Posted by David D'Aquin on May 24, 1997 at 09:28:45:

For those who do not see the difference between proposed Kelly bet spreads and traditional ones that are simply equal to advantage / variance, the entire thing has to do with the minimum and maximum bet in the spread only. All other bets are traditional Kelly bets. The biggest and smallest are exceptions. The smallest takes place in negative expectation situations and the largest must be some sort of composite average. The method that I used to determine whether one is over or underbetting for a proposed spread is not really different from Winston's. The only difference in the method of evaluation is average bet vs unit bet. My f divided by the average bet will be Winston's f. Winston's opitmal bank should be 1 / f, while I used average bet / f. But these are really only systems of evaluating a spread, not of finding the optimal spread. Both brh and Winston say that they find the optimal spread through a process of trial and error on a spread sheet. This doesn't really help much. If it is possible to do this, then it must be possible to find a formula to find it. The basic question is: For a given game and a given minumum and maximum bet size, expressed in number of units, at what count do you make your maximum bet? All the other bets are easy, advantage / variance for the count and a miniumum bet in negative expectations. But until that question is answered I don't think we have really gotten anywhere.