test 15vs9

[Kasi]

Published EV tables are available thru some online sites, but a very handy hard copy version is in the back of Professional Blackjack by Stanford Wong. This is strictly an "EV" chart, so it doesn't show the frequency of pushes -- but "push" frequencies are not needed when you have the net EV. The net EV is basically a product of taking half the pushes and assigning them to the win column, with the other half going into the loss column. That keeps the sample size the same, while revealing the net gain/loss percentage. For example, if a hand goes W=.50, L=.30 and P=.20, you gain 20 units on 100 hands. If you count each push as a half win and a half loss, you've got W=.60 and L=.40 for the same +.20 EV.

If you really want the "push" frequencies, there is a whole book of W-L-P sim results in 190,000,000 hands of Blackjack by Bill Brown.

Thank you for replying.

I do understand what you say.

I guess to re-phrase, how do you know, in your 15 vs 9 example, that the W%age isn't .19, the lose %age isn't 0.662 and the push %age isn't 0.148 since, I think they will yield the same EV?

No big deal, I think you mean if I really only care about W/L/T %'s I need a program to exactly figure them out - I can't figure them out from only knowing EV.

Which is what I think I thought lol.

Still curious how you arrived at your 0.232, 0.704, 0.0604 %'s in the first place? From a sim maybe? From a program? If you did it with pencil and paper, that's all I got pretty much, I'd like to know more about the process.

 

[k_c]

Contrary to the perhaps generally accepted opinion, and one that does seem to have an appealing logic, one actually doesn't only double a hand unless one has over a 50% chance of winning that hand.

If you count pushes as 50% win and 50% loss then yes it takes a winning percentage of over 50% to make a double profitable.

To k_c, does your stuff figure out W/L/T %'s?

My earlier programs computed the winning percent of a player hand counting a push as half a win and half a loss, but that was before I began to consider doubles, splits, surrenders, das, dan, etc. EV can be calculated EV=(W%-L%). But what should be done for a double? EV=2*(W%-L%)=2*W%-2*L%? Wouldn't that count the doubled hand as 2 hands? If not, why does W%+L%=200% for this hand? I concluded that this was not really 2 hands at all, but a weighting of the initial bet by a factor of 2. Similarly splits are only 1 hand as well. When strategy calls for splitting, the number of expected hands resulting from the split can be computed. If no resplitting is allowed the expected number of hands is always 2. If resplitting is allowed it will be more. The number of expected hands for 8-8 vs 10 for a full single deck when splitting to 4 hands is allowed is ~2.085. The extra bets placed on the table as a result of splitting and doubling after splitting is also a weighting of the initial bet.

Win-Loss-Push data is fine for analyzing simple unsplit hands but for more complex hands I think figuring EV based upon initial bet is the best way to go. If you try to use win-loss-push for complex hands you'll probably wind up in a comparing apples to oranges type of mode.

 

[Renzey]

How do you know, in your 15 vs 9 example, that the W%age isn't .19, the lose %age isn't 0.662 and the push %age isn't 0.148 since, I think they will yield the same EV?

That's right. If you're just looking at an EV table, you won't know the frequency of pushes. The W/L/P could've been .000/.472/.528, and you'd have the same EV.

Still curious how you arrived at your 0.232, 0.704, 0.064 %'s in the first place? From a sim maybe? From a program? If you did it with pencil and paper, that's all I got pretty much, I'd like to know more about the process.

It came from the sims in "190,000,000 Million Hands of Blackjack".

 

[Kasi]

If you count pushes as 50% win and 50% loss then yes it takes a winning percentage of over 50% to make a double profitable.

Hey I'm so glad you are still at least reading stuff here

I think we are saying the same thing here - one could win fewer than 50% of every 100 rounds dealt but a double could still be profitable.

