Blacjack Switch

[Kasi]

The game in Russia is still there BUT they have removed ES so the house edge is now around 0.2% rather than being player positive.

I knew I should have gone to St. Petersburg when I had the chance What was the player BS edge back then anyway? When did they wise up lol?

By no means am I an experienced player but it is one of my favorite games.
And I only have 3 lol.

So you're saying that an AP CC team actually beat this game? It is actually possible? Were they lucky or good? Probably not discussable here anyway.

Wow. Color me "absolutely blown away" whatever color that is.

Meant every word and thanks for the update on the rules variations.

 

[callipygian]

Where I would start, had I the skills, would be to at least duplicate the Wiz EV tables with his rules and infinite deck assumptions.

That's more or less what I did. Back in the era of Microsoft Works, I built an infinite deck simulator to calculate basic strategy. I moved it to Excel when I first started counting cards, and have built various pages around it - for example, calculating double-for-less breakeven points or Illustrious "18" calculations for my bet spread.

The EV tables for BJ Switch can be derived relatively easily - just set the probability of busting on dealer 22 to 0 and the payout for blackjack to be 1. That should replicate the Wiz table.

Since there are only 10,000 combinations of two hands, it's probably easiest just to create a table with each of the possibilities in four columns, use HLOOKUP or VLOOKUP for the EV's of switching vs. not switching, and then a normalization column for the probability of each hand.

Varying the count will then give the most efficient counting system (it's possible that Hi-Lo might not be the best) and the indices for switching switching.

I'll try it out this weekend and see what I get.

And then be kind enough to share it with the world?

Well, I'm guessing that the game isn't going to be any more profitable than regular blackjack. All of the variants - Double Exposure, Spanish 21, etc. - have published counting systems. They're just harder to execute and are less profitable.

 

[callipygian]

Alright, so I did the mechanics of crunching the numbers last night. There are no surprises based on what Wiz already posted. EV ranges from -0.2% to -1.3% depending on whether natural 21's count as blackjack or 21, and whether switched 21's count as blackjack or 21. I got fewer soft doubles than Wiz posted, but the EV's are really close and it might even be rounding error.

Counting is definitely possible: removing 2, 4, 5, 6 increases EV +0.5% each; removing 3, 7 will increase EV +0.3% each; removing A, 9 is worth -0.3% each; removing 10 is worth -0.5% each. Just about anything that works for blackjack should work for BJ Switch.

What's not obvious - at ALL - is basic strategy for switching. I have a hard time believing that any player can actually achieve the listed house edge with rules such as "hard 19 / 10-3: switch against dealer 2-4, 7-A".

I'm trying to organize all the switching strategy into a few clumps; I'm going to recalculate the EV based on the fact that nobody's going to memorize what I have now (which is essentially a block of 100,000 W/nW's).

 

[jack.jackson]

Alright, so I did the mechanics of crunching the numbers last night. There are no surprises based on what Wiz already posted. EV ranges from -0.2% to -1.3% depending on whether natural 21's count as blackjack or 21, and whether switched 21's count as blackjack or 21. I got fewer soft doubles than Wiz posted, but the EV's are really close and it might even be rounding error.

Counting is definitely possible: removing 2, 4, 5, 6 increases EV +0.5% each; removing 3, 7 will increase EV +0.3% each; removing A, 9 is worth -0.3% each; removing 10 is worth -0.5% each. Just about anything that works for blackjack should work for BJ Switch.

What's not obvious - at ALL - is basic strategy for switching. I have a hard time believing that any player can actually achieve the listed house edge with rules such as "hard 19 / 10-3: switch against dealer 2-4, 7-A".

I'm trying to organize all the switching strategy into a few clumps; I'm going to recalculate the EV based on the fact that nobody's going to memorize what I have now (which is essentially a block of 100,000 W/nW's).

