wanted for murder: formula that combines winning % + action points

[Dark Horse]

It's out there somewhere. And could be very deadly. For books.

Anybody else interested in trying to tackle this problem (for purpose of identifying a working formula)?

[Justin7]

I don't know of any books that still take action points. You can crush these long-term with a basic database analysis of games and spreads.

[Dark Horse]

I was also thinking about bet sizing.

There's something lost in translation between the mathematically perfect Kelly formula and the real life world of sports betting. This is admitted by anyone who bets fractional Kelly. There's an uncertainty principle, related to short term fluctuations of good and bad luck. Action points can wipe out that uncertainty.

If the advantage offered by action points can be translated into a mathematical formula, then that formula could be applied to upgrade Kelly from, for instance, 1/3rd to 1/2, or even to full Kelly.

[filter15]

I wish I knew what you were talking about

[roasthawg]

I was also thinking about bet sizing.

There's something lost in translation between the mathematically perfect Kelly formula and the real life world of sports betting. This is admitted by anyone who bets fractional Kelly. There's an uncertainty principle, related to short term fluctuations of good and bad luck. Action points can wipe out that uncertainty.

If the advantage offered by action points can be translated into a mathematical formula, then that formula could be applied to upgrade Kelly from, for instance, 1/3rd to 1/2, or even to full Kelly.

Good point... for the feint of heart it's always better to win less more often than it is to win more less often.

[MrX]

There's something lost in translation between the mathematically perfect Kelly formula and the real life world of sports betting. This is admitted by anyone who bets fractional Kelly.

No, it's not. I use fractional Kelly and the only thing that I'm admitting is that long-term bankroll growth isn't my only preference.

[Dark Horse]

Never mind. This is probably the last place on earth for creative thought.

[Desert Tortoise]

Am I missing something as to why this question is so hard? From my understanding Full Kelly is just advantage divided by variance. You can determine the Kelly bet for any distribution not just binomial.

[HailMary]

what are action points? is the kelly criterion a form of "Nick the Greek" Negative progression? i am interested in formulas that are created in real-time values,any suggestions? any of the 4 major sports will do. hail mary

[Dark Horse]

Just thinking out loud. There would have to be actions points for wins and losses, and these would be entered into a formula separately (probably something with square root and sample size).

Example from last NBA season (results, not desired formula):
Overs 8-0 ; action points for wins 98/8 for +12.25 average
Unders 14-5 ; action points for wins 242/14 for +17.28 average, and action points for losses -64/5 for -12.8 average.
Weakness: small sample size. Weakness largely neutralized by action points.

If I would normally bet 1/3 Kelly (for reasons that aren't relevant for this purpose), my goal here is to come up with a formula that would give me a single number (in the above case, one number for overs, and one for unders) by which I could multiply my fraction of Kelly, in a manner that is statistically 'accurate'.

I'm not interested in the discussion that full Kelly is the way to go, and that anyone who doesn't understand this is bordering on illiterate. This is only for fractional Kelly bettors, who would increase their bet size if action points would afford them an additional edge; and who are interested in precisely defining that edge.

Dogs may have to be weighted differently than favorites.

[TigerPawsSC]

[Dark Horse]

Wrecktangle, I got your pm's, but e-mail address was blocked, and your name was blocked in the second one. I can no longer respond because now your pm is blocked.

If you have any worthwhile thoughts on the matter, feel free to share them here. I may have to ask a math genius friend of mine.