My earlier programs computed the winning percent of a player hand counting a push as half a win and half a loss, but that was before I began to consider doubles, splits, surrenders, das, dan, etc. EV can be calculated EV=(W%-L%). But what should be done for a double? EV=2*(W%-L%)=2*W%-2*L%? Wouldn't that count the doubled hand as 2 hands? If not, why does W%+L%=200% for this hand? I concluded that this was not really 2 hands at all, but a weighting of the initial bet by a factor of 2. Similarly splits are only 1 hand as well. When strategy calls for splitting, the number of expected hands resulting from the split can be computed. If no resplitting is allowed the expected number of hands is always 2. If resplitting is allowed it will be more. The number of expected hands for 8-8 vs 10 for a full single deck when splitting to 4 hands is allowed is ~2.085. The extra bets placed on the table as a result of splitting and doubling after splitting is also a weighting of the initial bet.

Win-Loss-Push data is fine for analyzing simple unsplit hands but for more complex hands I think figuring EV based upon initial bet is the best way to go. If you try to use win-loss-push for complex hands you'll probably wind up in a comparing apples to oranges type of mode.

Well, as usual, you are light years beyond me. I've never even thought about considering EV as anything but a % of initial bet.

Truth be told I don't even know why I cared that much about W/L/T %'s anyway lol. People sometimes ask and I have no idea lol. Like you say, EV seems to matter alot more since that's the bottom line no matter what one's W/L/T/ %'s are.

But, if I did care, I'd express W/L/T %'s per "round" not per "hand" like I think you are saying.

I either end up with a net loss, a net win, or a net push per dealer upcard.

So, if I ever could figure this out, that's the way I'd do it lol.

 

[Kasi]

That's right. If you're just looking at an EV table, you won't know the frequency of pushes. The W/L/P could've been .000/.472/.528, and you'd have the same EV.

It came from the sims in "190,000,000 Million Hands of Blackjack".

Thank you - it all makes sense to me now lol.

 

[KOLAN]

bet anyway
for 180 hands played
win stand
hit -48
stand -34

 

[stophon]

so for 180
hit is -26.7%
stand is -18.9%

We're slowly closing in on the numbers we should be getting...

 

[Renzey]

bet anyway. For 180 hands played
hit -48
stand -34

It's not really a sample size of 180. It's just 90 hands played each way. You need 300 each way to be 90% confident that you've found the better play.

 

[KOLAN]

It's not really a sample size of 180. It's just 90 hands played each way. You need 300 each way to be 90% confident that you've found the better play.

i dont think so posh gonna help hit to rich stand

 

[shadroch]

I have no idea what you just said, but rock on!!!!:cool2:

 

[bj bob]

I have no idea what you just said, but rock on!!!!:cool2:

With ya there, ShadMan. It takes me as much time to translate KOLAN'S 10 word sentence as it does to read a chapter of the Aeneid in the original. As a linguist I find both equally entertaining.

 

[Xur]

WTB Kolan - > English Translator

 

[Kasi]

With ya there, ShadMan. It takes me as much time to translate KOLAN'S 10 word sentence as it does to read a chapter of the Aeneid in the original.

You should try Homer's Odyssey in Greek - brought back memories with the Aeneid stuff lmao.

Either way, whether one's heroes are Greek or Roman, KOLAN is a linguistic challenge lmao.

 

[Renzey]

It takes me as much time to translate KOLAN'S 10 word sentence as it does to read a chapter of the Aeneid in the original. As a linguist I find both equally entertaining.

Kolan, don't let'em dissuade you from completing your learning process. I'm anxious to see the results of those other 210 hands played each way with 15 vs. 9.

If you don't get them here, you owe it to yourself to complete them on your own.

 

[shadroch]

210 hands? I thought he wanted 10,000 players to each play twenty hands. Anything less and I'll feel used.

 

[KOLAN]

Kolan, don't let'em dissuade you from completing your learning process. I'm anxious to see the results of those other 210 hands played each way with 15 vs. 9.

If you don't get them here, you owe it to yourself to complete them on your own.

ok i just play 100 hit
results
lost 63
win 30
push 7

 

[Renzey]

ok i just play 100 hit
results
lost 63
win 30
push 7

Good. Now play 100 stand. Then add both of those results to the 90 + 90 that we already have.

 

[KOLAN]

100 stand
lost 72
win 28
push 0

 

[KOLAN]

Good. Now play 100 stand. Then add both of those results to the 90 + 90 that we already have.

bet 30 times win from 100 it becose i lucky

 

[QFIT]

Tell me when you hit 1,000,000,000 hands.