Monday, May 21, 2007

Assessing Trader Personality: The Role of Trader Psychometrics

Are there personality patterns common to successful traders? This is a question that interests aspiring traders and trading firms alike. The usual way of addressing the issue is to offer traders a set of pencil-and-paper questionnaires that evaluate personality traits and then see if the results correlate with trading outcomes.

There is much to be said for the questionnaire method of assessing personality and performance. Questionnaires are easy to administer and score. There are also highly reliable, valid, and standardized measures of personality traits in the psychological literature, such as the NEO PI-R based upon the work of Costa and McCrae. Research coming out of the London Business School has found that personality traits--including overconfidence and sensation-seeking--are significantly associated with trading results. This corresponds to my own, more informal, findings that high degrees of neuroticism, coupled with sensation-seeking and low conscientiousness, are associated with poor trading outcomes.

There are several potential weaknesses to the questionnaire approach, however. First, and most fundamentally, it is not at all clear how much of the variance in real world trading outcomes is accounted for by self-reported traits. My investigation with Andrew Lo's research team at MIT did not find any strong association in a sample of traders. Nor is it clear that the universe of traders is homogeneous, with a single set of optimal personality traits. Just as different personality styles are associated with specialties within medicine--psychiatry vs. surgery vs. radiology--it may well be the case that different personality patterns typify successful market makers vs. portfolio managers vs. mechanical system traders.

Equally problematic, questionnaires tend to be face valid; it is obvious to respondents what is being asked of them. This makes it easy to manipulate the results, particularly in employment settings where test-takers will provide the most socially desirable responses.

For these reasons, it can be worthwhile to supplement the use of questionnaires with measures of trader psychometrics. As David Norman of the Illinois Institute of Technology indicates, we can learn about traders from an assessment of their trading, just as we can learn about trading by asking questions of traders. Dr. Norman's Trader DNA program is an effort to collect metrics on one's trades, so that traders can identify their own patterns of successful and unsuccessful trading.

Having utilized this approach in my own work with traders, I would like to propose that trader psychometrics can reveal several important dimensions of trader personality that are useful supplements to the information gleaned from questionnaires:

1) Frequency of Trading - I propose that this is a manifestation of a trader's need for stimulation and tendency toward sensation-seeking. The trader who needs/desires to be active in the marketplace (e.g., the day-trader) displays different needs for stimulation than the long-term investor;

2) Size of Positions Relative to Portfolio Size - This is a direct measure of risk-taking vs. risk assumption;

3) Average Size of Losing Trades vs. Winning Trades - This assesses the prudence or conscientiousness of the trader in managing risk. Traders with large outliers among losing trades display a lack of control over risk.

4) Number of Different Positions and Strategies Employed - This is a measure of behavioral complexity, which in turn has been found to be associated with superior performance in a variety of work settings. Such complexity may also prove to offer a degree of insulation from changing market conditions.

The trader psychometrics address many of the shortcomings of pencil-and-paper self-report personality questionnaires. They are easily gathered by computer and need no psychological interpretation and administration. They represent objective, observable data that are not subject to subjective interpretive bias. The metrics also cannot be faked by traders.

Most important of all, however, the trader psychometrics create a process-based assessment. We can see how patterns of trading change over varying market conditions and across different markets and strategies. The research of MacCrimmon and colleagues, for example, found that risk-assumption has a very strong situational component: people who are risk-takers in one situation (personal life, for example) may be risk averse in other situations (finances). The trader psychometrics view personality as it is manifested in the specific trading context. We can detect personality at work as trading proceeds. This makes trading metrics a fine supplement to more global personality measures.

Finally, an advantage of trader psychometrics is that they can be collected by traders themselves and used to identify patterns of success and failure. For instance, traders can track their trading and learn whether they become more or less risk averse as markets become more volatile; whether they become more prudent over time; etc. In this sense, trader psychometrics are both assessment tool and learning tool.

Your personality affects your trading, but your trading also reflects your personality. A well-rounded approach to assessment looks at the trader from both angles, identifying elements of performance associated with the person and with the trading situation.

RELATED POSTS:

How Personality Affects Trading Performance

Steps Toward Becoming a Better Trader

A Personality Questionnaire for Traders

Interpreting the Personality Questionnaire

4 comments:

cpptrader said...

Brett,
Great post. These psychometrics could also be applied to gather information on automated systems. I can see specifically where their measurement can indicate whether the judgements you put into the system are valid or not. Generating metrics on trades, such as outliers, # of trades, position sizing, etc. is common, sure, but this provides a different aspect - how do my views affect the system?
Thanks, Matt cpptrader.com

Brett Steenbarger, Ph.D. said...

Hi Matt,

Very, very good point. One implication I didn't get into is that, if metrics can assess trading personality, then every system (and its metrics) has a personality as well. The psychometrics are very relevant to system traders and especially to hybrid traders who might add elements of discretion to trading rules.

Brett

Gurvinder said...

Can you please put the Printing options on your blogs so that one can easily printouts and read on paper

Thanks
GSA

Brett Steenbarger, Ph.D. said...

Hi GSA,

I have restored full syndication of the blog so that people can print out the posts from their readers. I'll maintain the full syndication unless, as before, I encounter trouble with websites posting the content without my approval. Thanks--

Brett