Kool! Keep up the good work, cal. Im actually surprised the eor isnt higher for the 2, than the five. But since their about the same, that sounds about right

 

[Geoff Hall]

Alright, so I did the mechanics of crunching the numbers last night. There are no surprises based on what Wiz already posted. EV ranges from -0.2% to -1.3% depending on whether natural 21's count as blackjack or 21, and whether switched 21's count as blackjack or 21. I got fewer soft doubles than Wiz posted, but the EV's are really close and it might even be rounding error.

Counting is definitely possible: removing 2, 4, 5, 6 increases EV +0.5% each; removing 3, 7 will increase EV +0.3% each; removing A, 9 is worth -0.3% each; removing 10 is worth -0.5% each. Just about anything that works for blackjack should work for BJ Switch.

What's not obvious - at ALL - is basic strategy for switching. I have a hard time believing that any player can actually achieve the listed house edge with rules such as "hard 19 / 10-3: switch against dealer 2-4, 7-A".

I'm trying to organize all the switching strategy into a few clumps; I'm going to recalculate the EV based on the fact that nobody's going to memorize what I have now (which is essentially a block of 100,000 W/nW's).

Excellent work Cal.

If you get to the stage where you can assign the count values to each card then I'll input it into the 'Switch' simulator and see how the ev changes per true count.

 

[Kasi]

That's more or less what I did. Back in the era of Microsoft Works, I built an infinite deck simulator to calculate basic strategy. I moved it to Excel when I first started counting cards, and have built various pages around it - for example, calculating double-for-less breakeven points or Illustrious "18" calculations for my bet spread.

The EV tables for BJ Switch can be derived relatively easily - just set the probability of busting on dealer 22 to 0 and the payout for blackjack to be 1. That should replicate the Wiz table.

Since there are only 10,000 combinations of two hands, it's probably easiest just to create a table with each of the possibilities in four columns, use HLOOKUP or VLOOKUP for the EV's of switching vs. not switching, and then a normalization column for the probability of each hand.

Varying the count will then give the most efficient counting system (it's possible that Hi-Lo might not be the best) and the indices for switching switching.

I'll try it out this weekend and see what I get.



Well, I'm guessing that the game isn't going to be any more profitable than regular blackjack. All of the variants - Double Exposure, Spanish 21, etc. - have published counting systems. They're just harder to execute and are less profitable.

That's fabulous C. You make it sound so simple. I never thought to even try to figure out the probability of occurrence of each hand kind of thing but I can visualize it lol. As in, if I had thought of it, I probbaly couldn't do it anyw ay lol.

My mind is kind of reeling at the moment, with visions of showing up with my 100 page BS card or a CC'ing version of same, lol, but great stuff.

I ended up just using a key column and vlooked up from there to combined EV's. Eventually all I really needed was the key column and a Switch/play decision based on one hand, just text, since that was all I cared about.

How the heck would a casino even know if they were hit by a card-counting team like Geoff said anyway? So maybe, it has been done before.

Anyway, a million questions to follow, no doubt

 

[callipygian]

After a few hours of playing with this, I've come to the unfortunate conclusion that the basic strategy for this game is just way too complicated to get even close to the theoretical house edge.

First of all, it's pretty easy to categorize the following hands:
Winners (EV > 0): hard 21, hard 20 (including 10-10), hard 19, hard 11, hard 10 (including 5-5), soft 21, soft 20, soft 19, and A-A.
Neutral (0 > EV > -0.1): hard 18, hard 9, soft 18, soft 14, soft 13, 9-9.
Losers (EV < -0.1): everything else

The simple strategy that you switch to the greatest number of winners, followed by switching to the greatest number of neutrals, will yield a house edge of 1.4% if you don't switch on ties, and 1.8% if you do.

I'm working on trying to break up losers into different groups, hopefully that will help the basic strategy.

To give you a hint of what this "basic" strategy is turning out to look like, here's one of the simpler rules.

