vpFREE2 Forums

S.C.O.R.E for video poker

#1

In the last Gambling with an Edge radio show, the S.C.O.R.E. value was mentioned for blackjack. This is supposed to be the hourly win rate assuming a Kelly betsize for a $10,000 bankroll. For video poker this would be:

$10,000 x (Kelly ratio, approx. ev/variance) x hands/hour x ev

this comes out to:

$10,000 x ev^2/variance x hands/hour

For FPDW at 1000 hands/hour:

SCORE=$10,000 x .0076 x .0076 / 25.84 x 1000 = $22.35 per hour

If I use the exact Kelly ratio:

SCORE=$10,000 x (1/2925) x 1000 x .0076 = $25.98 per hour

This could be a way to compare table games to video poker.

It's clearly related to the Sharpe Ratio (ev/sd) which is a standard way to rate investments.

[Non-text portions of this message have been removed]

#2

To earn the $22.35/hr you need to bet $2941/hr (22.35/.0076) or $2.94/hand at 1K hands/hr.

G'luck all,
Gamb00ler

nightoftheiguana2000 wrote :

In the last Gambling with an Edge radio show, the S.C.O.R.E. value was mentioned for blackjack. This is supposed to be the hourly win rate assuming a Kelly betsize for a $10,000 bankroll. For video poker this would be:

$10,000 x (Kelly ratio, approx. ev/variance) x hands/hour x ev
this comes out to:

$10,000 x ev^2/variance x hands/hour

For FPDW at 1000 hands/hour:
SCORE=$10,000 x .0076 x .0076 / 25.84 x 1000 = $22.35 per hour
If I use the exact Kelly ratio:
SCORE=$10,000 x (1/2925) x 1000 x .0076 = $25.98 per hour
This could be a way to compare table games to video poker.
It's clearly related to the Sharpe Ratio (ev/sd) which is a standard way to rate investments.

[Non-text portions of this message have been removed]

#3

Gamb00ler wrote: "To earn the $22.35/hr you need to bet $2941/hr (22.35/.0076) or $2.94/hand at 1K hands/hr."

Right. This is a rating system. The famous Don Schlesinger (has a wiki page) invented it I believe so if you don't like it take it up with him. In the real world of course you can't bet $2.9412... per hand, whether at table games or a machine. However, under the Kelly system, you can always bet less, in fact that's a real good idea, to cover the unknown unknowns. It's a fallacy that you have to always bet the exact optimal Kelly bet, which is a good thing because almost always that is not an allowed bet size. If you'd like you can adjust the S.C.O.R.E. rating to cover allowed bet sizes less than or equal to the optimal Kelly bet, the allowed bet sizes on machines being different than the allowed bet sizes at table games. And your bankroll may actually be several multiples of $10,000. In my mind the way to think of this rating system is that it represents the hourly return assuming what is probably a hypothetical optimum Kelly bet and per $10,000 of bankroll. If your bankroll was $20,000, you'd double it, and so on. If you're not playing 1000 hands per hour on machines or 100 hands per hour on tables or IGT headsup poker, you can make that adjustment as well.

[Non-text portions of this message have been removed]

#4

Personally, you can get more or less the same benefits as S.C.O.R.E. with the following system for rating any gamble/investment:

1. Do you think you have an edge? You need an edge if you expect to win in the long term.

2. Do you think you have the bankroll? I'd recommend a bankroll always greater than the betsize times the variance divided by the edge, which is the approximate Kelly optimal. If you want you can solve for the exact Kelly optimal using a spreadsheet, for FPDW it is 2925 times the betsize. The advantage of using the estimate (variance/edge) is that it can be calculated on the fly in your head as long as you think you know the variance and edge.

3. Do you think you have the time? You need to be able to play over Nzero (variance/edge/edge) hands in the current tax year. Nzero is the point where the edge equals one standard deviation. Less than Nzero is clearly still short term.

If you can answer a firm yes to all of these questions, you have a play. If you have a choice of several plays, you need to rank them somehow, the typical way to rank plays would be by dollars/hour of ev. Another way to rank would be by the Sharpe Ratio which is ev/sd or ev/sqrt(variance).

[Non-text portions of this message have been removed]