Streak Sidebet

[CoopOwnz]

A friend has come across the streak sidebet.
It's listed on wizard of odds here: (first paytable)
https://wizardofodds.com/games/blackjack/appendix/8/

I've been running tests on my own software to see if this sidebet becomes better with the count (hi-lo).
I originally assumed it would only get very slightly better because our win/loss percentage doesn't vary much by the count.
However, when running simulations on the 4 streak, it shows that this becomes beatable at +3 TC.

My sims:
Streak of 4 (pays 18:1)
Rules: 6D, S17, DAS, DOA, 75% pen.
5 million shoes ea.

Take Sidebet at	Return
Every Hand	-3.2155%
+1 TC or above	-0.8656%
+2 TC or above	-0.3472%
+3 TC or above	0.2849%
+4 TC or above	0.8749%

I've checked my end over many times but I can't make sense of these results. Has anyone looked into this sidebet?
Can this sidebet be tested in Norms software?

Thanks

[Freightman]

There's is apparently a point whereby you are more than likely the house to win the next the next hand. I believe that point us around true 4 or 5.

Your table states a true count and above. It's the above that appears to be distorting your results. Confine the results to 3.0-3.99, 4.0-4.99 etc. Confining to the true count bucket alone will remove those positive results in the over category, and thus, should reduce the results of where the bet is beatable.

If my thoughts are correct, then the sidebet at true 3 should still show a loss.

[CoopOwnz]

There's is apparently a point whereby you are more than likely the house to win the next the next hand. I believe that point us around true 4 or 5.

Your table states a true count and above. It's the above that appears to be distorting your results. Confine the results to 3.0-3.99, 4.0-4.99 etc. Confining to the true count bucket alone will remove those positive results in the over category, and thus, should reduce the results of where the bet is beatable.

If my thoughts are correct, then the sidebet at true 3 should still show a loss.

That table is showing 5 separate simulations.
In the TC +2 and above row, the player bet on a 4 streak anytime the TC was +2 or above.

[DSchles]

But you want to know how to size your bet. The greater the edge, the larger the bet. You need to do it by TC buckets.

Don

[CoopOwnz]

But you want to know how to size your bet. The greater the edge, the larger the bet. You need to do it by TC buckets.

Don

Yeah, I will dig deeper into this. I first wanted to make sure that it was beatable. It also may be worthwhile to make some basic strategy adjustments for this SB.
I was hoping someone would be able to find some information on this one.

[DSchles]

Go here and enter the proper rules: http://www.blackjacktheforum.com/res...p?do=statspage

For the TC = +3 bin, a hand wins 49.05% of the time. Therefore, .4905^4 = 0.0579, which makes the 18 to 1 payoff very profitable. You win 18 579 times out of 10,000, and lose 1 9,421. Net profit is 1,001/10,000 = 10.01%. The sim appears to be way off.

At TC = +4, a hand wins 49.69% of the time. The math shows an edge of 15.83%.

At TC = +5, a hand wins 50.21% of the time, and the edge grows to 20.76%.

Nice work if you can get it!

Don

[DSchles]

Something must be wrong. There's a 3.67% edge even at +2. Are you sure the payoff is 18 TO 1 and not 18 FOR 1? There's even a 0.28% edge at +1. APers could destroy this game. Too easy. Something is wrong.

Don

[CoopOwnz]

Something must be wrong. There's a 3.67% edge even at +2. Are you sure the payoff is 18 TO 1 and not 18 FOR 1? There's even a 0.28% edge at +1. APers could destroy this game. Too easy. Something is wrong.

Don

Yes it's 18:1 for the 4 streak. I know that my result for the HE off the top matches wizard of odds (3.2%).
Comparing my simulation win/loss/tie per TC was lower then what I got from the form you linked. I will have to go back through everything again to make sure there wasn't some baseline mistake on my end.

My sim was also not using indices but still wouldn't be cause for 0.5% differences in the win/loss percents.

Edit: I will actually have him double check the sidebet at the store.

[DSchles]

You're misunderstanding. Ties don't count! The stats are for RESOLVED hands: you either win or lose. And, when the winning streak pays 18 TO 1, you don't need to win more than 50% of the hands to have an edge. In any event, you DO win more than 50% of RESOLVED hands at +5 and higher -- even lower at single deck.

There are other comments to be made. Even with a very large edge of 15% or higher, there is great variance, because of the 18 to 1 payoff. So, you have to divide edge by 18 to get the correct Kelly wager, which then becomes less than 1% of the bank. Clearly, you need a large bank to capitalize on this, if you're going to take a stab at making significant money.

Finally, although the TC, on average, tends to remain the same, it obviously can fluctuate over the course of four straight hands or more. So, the edges are calculated with the premise that the percentages for winning a hand remain constant over the duration of the wager, when, in fact, that would rarely be the case.

In any event, everyone needs to stop reiterating the myth that you can't have a count where you win more hands than you lose, or that, with ties excluded, you can't win more than 50% of the resolved hands. Both of those statements are patently false.

Don

Don