If you are not dealt a blackjack but can switch to a blackjack, switch. But if you have a 10-10 and a soft hand, don't switch, except 10-10 / A-2 through 10-10 / A-8 against dealer 10 and 10-10 / A-2 through 10-10 / A-7 against a dealer A. Also,

Hand 1	Hand 2	10	9	8	7	6	5	4	3	2	1
10-9	9-A	W				W	W	W			
10-9	8-A	W									W
10-9	7-A	W	W			W	W	W	W	W	W
10-9	6-A	W	W	W		W	W	W	W	W	W
10-8	9-A	W									W
10-8	8-A	W	W	W	W	W	W	W	W	W	W
10-8	7-A	W	W	W		W	W	W	W	W	W
10-7	9-A	W	W	W		W	W	W	W	W	W

And this is allegedly an "obvious" decision ...

 

[Kasi]

What's not obvious - at ALL - is basic strategy for switching. I have a hard time believing that any player can actually achieve the listed house edge with rules such as "hard 19 / 10-3: switch against dealer 2-4, 7-A".

Am I understanding your example correctly? Like your left hand (top to bottom) is 10,9 and your right hand is 3,10 and you could switch to 10,10 and 9,3 if you wanted?

If so, I'm getting that one should switch to the 10,10 and 9,3 vs all dealer upcards based on the Wiz'z EVs?

In the course of doing that, it seems that the EV's on the Wiz website have been changed from what they were years ago. Can you, Geoff, anyone confirm this?

Regardless, ultimately why even care about learning "rules" for switching decisions? What would be illegal about showing up with a large "basic strategy for switching" printout? I think such a monstrosity could be narrowed down to only 1035 unigue hands per dealer upcard lol. I can't see how any loss in speed of play could occur - I doubt one would need more than 1 sheet per dealer upcard, especially double-sided on legal paper. With small fonts

On what basis could a casino object to this? If they can't, I do see a market for such a product.

Do casinos even currently sell a "Basic Strategy card for just playing decisions" like they do for blackjack games? I can't imagine they'd even actually know what that would be, simple as it is, let alone when it would be proper to switch or not.

 

[callipygian]

Am I understanding your example correctly? Like your left hand (top to bottom) is 10,9 and your right hand is 3,10 and you could switch to 10,10 and 9,3 if you wanted?

The format in the code is (bottom card #1),(top card #1) / (bottom card #2), (top card #2). So yes, you are understanding it correctly.

10,9 / 3,10 can be switched to 10,10 / 3,9.

If so, I'm getting that one should switch to the 10,10 and 9,3 vs all dealer upcards based on the Wiz'z EVs?

That doesn't seem right. You'd be trading a 20 and 12 for a 19 and 13, both of which are inferior. Are you sure you don't have the EV's backwards?

I do see a market for such a product.

I think the chart on wizardofodds.com would be all you need to make all the decisions. If the decision isn't obvious, just add the EV's. The goal was to make something simpler than that.

I'm assuming for basic strategy play, the casino wouldn't care what you had out on the table. However, if you were bet spreading and winning, I'm pretty sure the heat would come down pretty quickly as your devil chart would be quickly blamed for your massive pile of chips.

 

[Kasi]

That doesn't seem right. You'd be trading a 20 and 12 for a 19 and 13, both of which are inferior. Are you sure you don't have the EV's backwards?

Maybe lol. Sure wouldn't be the first time.

More specifically, for 10,9 & 3,10 vs dealer 5, from the Wiz, I get +0.3139+(-0.2551) for a combined EV of +0.08. Versus a 10,10, and 3,9 of +0.5752 +(-.0.2538) for a combined EV of +0.3214.

So one would switch from a 10,9 & 3,10 to a 10,10 & 3,9 vs dealer 5.? I don't find that inconsistent with a rule like "always switch if 1 hand is a higher winning hand" like I think Snyder may suggest as a "switching rule".

Although the EV's my sheet uses differ from those currently posted by the Wiz's website, I still got the same "switch" answer anyway.

Maybe your EV's are based on different rules than the EV's the Wiz's EV's are based on?

I don't even understand the low range of 0.2% HA since, I think as I remember it anyway anyway, the old Playtech game was less than 0.1% whatever those rules were. Have they changed maybe?

