Online live dealer blackjack - New "cash out" feature (poor man's Surrender)

[McSweeney]

The classic Surrender rule allows you to forfeit your hand to get a refund of half your bet. An online casino I play on sometimes just implemented an interesting new feature that is somewhat similar called "Cash Out" but with the twist that you get a dynamic % of your bet back based on what your hand is vs the dealer's up card. The better your hand is vs the dealer's, the more money you get back. Obviously, the casino only gives you pennies on the dollar compared to the EV of the optimal basic strategy decision. I have confirmed that the cash out offer is always the same % of your bet based on your hand vs the dealer's and does not change based on what the remaining shoe composition is.

Some examples:

A terrible 16 vs dealer 10 situation. As you can see, with the Cash Out offer you only get a measly 40% of your bet back; much worse than the classic Surrender rule (50%).
16 v 10.jpg



On the other end of the spectrum, we have a favourable 11 v dealer 6 situation where you can choose to get 165% of your bet back rather than do the optimal basic strategy decision of doubling down.
11 v 6.jpg


So my question is: are these offers so bad that they are NEVER worth taking, or might the count might get so extreme one way or another that it's actually worth doing in rare situations to increase your EV by a small amount? For example, perhaps you've got a 16 vs dealer 10 and the count is so sky high that even taking the 40% is better than doing the optimal card counting decision of standing. Or maybe you've got an 11 vs dealer 6, but you know that not only are a lot of low cards left in the shoe, but also aces! Meaning it's likely you will draw a dud card to put you below 17, and the dealer also has a good chance of drawing low cards/aces to make a hand, so it's better to take the 165% of your bet rather than double down.

[JohnGalt007]

I used my combinatorial analyzer to investigate this question with the two scenarios you described above: 9-7 vs T with a cash-out percent of 40% (Cash Out EV = -0.6), and 9-2 vs 6 with a cash-out percent of 165% (Cash Out EV = 0.65). I assumed that the dealer stands on soft 17. I calculated the EVs of standing, hitting, and doubling down for both of these scenarios while adjusting the running count by changing the composition of the deck before recomputing the EVs. I increased the running count by removing 3s, 4s, and 5s; I decreased the running count by removing 10s. When the EV of the optimal play of 9-7 vs T dipped below -0.6, I noted the current running count as my index at which I should opt for "CashOut". Similarly, when the EV of the optimal play of 9-2 vs 6 dipped below 0.65, I noted the current running count as my index at which I should opt for "CashOut". These were my results:

9-7 vs. T
1 deck index = 9
2 deck index = 16
4 deck index = 30
6 deck index = 44
8 deck index = 58

9-2 vs 6
1 deck index = -1
2 deck index = -2
4 deck index = -2
6 deck index = -3
8 deck index = -4

Although my approach described above is a bit crude, this should at least demonstrate proof of concept that this bet is analyzable and exploitable. In particular, although the running count is unlikely to be quite as high as needed to justify the "CashOut" option in the 9-7 vs. T situation, it's quite possible to get a running count around a true -1 to justify the "CashOut" option with 9-2 vs. 6. Hence, it may be helpful to tabulate all of the indexes and the specific "CashOut" percentages for each situation so you could make a list of only those indexes which are most likely to arise during actual gameplay. By weighting the increase in EV in each of these situations according to their probabilities of occurrence and then summing these products, you could also calculate the overall improvement in your EV for using the "CashOut" options strategically, and see if that amount justifies the extra effort. Since it's an online casino, you could consult these indexes and situations on a piece of paper to make things easier for you. By the same token, though, make sure that the cards are in fact dealt from the shoe without replacement, or else this entire discussion is moot.

[McSweeney]

Thanks! This is most helpful. I should mention that the game is S17, NO double after split, no resplitting ANYTHING, dealer peaks for blackjack if ace upcard but NOT with a 10 upcard (however, you only lose the original bet if you get blackjacked on a double down), 8 decks with an abysmal 4 decks cut out. So yeah the count does not get overly high, although I am using the CAC2 system and you'd be surprised how often the count gets up past true +8 when you get a lucky run of +2 cards! It goes without saying that the game quality is horrible and I need all the help I can get boosting the EV of such a game. But I've actually made a few thousand bucks on it after grinding for many hours just because my bet spread is so ridiculous ($0 - $900).

Some other Cash Out values I've gathered:

· 16 vs dealer 10 – 40%
· 15 vs dealer 10 – 43%
· 16 v dealer A – 45%
· 14 v dealer 10 – 46%
· 15 v dealer 9 – 49%
· 12 (10 2) v dealer 10 - 55%
· 15 v dealer 7 – 60%
· 14 v dealer 7 - 65%
· 15 vs dealer 2 - 68%
· 15 v dealer 4 – 77%
· 17 v dealer 5 - 93%
· 19 v dealer 9 – 126%
· 19 v dealer 5 – 141%
· 10 (5 5) v dealer 4 - 145%
· 11 v 5 – 160%
· 11 v 6 – 165%

[Archvaldor]

By weighting the increase in EV in each of these situations according to their probabilities of occurrence and then summing these products, you could also calculate the overall improvement in your EV for using the "CashOut" options strategically, and see if that amount justifies the extra effort.

