View Single Post
Old 07-08-2017, 11:38 PM   #72
Dorianmode
always learning
 
Dorianmode's Avatar
 
Join Date: Jul 2011
Location: Minneapolis / Rancho Santa Fe
Posts: 277
Kelly Criterion

http://en.wikipedia.org/wiki/John_Larry_Kelly,_Jr.

One thing I meant to mention was, that the Kelly Criterion (number) that his formula will produce, (which he either discovered, or realized worked .... he was an engineer for the great, famous Bell Labs, ... he noticed that the formula for the most efficient way to transmit electricity in electrical lines was also a formula for the most efficient way to grow a pool of investment money). Anyway the number the formula kicks out is WAY to high to use on a practical basis for actual wagering ... and this has long been recognized. Huey says in his piece to bet some fraction of it. For example, if your 2 horse win rate is 60 %, and your payoff rate for those is 1.20, then the Kelly number is 0.27. No one in their right mind would bet 27 % of their bankroll. So it's a very useful number to tell you how (and if) a wagering line of data is profitable, and how much so, but just as a disclaimer ... use the numbers to see how profitable your race sets are, not to determine how much of a bankroll to actually bet ... as if you're good, the Kelly can be high (like 50 % or more, and that's not a way to wager).
Dorianmode is offline   Reply With Quote