Stock Markets Exposed

How do you apply the Kelly Criterion to stock investing?

How do you apply the Kelly Criterion to stock investing? Any ideas of how to do this?

Public Comments

1. I've done lots of testing using Kelly Criterion in my trading. The real problem is, you cannot accuratly calculate probabilities of winning the trade, how much you will win (odds on wager), etc. I've toyed with the idea of simply putting a profit target and stop loss even distances from my entry, say .50 away, and then make the assumption that you have 50% probability of winning, and the odds are 1:1. However this is an inaccurate assumption, as your stop order will trigger immediatley when your price is hit, but the price has to go almost through your limit price before getting a fill. This means that the probability is actually less that 50%. Hope this is helpful!
2. It doesn't really apply since investing outcomes aren't binary but you're on to something with that kind of thinking. I'm reading The Dhando Investor by Mohnish Pabrai. He offers up this website for a free calculation of a derivation of the Kelly Formula: http://cisiova.com/betsizing.asp It's a derivation because the Kelly Formula is binary, either win or lose.
3. You should do a search for "Ed Thorp." He's probably the best known practitioner of applying Kelly to stock trading. He has written a number of papers publicly available on the subject and are quite interesting, if abstruse. To use Kelly with a multi-asset portfolio, you need three things - the expected return, the volatility, and the covariance between your prospective stock holdings. Then you have to implement some linear algebra to arrive at the "optimal" portfolio construction. Obviously, this is not straightforward. You can also look up papers by William Ziemba. He covers the same subject, which he describes as capital growth theory.