Portfolio construction and position sizing are key elements in investing. For every investor, there exists a tension between the desire to maximize the rate of growth of the portfolio while simultaneously minimizing the chance of blowing up. The Kelly Criterion is the method to determine the optimal portion of the portfolio to be invested in any given opportunity. Buffett, Munger, Whitman and Pabrai are all proponents of the theory.
John L. Kelly, Jr, the developer of the Kelly Criterion, seems to have been a remarkable character. According to his entry in Wikipedia, he was a physicist, “recreational gunslinger”, daredevil pilot, developed the vocoder, the first demonstration of which was the inspiration for the HAL 9000 computer in the film 2001: A Space Odyssey, and was a keen blackjack and roulette player, which is a little odd, because his criterion recommends against a bet on the roulette wheel. He died of a brain hemorrhage on a Manhattan sidewalk at age 41, never having used his formula to make money.
The Kelly Criterion output varies depending on two things: the investor’s certainty about the outcome of the investment (the “edge”) and the expected return (the “odds”). I have found it difficult to apply in practice. Hunter at Distressed Debt Investing has a great post on Peter Lupoff’s application of Kelly Theory to event-driven investing in Tiburon Capital Management’s portfolio. Lupoff’s post deals with some of the issues I have had, and is well worth reading.
[…] we’re prepared to put up with a little extra volatility for a little extra return, we can also Kelly weight our best ideas). Here are the equal weight return statistics for dividend […]
LikeLike
[…] we’re prepared to put up with a little extra volatility for a little extra return, we can also Kelly weight our best ideas). Here are the equal weight return statistics for book […]
LikeLike
[…] prepared to put up with a little extra volatility for a little extra return, we can also Kelly weight our best ideas). Here are the equal weight return statistics for the cashflow […]
LikeLike
[…] dividing the total portfolio capital over the total number of positions, say 10 to 30 stocks) or Kelly weight our best ideas. The equal weight returns are therefore more useful for most investors. For equal […]
LikeLike
Good article. According to Wiki (which I rarely trust), the Kelly Criterion was developed in 1956. I have a document entitled “Valuation Techniques” by Benjamin Graham that was published in 1947. The equation is nearly identical to the Kelly Criterion. Therefore, it is my belief that Graham composed the equation long before Kelly did and although Pabrai has written about the Kelly Formula in his book, I believe Buffett, Munger, Whitman, and especially Schloss (as I have Schloss documents using the formula) all used Benjamin Grahams formula which is virtually the same but preceded the Kelly Formula.
It is:
G = Gain in points if successful
L = Loss in points if unsuccessful
C = Chance of success in %
Y = Time of holding in years
P = Current Price
(GC) – L(100-C) / YP = %
LikeLike
The big difference between Graham’s formula and Kelly’s is that Graham’s formula doesn’t tell you how much of your portfolio should be allocated towards the investment, it only gives its chance of success.
LikeLike