Higher c-socre in cvcx with more players?

[Charlie Mosby]

Hi, all, this is another cvcx problem which I hope is not so foolish as before, thanks!

I just noticed a series of cvcx sims with the identical setup (with only the number of players changed) show the following:

# of total players: __2,____4,____6
highest c-score: ~251, ~259, ~268
(all at the fixed RoR = 0.5% and varying the spread size, with the '~' indicating that c-score is roughly at that level)


The exact values don't matter, the point here is clearly 'with more players the c-score is higher'.
As for the details, basically when there are more players both the win rate and standard deviation per hour (not per round) are slightly lower and the c-score came out higher......and N0 is lower.


These are completely opposite to what I so far believe is true.


The above numbers are for playing a single spot. This increasing trend will be more obvious when playing two hands (at TC >= 1 to TC >= 3), something like c-score = ~300 for 2 players and ~350 for 6 players. (I know the results for playing two hands is just an estimate, but still....)

I know that it takes two points to make a straight line and 3 points can hardly be considered showing any trend....but I don't think this is just some artifacts :P


I'm wondering if I made some mistakes in the setup that I'm not aware of.....though not very likely as I double checked many times.


All are 2*10^10 rounds (I think this 20000 million is pretty enough) with the following rules and setup:

6D, H17, DAS, RSA, LS, pen = 80%

Wong in at TC = 1, wong out at TC = 0, stop observing at TC = -1, no time lag between shoes
(note that is it NOT checking the boxes 'back counting' and 'departure adj' for a play all sim....these are in the setup)

allow betting at half true count checked

same set of r-a indices generated previously with CVIndex, same bankroll, same number of hands per hour, etc.

(For only 1 player the c-score is dramatically lower than 2 players, like ~120....but I don't think checking the box for back counting is very realistic for a total number of players = 1 simulated for play all)





oh btw here's a little separate question for checking the 'play two hands' box:

as the cvcx sim is itself setup with wonging in at TC = 1, it makes no sense to 'start playing two hands at TC = 0'.
However, the results (the betting schedule, the c-score and all those numbers below) are different for playing two hands starting at TC = 0 vs TC = 1.
I know that this might be an artifact due to playing two hands being an estimate.....but then what about starting at TC = 1 vs TC = 2?
I guess I just need to know that the results for starting at TC = 1 (the 'critical point' of the setup) is reliable.

[Norm]

Makes sense. CVCX assumes that you sit at third base. The more players, the more accurate the play at third base, and the deeper the actual penetration.

I don't understand the second question.

[Charlie Mosby]

Makes sense. CVCX assumes that you sit at third base. The more players, the more accurate the play at third base, and the deeper the actual penetration.

I thought this 3rd base advantage is nonexistent in face up shoe game; you still see all the cards there is to see even if you're on 1st base......I don't understand how the play at 3rd base can be more accurate........oh is it actually not about the accuracy of the play, but that the 3rd base player is always at a 'deeper' place compared to the 1st base (when comparing round by round)? At a 'deeper place' being better is essentially what the floating advantage is about (slightly higher advantage even at the same TC).

Put differently, if one matches up the results like this: 1st base at (n+1)th round vs. 3rd base nth round.
then the 1st base is always 'deeper' in every match-up, however, the combined result of these match-ups will be over come by what's left off: the very first round for 1st base vs. the very last round for 3rd base, where the former basically is at TC = 1 and the latter can be at higher TC (or as low as TC = 0 right before Wonging out), plus there's floating advantage effect.



Besides that, the implication seems to be this:
even though there are fewer rounds played when there are more players present, the HOURLY result will be better if you play at an almost full table.
sure, c-score is unit-less .....but with win rate and standard deviation can be both per round or per hour, we might as well view the c-score being an hourly result.

Is this interpretation correct? I shall from now on seek tables with more players than just 1 or 2 (as few as possible) existing players?