It would be pretty funny if every player showed up with a 10 page printout and they kicked out everyone who won some money lol.

 

[Geoff Hall]

Switching Strategy

I developed a 'Switch' table that seemed reasonably effective and was only a table with about 20 numbers in.

I'll see if I can dig it up although I'm not sure if I have it anymore. If so, I'll post it for anyone who wants to see how accurately it compares with the exact decisions.

Heading to Indiana tomorrow ready for the game to go live at Horseshoe Hammond on Saturday. If anyone in that area then drop in and say 'hi' as I'll be there. Also, in Kansas City next week training the dealers ready for the installation at Isle Of Capri there.

 

[WRX]

So you're saying that an AP CC team actually beat this game? It is actually possible? Were they lucky or good? Probably not discussable here anyway.

Some posters here are reinventing the wheel. Just get yourself a copy of Arnold Snyder's Big Book of Blackjack, where he gives correct basic strategy, a simplified switching strategy, and information on how to beat the game with the Red 7 count. Additional information on index plays for the Red 7 count may show up from time to time on his Web site.

Or go to Geoff Hall's own Web site, or to the Wizard of Odds, for basic strategy information.

One point that's unfortunately unclear is how much edge the player gives up by using Snyder's switching strategy, as opposed to theoretically perfect switching. A sim would be nice. Geoff's promised table of simplified switching rules, with edge calculation, will be a big help. (Thanks in advance!) I'm thinking that the sacrifice for less than perfect switching should be small. Most decisions are obvious, and the close ones probably make little difference one way or the other.

Watch out for rules variations, as they can affect the edge significantly. I've even seen the game dealt out of CSMs. The early hit by a counting team that Geoff has described has led to rules being monkeyed with a lot, and they differ between casinos. In many cases, what you end up with is a six deck game with a basic house edge that's a bit higher than in a decent regular six deck blackjack game. The advantage is that you may be able to find very good penetration, and the heat level may be much lower than in a regular blackjack game. So you might find a good opportunity. As always, YMMV.

 

[callipygian]

Some posters here are reinventing the wheel.

I think you're confused about the limitations of what's already posted out there.

Snyder gives a very simple approximation of strategy, but it's not very accurate. Wizardofodds gives a very accurate approximation of strategy, but it's not very simple (you have to memorize ~150 EV's to three decimal places, add two together, add a different two together, and then compare the two sums).

The goal is to find a system which is both simple (maybe 10-15 rules to remember) and accurate (to lower the house edge to at least less than 1%).

Arnold Snyder's Big Book of Blackjack, where he gives correct basic strategy, a simplified switching strategy, and information on how to beat the game with the Red 7 count.

The numbers I posted are essentially Snyder's simplified switching strategy, except that I count 17 as a loser and 18 as a push whereas he counts both as winners.

If I'm right about winners and losers, Red 7 isn't a good strategy to use. If, as I believe, his switching strategy gives a 1.5%+ house edge, the pivot for an unbalanced count should be somewhere between +3 and +5.

Red 7 will only work well for an optimal switching strategy where the house edge is under 0.5% because the pivot is +2.

One point that's unfortunately unclear is how much edge the player gives up by using Snyder's switching strategy, as opposed to theoretically perfect switching.

The numbers I posted suggested this is somewhere in the 1-1.5% range for my simplified strategy and worse for Snyder's. I'll get the exact numbers for Snyder's strategy this weekend.

It's unreasonable to expect players to use theoretically perfect switching. I posted an example of how complex it is in a "simple" case - where player is not dealt a blackjack, but can switch to blackjack.

Trust me, it's much more complex for a "complex" case, where player is switching between two losers.

 

[Geoff Hall]

One point that's unfortunately unclear is how much edge the player gives up by using Snyder's switching strategy, as opposed to theoretically perfect switching. A sim would be nice. Geoff's promised table of simplified switching rules, with edge calculation, will be a big help. (Thanks in advance!) I'm thinking that the sacrifice for less than perfect switching should be small. Most decisions are obvious, and the close ones probably make little difference one way or the other.

