GM Nigel Davies Improves Himself

December 14, 2006 |

I started reading Dr. Brett Steenbarger’s Enhancing Trader Performance this morning and it’s looking like one of the best guides to self-improvement I’ve ever read and in any field.

Never before have I seen it explicitly stated that one must love the field in which one hopes to succeed and that the process of improvement is driven by this.

Here are two great sentences from page 36 which sum up most of what good chess players have done to acquire mastery: “Explore. Play.”

Bill Rafter relates his rowing experience:

As a former elite-level rower who had gone the “metrics” route, I could reduce success in that sport at the elite level to 2 dimensions. The dimensions could be monitored in real time, and the athlete capable of achieving the best ratio of the two dimensions would win. The problem with the market is that there are N dimensions, with N being a large number, and that number being subject to considerable disagreement among practitioners. Also, “winning” in a boat race or other sport is relatively easily defined. I have been trading for decades, and regularly change my own definition of winning.

Some of the most accepted market metrics are poorly chosen. Sharpe on his own site points out times when the Sharpe Ratio is inappropriate, yet others follow it blindly. My suggestion is that you pick you own “winning” metric, and target that.

“Tis with our judgment as our watches; none run alike, but each believes his own.” -Alexander PopeBut get ready for a lot of testing, as N is a very large number indeed.

Jim Sogi adds:

Dr. Brett writes about recording and studying performance metrics in his book. Though he is dealing with high frequency traders, some good measurement and statistical analysis of trades made, would be very helpful to fine tune and improve the performance of any trader. He claims that this can improve performance and help to avoid blow-ups.

The joke here is that I do not like to watch movies of myself surfing or hear recordings of my music because in my own mind I was making huge dynamic turns on a wave, and sounded like John Lennon on the guitar, but when I see or hear the actual recording, in reality it was just a disappointing knee high wave and the singing is all off key. A legend in my own mind — as Jim Lackey would say, “Get the Joke.”

Sharpe ratios, return, max drawdown, and time to break-even are good basic measures of overall return path, but are there better ways to measure trade by trade statistics? Is there software that will compile or record metrics even on the fly, or from the brokers statements or the trading platform? Do you exit too soon, enter too soon, or leave too much on the table. Is the performance well suited to the current market cycle or are you fighting it? Are you getting killed on breakouts after 3 years of ranges? The stats would need to include not only the trade, but what the market did either before or after, and during your trade, to see how much is left on the table, how much slop there is in your entry, how much wasted commission, how much unnecessary vig. did you pay, time of day issues, seasonal issues, personal issues, were you hold too long, not following systems, was there efficient use of capital?

The top athletes now use scientific study of their performance metrics, and the fine tuning of their mechanics. Baseball players have a detailed set of stats. Any suggestions for a trader to compile his stats on either metrics to consider, or software?

G.M. Nigel Davies sheds further light on the chess world:

In chess the main thing is rating, the numbers don’t lie. If you play well then this is reflected in your rating performance. People often take ‘a view’ on what they did well or badly, but this tends to be corrupted by subjectivity even if they try to be scientific/objective about it. How should one rate trading performance? For short term trading how about calculating the t-score of weekly returns as a percentage of trading capital? A fixed number of weeks would be good with the performance rolling forward over time. Longer term traders might use monthly returns to calculate t-scores. Probably this would be deeply unpopular as people find out how random their trading is. But it has the merits of simplicity, objectivity and zero reliance on anyone’s ‘judgement’.

Jim Sogi raises some open questions and adds a new stat technique to the mix:

Diebold also spoke about distinguishing between alpha with beta. Are you adding value or just adding more risk and just got lucky? As The GM said, the T score of your returns combined with yield vs. the market should sort the beta out. Looking at the stats shows how hard it was to beat the market in the 3rd and 4th quarters where anything less than leveraged buy and hold would have underperformed the 10%, 140 point rise in the S&P.

Performance optimization is a multifactor problem and will vary depending on the Bayes utility factor which ought to be factored in as a risk preference metric. The formula would solve for the highest the amount of change (but not more) needed in various performance factors in order to achieve the greatest increase in the absolute and relative yields relative to the amount of risk preference. It might have similarities to maximum risk formulas with the addition of performance factors.


WordPress database error: [Table './dailyspeculations_com_@002d_dailywordpress/wp_comments' is marked as crashed and last (automatic?) repair failed]
SELECT * FROM wp_comments WHERE comment_post_ID = '188' AND comment_approved = '1' ORDER BY comment_date




Speak your mind


Resources & Links