Bear in mind also that there is a significant risk reduction when using the cash-out option. This allows you to use more aggressive spreads with the same risk, so you make more money from this reduction.

Technically speaking you might even do this with slightly negative ev because of the significantly reduced risk.

[DSchles]

This is very interesting. I would point out that the Appendixes of BJA3 plus Dave Brolley's charts of the EV of every holding vs. every dealer upcard for every count (or Norm's similar, but not quite as extensive, online charts) furnish all the information you need to compare all of the values.

Archvaldor's observation, above, is important, as well. The two doubles' cash out values are two percent lower than the EV-maximizing ones off the top of the deck, but they surely are worth more in negative counts. In addition, there are risk-averse plays here. You may prefer a certain 165% payout over a risky 167% one (doubling 11 vs. 6 off the top).

In short, yes, this is exploitable. But, overall, for how much is the question.

Don

[JohnGalt007]

Bear in mind also that there is a significant risk reduction when using the cash-out option. This allows you to use more aggressive spreads with the same risk, so you make more money from this reduction.

Technically speaking you might even do this with slightly negative ev because of the significantly reduced risk.

Agreed. The variance is 0 if you decide to CashOut. Thanks for pointing that out!

[JohnGalt007]

In short, yes, this is exploitable. But, overall, for how much is the question.

Don

I suspect that some plays are only marginally profitable due to how large the magnitude of the index has to be to opt to cash out (and therefore how rarely those situations arise), and other plays are more profitable due to how small said magnitude has to be to opt to cash out (and therefore how often those situations arise). The real task then would be to tabulate all these situations and their cashback percentages so we can work out those indexes and then focus on those which are either the most common or yield the greatest increase in EV.

[moo321]

This doesn't look like it's going to be worth much. Especially not with 4 decks cut off of an 8 decker. That game would not be particularly interesting if you could spread 1-100 without heat. It would have to allow pretty high stakes to even start to merit consideration.

[McSweeney]

It is true that the potential playing deviations for Cash Out won't be worth much, especially if the most common ones are biased toward negative counts; the ease of wonging in and out of the game in online blackjack means you simply won't be betting during negative counts. I am actually minimum betting through mediocre counts right now in an attempt to gather the Cash Out value for every possible combination of hands since they don't provide a complete list. I feel like I'm opening packets of Pokemon cards trying to find the rare ones to complete my collection.

The game DOES allow high stakes with max bets ranging from $500 - $5,000 depending on the table. According to a CVData simulation, the game is worth $32.27 an hour with my $0 - $900 bet spread and extreme wonging. I have actually overachieved that greatly in my limited 31 hour sample size. I think it's only a matter of time until I get digitally "backed off" (I can't imagine why I haven't been already since it must be ridiculously easy for them to implement software that automatically detects card counters).

[James989]

Thanks! This is most helpful. I should mention that the game is S17, NO double after split, no resplitting ANYTHING, dealer peaks for blackjack if ace upcard but NOT with a 10 upcard (however, you only lose the original bet if you get blackjacked on a double down), 8 decks with an abysmal 4 decks cut out. So yeah the count does not get overly high, although I am using the CAC2 system and you'd be surprised how often the count gets up past true +8 when you get a lucky run of +2 cards! It goes without saying that the game quality is horrible and I need all the help I can get boosting the EV of such a game. But I've actually made a few thousand bucks on it after grinding for many hours just because my bet spread is so ridiculous ($0 - $900).

Some other Cash Out values I've gathered:

· 16 vs dealer 10 – 40%
· 15 vs dealer 10 – 43%
· 16 v dealer A – 45%
· 14 v dealer 10 – 46%
· 15 v dealer 9 – 49%
· 12 (10 2) v dealer 10 - 55%
· 15 v dealer 7 – 60%
· 14 v dealer 7 - 65%
· 15 vs dealer 2 - 68%
· 15 v dealer 4 – 77%
· 17 v dealer 5 - 93%
· 19 v dealer 9 – 126%
· 19 v dealer 5 – 141%
· 10 (5 5) v dealer 4 - 145%
· 11 v 5 – 160%
· 11 v 6 – 165%

Few months ago, I found that casino keep changing those cash out values for the same hand based on the deck compositions, are you sure those cash out values are pre-determined and fixed through out the shoe ?

[McSweeney]

Quite certain, although I could take some more duplicate screenshots to make absolutely sure. I mainly looked at multiple instances of 15 vs dealer 10 and 16 vs dealer 10 and saw it was the same 43% and 40% respectively. It's possible that every time I took a duplicate screenshot it was coincidentally the same count.

Does the game you played look similar to my screenshots above? Almost every online blackjack game, regardless of which online casino it is, is run by the same software from Evolution Gaming. So what goes in one game should line up with the other.

[bryangarrett]

Your post about the new "cash out" feature in live dealer blackjack. It sounds like a handy tool, especially when you need more time to feel lucky with your hand. It’s like having a backup plan if the game isn’t going your way. Instead of just sitting there and losing everything, you can grab a bit of your bet back. It’s almost like a poor man’s surrender, as you said! It might not be as satisfying as just winning, but it could save you from a more significant loss. It’s worth checking out if you play often. I found more info on it at https://nagad88.lol – seems like this could be a nice addition to the strategy.

[Anton20]

Is that cash out feature similar to Asian surrender?