I checked the files on my laptop and haven't got my 'simplified switch' table with me. I'm back in the UK on August 13th so will dig it up from my PC when I return (providing I remeber by then ).

I did run some sims, a few years back, which graded the 'switch' decisions according to their difficulty and certain levels of player would get a % of 'switch' decisions correct. For example, an 'expert' player was rated at getting 9,850 'switch' decisions correct and guessing the remaining 500 (of which 50% would be correct). This added close to 0.3% to the house edge - so, for example, playing at Casino Royale, 6 decks, base house edge 0.16%, then a player who gets all but 500 'switch' decisions correct will be playing at approximately 0.46% house edge.

I'm not sure if Snyder (or my) 'switch' table achieves a better 'switch' % as I haven't checked it out. Working on memory, I think that Mike Shackleford's switch table made roughly 30 errors and they were programmed into a sim to add 0.008% to the house edge.

I have an exact 'switch' table in the UK in case anyone would like to compare it to other stategies. It will hav eto wait until after August 13th but if anyone would like a copy then I'll let a few go to those that email asking for one.

 

[callipygian]

For example, an 'expert' player was rated at getting 9,850 'switch' decisions correct and guessing the remaining 500 (of which 50% would be correct).

There should be far fewer decisions than that. ~10,000 is the total number of hands possible, but ~5,000 of those are hand duplicates (e.g. 10-8 / A-4 vs. A-4 / 10-8) and another ~1,000 of the remaining ones are switch-neutral (the top card or the bottom card is the same, thus, switching is useless). Then there's another set of duplicates which are more difficult to identify (e.g. 10-8 / A-4 vs. 4-A / 8-10) which actually are the same two hands with the bottom and top cards reversed. I think there may be close to ~2,000 of these.

Of the remaining ~2,000 switch decisions, at least ~1,000 are obvious (e.g. two losers -> two winners) which actually only leaves about ~1,000 non-obvious switch decisions (which is actually 10,000 entries because each of the decisions goes against a dealer upcard). I've found that these ~1,000 non-obvious decisions can add 1.0% to 1.4% on the house edge.

 

[Automatic Monkey]

There should be far fewer decisions than that. ~10,000 is the total number of hands possible, but ~5,000 of those are hand duplicates (e.g. 10-8 / A-4 vs. A-4 / 10-8) and another ~1,000 of the remaining ones are switch-neutral (the top card or the bottom card is the same, thus, switching is useless). Then there's another set of duplicates which are more difficult to identify (e.g. 10-8 / A-4 vs. 4-A / 8-10) which actually are the same two hands with the bottom and top cards reversed. I think there may be close to ~2,000 of these.

Of the remaining ~2,000 switch decisions, at least ~1,000 are obvious (e.g. two losers -> two winners) which actually only leaves about ~1,000 non-obvious switch decisions (which is actually 10,000 entries because each of the decisions goes against a dealer upcard). I've found that these ~1,000 non-obvious decisions can add 1.0% to 1.4% on the house edge.

One simple rule that will help on a lot of hands is when choosing between two pairs of stiffs, choose the pair with the larger difference between the two, e.g., a 13 and a 15 instead of two 14's.

 

[Kasi]

There should be far fewer decisions than that. ~10,000 is the total number of hands possible, but ~5,000 of those are hand duplicates (e.g. 10-8 / A-4 vs. A-4 / 10-8) and another ~1,000 of the remaining ones are switch-neutral (the top card or the bottom card is the same, thus, switching is useless). Then there's another set of duplicates which are more difficult to identify (e.g. 10-8 / A-4 vs. 4-A / 8-10) which actually are the same two hands with the bottom and top cards reversed. I think there may be close to ~2,000 of these.