The commen knowledge is that when the count is high, we don't want other players to compete for the good cards (thus the advantage)....and this is also part of the 'play 2 hand or not when heads up' issue. I previously thought other players will literally 'dilute' the advantage.
(I know that the changes in wagering TC vs. playing TC will average out in the long run so this is not a problem)

I don't understand the second question.

yeah I just realized that question was stupid......one can indeed play two hands at TC = 0 since the count can drop after you wong in at TC > 1.
My bad :P

[Norm]

I thought this 3rd base advantage is nonexistent in face up shoe game.....I mean, you still see all the cards there is to see even if you're on 1st base.

Not quite all of the cards. The difference is tiny, as your can see from the results.

If you run a sim with the Wong in at TC > 1 setup, and then when viewing the results you check 'play two hands' and scroll the spinner to be @ TC = 0...you'll get a result different from having the 'play two hands' spinner to be @ TC = 1.

Hmm, it's been a long time since I wrote that support. Probably shouldn't allow Wong in the setup and two-hand estimation at the same time. You need CVData for that kind of detail.

[Charlie Mosby]

Not quite all of the cards. The difference is tiny, as your can see from the results.



Hmm, it's been a long time since I wrote that support. Probably shouldn't allow Wong in the setup and two-hand estimation at the same time. You need CVData for that kind of detail.

Thanks for the quick response......but I edited and added many stuff. Hope you can read it.

I guess I need to re-think what I know about the game...... I don't understand what more info can be gained on the 3rd base. Yeah sure you do see a few more cards than the 1st base, but I don't see how that is more info to make the play 'more accurate' (or info relevant to 1st base's play, which proceeds those cards; these few more cards will be relevant to the 1st base for the next round, but 1st base of course will have already counted them by then); the 1st base is also obtaining the accuracy his/her count system allows.

[Norm]

First base does not see the other players' hit and split cards before making decisions. The more you see, the better.

[Charlie Mosby]

The difference is tiny, as your can see from the results.

First base does not see the other players' hit and split cards before making decisions. The more you see, the better.

Actually the difference is not that tiny.

For the sims described above, if 'play two hands' is checked and spinner set at TC >= 3, then

# of total players: __2,____4,____6
highest c-score: ~299, ~327, ~350

I don't think this can be explained by the 3rd base advantage as mentioned.

so I'm wondering if there's something else missing in my sims to make the comparison valid.


In addition, haha sorry I still don't quite understand why seeing more is better. Are you saying that those hit and split cards (of other players) that will come AFTER 1st base's play in fact help make his decision more accurate?

If that's the idea, it's even more bizarre to me. Consider this extreme example: suppose I know (see/count) every card that will come after my play, down to the very last card (or to the cut card) of the shoe, will that make my play more accurate? I hardly think so.
(This extreme scenario is equivalent to having like 100 players at the same table. Within this one and only round, against the same dealer up card, the 1st base will play at the beginning of the shoe while the 3rd base will literally play at the end of the shoe)

Counting is to use the cards dealt to estimate the composition of the remaining (undealt) cards. Within the same round, the hit and split cards (of other players) is NOT part of the dealt cards to the 1st base; they are part of the remaining cards and 1st base actually do NOT want to count them for his/her play, which proceeds them.

Within the same round, other players' hit and split cards IS part of the dealt card (that needs to be counted) to the 3rd base, but not to the 1st base.

[Charlie Mosby]

Come to think of it, we also treat the dealer's hole card and those unseen in the discard tray as part of the remaining deck, even though physically they come out before our play.

hmmmm......I think my argument against the 3rd base advantage still stands, but I need to be more careful with the language.

I'm not saying 3rd base absolutely has no advantage....maybe there is, I'm not sure. I tried to argue in the previous messages that the explanation for the tiny advantage (if there's any) cannot be that 'the play is more accurate from seeing more cards'.

message #1:
All sims are 2*10^10 rounds (I think this 20000 million is pretty enough) with the following rules and setup:
... same number of hands per hour, etc.


message #7
...
...
so I'm wondering if there's something else missing in my sims to make the comparison valid.

At one point I was thinking maybe this is again the "number of rounds per hour" issue, like that for 'play 1 spot vs. play 2 spot when heads up'.

