IP Biloxi situation

Ip

Is the total on the left before or after the bet amount?
Larry

 

Ib

Stay a perfect world, the next card is a Ten, and or the dealer has an Ace in the hole

 

Good Bet Too

Since you bet 4th, near the end except for the 5th player to act who is almost out of it; your bet gives you the second high and the second low. Any bet from 2950 to 3075 gives the max probability of 67.88% to advance (using tables for 5 players).

Standing with 15 gives a 37.46% chance to advance

Attempting to hit to a minimum of 17 gives a 43.70% chance to advance.



.................................BlueLight

 

If you stand on 15, your only chance to advance is dealer bust or hitting to 21.
Dealer’s 20 makes BR2 push and BR3 having more money unbet. Any other dealer’s nonbusting hand makes BR2 advancing with one other player but you. Depending what tables for dealer final hand you use, infinite deck or removing the cards used in that hand, or actual count, your chances of advancing are about 32%.

A quick look at my tables shows hitting to 16 better by about 2.5% and starts diminishing (0.5%) by hitting to 17
.
At the table I would make a similar bet, but would consider betting 700-775 (losing locks out BR5 and winning covers BR3 bj), especially that BR3 less than optimal bet lowers chances of his unconventional doubling. Betting 1800 opens chances for BR5 but preserving lows (and push to BR4) but ads beating straight wins by BR2 and BR3 by doubling- if they have good hands. Still, 2975 to 3025 seem a “much” better choice. Betting 2950 doesn’t cover BR2 full double and by the same token splitting 3050 loses more than his full double.

S. Yama

 

I stand corrected

Yes standing with 15 gives a 31.41% chance to advance. Now attempting to hit to 18 gives a 37.76% chance to advance. Couldn't find what I did wrong before, probably inputted a wrong bankroll or bet somewhere.

Still your bet of 3025 gives you an initiall chance of 67.88% to advance.


............................BlueLight

 

Sure enough, I am guilty of pulling a wrong count adjusted table.
Note that BR5 should hit to at least 19 and one point higher than what BR1 ends up with.

With no count known my paper and pencil results are:
stay 15.........hit to 16....... hit to 17....... hit to 18....... hit to 19
31,4% ..........34,4% ......... 35,2% ......... 35,6% ......... 32,2%

Anybody would like to crunch numbers for bet of 1800 and BR1 doubling if need to beat BR2?

S. Yama

 

Ip

Out of curiosity, looking at Wong's table 3 for 4 or 5 players in contention the best play is High if all win. So would a higher bet have been the best thing to do?

Larry

 

Ip

I used Wong's old CTS program and ran 5000 iterations on each of the 3 scenarios. It appears that the 3025 and 1800 bets would be about the same approximately 61% advancement rate. I also used 3200 and it looked like it would be around 54%. I always look at the program as a general indication with multiple player scenarios, especially if anyone is playing behind the simulation point. I just don't know what the program does on some of the situations. It is quite possible that more iterations would have shown more differentiation between the 2 better bets.

Playing 4th you had enough information to make a better bet than just 1st high. Your instinct of 2nd high or 2nd low if all win or lost appears to be a good one. Page 116 (of the version I have) you are BR1 and not betting 1st, appears to fit your situation and keeping the 2nd most chips unbet is the quick answer.

Larry

 

These teasers make great test cases for my software, especially when baseline results are given for comparison. Before attempting the alternatives above, I decided to try to reproduce the results obtained by Ken and Bluelight.

My hand strategy generator correctly determined that hitting to 18 or better is the optimal strategy.
For the initial probability of advancing given the original bets, I got 66.99% (Bluelight reported 67.88%).
For standing on 15 in the original situation, I got 31.45% (Ken and Bluelight both reported 31.41%)
For hitting to 18 or better, I got 37.49% (Ken and Bluelight both reported 37.76%)

All of the results are within 1%, however, since Ken and Bluelight both computed identical results, I feel compelled to find an explanation. Possible contributing factors are:

1) My software uses a simulated shoe and my model used 8 decks and S17 for the dealer, where Ken used infinite deck.
2) My software uses iterative simulation of the situation to determine the probability of success. Perhaps Ken and Bluelight's numbers are the actual computed probability?
3) My model assumes that the player acting after Ken plays basic strategy. I don't know what assumption Ken and Bluelight made here.

Ken and Bluelight, if you could comment on your methods vs mine, either here or via PM, it would be very much appreciated.

With all the being said, my computations for the initial situation are at least in the ballpark, so I ran the additional two suggested scenarios:

1) A bet of 1800 by Ken who would double, if necessary, to beat BR2.

I had my software generate the optimal playing strategy for Ken with this bet and then simulated the situation. The result is that Ken would advance 64.81% of the time. This differs from hopinglarry's reported 61%. My model was, once again, 8 deck simulated shoe with S17 and all players other than Ken playing basic strategy. I ran 360 million iterations vs 5000.

2) A bet of 3200 by Ken

Using the same model as case 1 (including optimal strategy for Ken and basic strategy for the other players) I got a 52.95% of advancing (vs 54% reported by hopinglarry). 235 million iterations were run.

So, it looks like Ken's instincts may have been correct to make the original bet of 3025.

 

Simulator

Thanks for running this stuff. The old CTS program I ran for simulation took a number of hours to run just 5000 iterations. The program does not run fast. It would take me years to run hundreds of million of iterations. I seldom ever piddle with the program, but with enough information it probably at least gives a reasonable estimate of whether to hit or stand. I would not even if have attempted to analize the situation if Ken would have been 1st or 2nd in the playing order. Too many players at the table and I don't know what actions the program takes. To find the best number to stop at (17, 18, ?) would have taken a long time and I would not trust the minor difference between 17-18 as to which would be best.

Larry

 

This was my feeling as well. I did check the intermediate results as the simulation progressed and the numbers did seem to be creeping closer.

My software has this capability as well. I think I'll rerun the situations using that assumption. For the initial bet situations, I assume this means optimal strategy for all?

I hear you. There is so much that I want to add to my software, but I have a family and a full time job!

 

Oh, sorry if I mislead you. My software does not generate optimal bets (yet) either, but given the bets of each player, their goals and which ones play basic strategy, it is able to generate optimal playing strategies for each of the remaining players.

The potential problem for my software here is that, with 5 players, the number of permutations may make it infeasible to compute 5 simultaneous optimal strategies in order to answer questions like "what are Ken's chances of advancing with a bet of X". My previous simulations simplified things by assuming everyone, except you, played basic strategy.