Of the remaining ~2,000 switch decisions, at least ~1,000 are obvious (e.g. two losers -> two winners) which actually only leaves about ~1,000 non-obvious switch decisions (which is actually 10,000 entries because each of the decisions goes against a dealer upcard). I've found that these ~1,000 non-obvious decisions can add 1.0% to 1.4% on the house edge.

I got it down to 1035 avg unique decisions per dealer upcard after eliminating the duplicates etc. So, 10350, in total, unique 4-card switch decisions vs all dealer upcards.

Not that I employed it, mind you, as I found it easier to use the brute-force 10,000 decisions per dealer upcard. Easier to use 5 keystrokes for me on the 10,000 hands per dealer upcard (100,000 total) than look up the 10350 hands in some table as I would have to think about the order and if the 4 cards did not turn up then it was a "did not matter anyway" etc. When I was Peter Pan in Never-Never land anyway. aka the Internet lol.

If it's of the slightest help to you, I have the 10350 hand list I could have used.

I'm not sure if you ever replied to that switch decision vs 5 & 6 in some post you earlier made where I got always switch but you got don't switch vs 5 & 6.

I was just using the Wiz EV tables and you may be using your own, or different rules, which could possibly account for the different switch decisions. I think it was like 9,3 & 10,10 or 10,3 and 10,9 vs dealer 5 or 6 or something.

All I remember from the old days is that I think BJS had less than a 0.1% HA at Playtech casinos. I forget the exact rules. I only remember because they gave a 0.1% of total-wagered comp which made it a +EV game for me. It was fun being the casino for a change. Even without a bonus.

:cry: :cry: :cry: :cry: :cry: :cry: :cry:

So I don't fully understand your low-limit of 0.2% but assume it's real-life rule-crap.

 

[Geoff Hall]

I got it down to 1035 avg unique decisions per dealer upcard after eliminating the duplicates etc. So, 10350, in total, unique 4-card switch decisions vs all dealer upcards.

Not that I employed it, mind you, as I found it easier to use the brute-force 10,000 decisions per dealer upcard. Easier to use 5 keystrokes for me on the 10,000 hands per dealer upcard (100,000 total) than look up the 10350 hands in some table as I would have to think about the order and if the 4 cards did not turn up then it was a "did not matter anyway" etc. When I was Peter Pan in Never-Never land anyway. aka the Internet lol.

If it's of the slightest help to you, I have the 10350 hand list I could have used.

I'm not sure if you ever replied to that switch decision vs 5 & 6 in some post you earlier made where I got always switch but you got don't switch vs 5 & 6.

I was just using the Wiz EV tables and you may be using your own, or different rules, which could possibly account for the different switch decisions. I think it was like 9,3 & 10,10 or 10,3 and 10,9 vs dealer 5 or 6 or something.

All I remember from the old days is that I think BJS had less than a 0.1% HA at Playtech casinos. I forget the exact rules. I only remember because they gave a 0.1% of total-wagered comp which made it a +EV game for me. It was fun being the casino for a change. Even without a bonus.

:cry: :cry: :cry: :cry: :cry: :cry: :cry:

So I don't fully understand your low-limit of 0.2% but assume it's real-life rule-crap.

Some 'switch' decisions do change depending on the dealer upcard which is why we looked at all 10,350 decisions in order to eek out the difficult ones.

The difference between Playtech (original) at 0.03% and Casino Royale at 0.16% is that Playtech originally had ENHC. This meant that players could switch to a natural, BEFORE the dealer checked for a Blackjack, and end up with a push. Although, in general, the ENHC rule is worse for players, the 'switch before dealer checks' added 0.13% to the players edge.

Even now, the Playtech version is only at 0.16% edge - same as Casino Royale. In fact, one mathematician calculated the edge using combinatorial analysis and got an answer of 0.13% which I believe may be a more accurate answer.

 

[Kasi]

Some 'switch' decisions do change depending0 on the dealer upcard which is why we looked at all 10,350 decisions in order to eek out the difficult ones.

