Another view of the Kelley Criterion....(From Worden Bros.)
We haven't done it before but we are knighting this gentleman for his vigorous and sincere criticism of the Kelly Criterion, as put forth by Sir Gambler in Friday's Worden Report. He gets a little over-zealous at times, but I have often been guilty of the same character flaw, and I'm sure he doesn't mean this as a personal attack on Sir Gambler. It is important that we expose ourselves to all sides of a story. Sir Makis definitely thinks for himself and he thinks well. He belongs at the Roundtable and we welcome him, though we caution him always to wear his armor and shield. (We check swords at the door.) May he enjoy the celebratory bottle of Veuve Clicquot Ponsardin.
Dear Don, With great surprise I read your report on The Kelly Criterion. This is a method that will, in my own opinion, lead anyone who uses it to lose all of his or her trading capital, as it happens eventually to all professional gamblers. This is why:
1. The method makes the tragic assumption that the future marginal profitability rate, or bias towards profitability, can be predicted. This is of course not right since it makes trading an almost deterministic game with guaranteed profits...
2. This very same assumption of a profitability bias is what accounts for the successful compounding. If the assumption is wrong, however, instead of mounting profits you get devastating losses.
3. As with other misapplications of Engineering concepts and methods to trading, the Kelly criterion takes a small initial sample of trades to calculate the winning bias. In my view, there is no sample size of trades that will tell you anything about the profitability of future trades, otherwise everyone would be rich. Furthermore, do not forget "beginner's luck". But if you must take a sample, it must be in the order of 1000 to 5000 trades to be right...
4. Most beginners in trading are "done" by 30 or 50 trades, no time to think about Kelly criterion.
5. Near maximum profitability this type of criterion will cause overtrading with excessive losses when "luck turns against you". By the time risk parameters are adjusted, a good chunk of the trading capital is gone.
In my opinion, the best way to utilize your trading capital is to risk a small, constant percentage of it, every time you trade. That's the good old technique. You set your stops accordingly, although some freedom is allowed to deal with resistance and support levels. Risking in the order of 2% to 5% on each trade is a good practice. For instance, a 2% risk requires 50 consecutive losses to completely wipe out the trading capital.
In my 15 years of trading experience I have realized that the markets love simplicity and reward those that exploit it. Fancy ways of maximizing future gains that you do not have in your pocket right now are punished by the market. The main objective of every serious trader should be to maintain profitability and not try to maximize it with fancy ways, believe me in that, I have tried many. Your service is the best in the world. Thanks. Makis
I do not agree that all gamblers come to a bad end, though I do agree with much of what Sir Makis has to say. The ?sample? of past trading wins and losses is what I termed ?the Achilles heel? of the Kelly Criterion in Friday's Report, and this, put another way, is the thrust of Makis's critique. I think Makis in his paragraph 2 misses the point that compounding results in a contraction of risk when things are going against you ? not increasingly devastating losses. And in paragraph 3, he seems to be denying that strategic systems are not as important as the ability to pick winners, where in fact strategy is by far the more important factor in the long run. He also misses the point that if the past is no indication of the future as he contends in paragraph 1, then 5000 trades is no better than 50. His last two paragraphs take a positive approach and alone justify his Knighthood. As a matter of fact, his proposed method also takes advantage of compounding, since he deploys a fixed percentage of available capital to each trade.
Having thought about it some more, I have about concluded that the Kelly Criterion doesn't transfer to the stock market without modification. I believe that the ?risk factor? should be applied to the portfolio as a whole rather than individual trades. It's a matter of looking at your portfolio as a ?unit? ? like a mutual fund. And I do like the advice given by Sir Makis in his last two paragraphs. -DW |
