mode of execution, average profit by broker or counterparty as well as distribution of profits and losses on longs versus shorts and across sectors. In addition, we can derive the following daily time series regarding traders including the P/L, size of the portfolio (measured in currencies or trade units), number of trades and volatility index (for example, value at risk).
Risk measures may be used to recommend reallocating capital. Profitable traders can be encouraged to get bigger, while those who are not profitable can reduce or redeploy their capital allocations. In effect, these statistical measures can be used for policymaking and strategic planning.
From the viewpoint of risk management, it is useful to find the best ways of combining the performance of portfolio managers where there are negative or statistically insignificant correlations by reducing redundancies. Identifying traders whose portfolios do not have positive correlations allows us to suggest capital allocation alternatives. Where traders are positively correlated with the best performers but whose contribution to the firm's trading profitability may not be optimal, it is possible to redeploy capital to maximize the performance of those contributing the most profitability to the firm.
On reviewing this range of data, it is possible to identify the strengths and weaknesses of traders and to determine patterns of trading behavior that have a negative impact on trading. Doing so can then provide a profile of behaviors that can be identified and then worked with to improve the trader's performance and increase his or her profitability.
Further, we are able to determine such things as whether a trader is overtrading in too many tradables, scalping or trading too rapidly or too slowly and missing opportunities. We can determine whether the trader's losses are more excessive than is warranted, whether he is gambling, whether he is foolhardy or whether he is reasonably cautious.
Whereas the risk manager uses transactions data to determine patterns of profitability and risk assumption in the actual trading of the portfolio managers, the same data has value in understanding the psychological dimensions of trading patterns in different traders. This approach is valuable in finding ways to increase profitability and performance capabilities.
Excerpted from an article originally published in the August 1998 issue of Technical Analysis of STOCKS & COMMODITIES magazine. All rights reserved. © Copyright 1998, Technical Analysis, Inc.