The difference between Playtech (original) at 0.03% and Casino Royale at 0.16% is that Playtech originally had ENHC. This meant that players could switch to a natural, BEFORE the dealer checked for a Blackjack, and end up with a push. Although, in general, the ENHC rule is worse for players, the 'switch before dealer checks' added 0.13% to the players edge.

Even now, the Playtech version is only at 0.16% edge - same as Casino Royale. In fact, one mathematician calculated the edge using combinatorial analysis and got an answer of 0.13% which I believe may be a more accurate answer.

I was just trying to perhaps narrow the gap for Callipygian but on re-reading I see he has it down to less than that anyway from an "obvious" switch point of view. Anyway, I'm glad we agree on the total 10350 cases that need to be analyzed. Also, like he said, I do remember a bunch of hands where switching 1 hand to a BJ would not always be correct even though that might be the "obvious" thing to do. Maybe something like 10,9 and 7,A where you could switch to BJ and 7,9. Like you'd play the first hand vs dealer 8 or 7 maybe but not all other dealer upcards. And maybe give up a substantial amount of EV on that hand, especially vs the 7. Not that that would even be illogical if you thought about it long enough but it would be easy to just quickly go with the simple "I can switch to one BJ, how bad can that be" strategy.

Anyway, and this will be about as vague a question as you can get lol, take a knowledgeable BJ player and what would contribute more to any increased HA, switching decisions or BS plays after the switch decision?

Maybe I mean, say take for example Casino Royale with a perfect switching player who played "normal" multi-deck BJ BS. What would he be giving up? Or vice-versa under any assumptions you want to make about the skill of the switch player lol. Which is relatively more important in general would you say?

Would you say casinos use the "hold %" to comparatively evaluate the game to others and, if so or even not, how do you suppose they go about figuring out comps for an avg player for which I assume they use some avg HA they think they play at?

If any of the above is not fit for the public, that's cool.

Thanks for the update on the old Playtech rules. Wow, 0.03%, lower than I think I remember. Who couldn't love that with a 0.1% comp bonus?!

Oh no. I have to :cry: me a river again just thinking about it.

Does the Super Match Bet, or whatever it is, stay the same at all casinos? Or even still exist? Didn't seem like the worst side-bet to me.

And good luck to you on your trip in the States here. It just can't be easy competing for shelf space at the supermarket, let alone keeping it there, yet, somehow, you've done it.

 

[WRX]

Quote:

Originally Posted by Geoff Hall:

"For example, an 'expert' player was rated at getting 9,850 'switch' decisions correct and guessing the remaining 500 (of which 50% would be correct)."

There should be far fewer decisions than that. ~10,000 is the total number of hands possible, but ~5,000 of those are hand duplicates (e.g. 10-8 / A-4 vs. A-4 / 10-8) and another ~1,000 of the remaining ones are switch-neutral (the top card or the bottom card is the same, thus, switching is useless). Then there's another set of duplicates which are more difficult to identify (e.g. 10-8 / A-4 vs. 4-A / 8-10) which actually are the same two hands with the bottom and top cards reversed. I think there may be close to ~2,000 of these.

Of the remaining ~2,000 switch decisions, at least ~1,000 are obvious (e.g. two losers -> two winners) which actually only leaves about ~1,000 non-obvious switch decisions (which is actually 10,000 entries because each of the decisions goes against a dealer upcard). I've found that these ~1,000 non-obvious decisions can add 1.0% to 1.4% on the house edge.

Well, I'm not 100% clear on what you mean by this, but if you are saying that getting these 1,000 decisions wrong adds 1.0% to 1.4% to the house edge, that seems consistent with Geoff's old sim finding that an "expert" getting 250 decisions wrong gives up about .3% added house edge.

If it turns out that Snyder's simplified switching strategy, or Geoff's own, gets the player to the "expert" level, giving up that added .3% house edge, the Casino Royale game will be very playable. The game in some other casinos may be not so great. I hope that Geoff will be able to follow up on this topic.

Snyder never counts 17 or A6 as a "winner." The best it can be is a "chance" hand. See The Big Book of Blackjack pp. 232-33.