Reasonably, with more players the game is slower. However, since c-score is unit-less this should not matter.

Come to think of it, why did it matter that play 2 spots is slower (for the 'heads up play 1 spot vs. play 2 spot' issue)?
I know that the SCORE is defined with 100 round/hr so with different speed the comparison is unfair, but really the c-socre doesn't change when you tune the number of hands per hour......

Is it that tuning rounds/hour in viewing the statistics doesn't really give us what we want (except for estimating goals probability and hours needed etc)? We need to use CVdata to run entirely separate sims?

[Norm]

There are two separate effects here. In your first post, the difference was tiny. This is the difference in advantage for the third base player because he gets to see some more cards befor playing and plays at a slightly higher effective penetration as far as playing efficiency. This difference is well-known. Later, you posted sims where you swich from one to two hands at TC>=3 and showed a much larger difference. This is quite different. When you switch from one to two hands at a high count and are the only player, there is a gain. When there are other players, the gain is much higher because you are giving yourself a higher percentage of the +3 hands. If there are four players, you normally get 1/4th (25%) of the hands. But, you are giving yourself more like 2/5ths (40%) of the good hands. Not exactly, but enough of a difference to give yourself a substantial boost.

NOTE: This explains the differences in the sims you are running. But, it does NOT mean that switching to two hands at +3 makes sense. Thats' an entirely different question.

[Charlie Mosby]

NOTE: This explains the differences in the sims you are running. But, it does NOT mean that switching to two hands at +3 makes sense. Thats' an entirely different question.

yeah I talked about switching to two hands at +3 just because the difference is bigger.

Curiously if spreading to 2 hands at +1, the difference in c-score is much smaller for 6 vs. 2 players: ~377 vs. ~364

There are two separate effects here. In your first post, the difference was tiny. This is the difference in advantage for the third base player because he gets to see some more cards befor playing and plays at a slightly higher effective penetration as far as playing efficiency. This difference is well-known.

After some more thoughts about this 3rd base advantage, I now know what I was missing. I was wrong saying that 'play being more accurate cannot be the explanation'. Actually my hypothetical scenario with 100 players at the same table exactly illustrates (magnifies) this effect.

Later, you posted sims where you swich from one to two hands at TC>=3 and showed a much larger difference. This is quite different. When you switch from one to two hands at a high count and are the only player, there is a gain. When there are other players, the gain is much higher because you are giving yourself a higher percentage of the +3 hands. If there are four players, you normally get 1/4th (25%) of the hands. But, you are giving yourself more like 2/5ths (40%) of the good hands. Not exactly, but enough of a difference to give yourself a substantial boost.

Thanks.

I understand that with more other players spreading to two hands gives myself a higher percentage increase. However, in the end the question is that, as a result, is playing at a table of 6 (with 5 others) really having a c-score that much higher than at a table of 2 (with another guy)?
(given that you spread to two spots at TC = 3)

I used a bankroll of only 7.5K, and the c-score of ~350 for 6 players vs. ~300 for 2 players just seems unreasonably high a difference.

I don't know....it's just hard to swallow that for this specific style of play that playing at a full table is much better than a total of 2 players.

I'm more inclined to think that it's unfair to directly compare cvcx sims of different players (all other settings being identical), and if one wants to compare the effect of total players, one needs to run CVData sims while specifically taking into account the speed issue.

[Norm]

I am not sure how you would simulate what happens backcounting at a casino in terms of game speed. I know assumptions are made by the computer.

Nope. I agree with you that the computer does not have enough info to make this assumption. Therefore, it has always been my contention that the computer should not make any such assumption. That is why the users must specify speed based on their experiences in particular situations.

[Charlie Mosby]

...
...
My point is that game speed not the number of players is what gets you more hands played
...
...
If you were just worried you had made a mistake Norm covered it well.
If you were thinking toward casino applications it is a moot point.

Yeah, I think I over-reacted to seeing that somewhat counter-intuitive (to me) sim result.
Thanks Tthree.

And thanks again, Norm.