Where should you sit at a table?

Yeah. I know. "On my rear end." But seriously, folks...

Has anyone done any research on the best position at a crowded table? I agree that one shouldn't choose any seat at a crowded table if an empty table is available, but if there are no empties and one is determined to play, which seat has the best return using a good counting system?

This is not idle curiosity. I recently wrote a little computer program to emulate and track several players in a 2-deck game. The cards are dealt face-down, each player's running count is updated according to his position (he can see busted hands, snappers, double-downs and can peek at his neighbor's hole cards), and insurance bets are included. The results were pretty interesting, and to my chagrin, I found that my usual favorite position is if fact the worst position.

I won't divulge the results at this time, for I think it would spawn a bunch of subjective posts arguing about personal favorites and other opinions. I am wondering whether someone else has done a similar study, or has published anything on the subject.

So far, I have only run the 2-deck version because I hate those damn shoes, and because several of the casinos in my area offer a pretty good 2-deck game.

I thank you for your time.

If you have a sim showing a difference in results due to seating position that is more than .01% over several million hands, then something is wrong with your sim.

you didn't answer my question: Have you done a similar study, or do you know whether anyone has published any information on this subject?

I've been programming long enough to know that virtually no computer code can be called perfect or bullet-proof, but I will say in defense of this program that each of the emulated players are using the same PLAYER class, strategy matrix, card-count object and so on. The numbers LOOK good, but I don't offer that as proof.

Assuming that the code is correct, I can only offer conjecture as to why there is such a divergence from theoretical expectation. That's the main reason that I asked for some practical input.

Thanks for your time.

Try having all the players play basic strategy. If they do not show equal results based on seat position, something is wrong.

For players who alter their playing strategy based on the true count, third base will yield a small advantage, because they see more cards dealt before they make their play. (they are in effect getting consistently deeper pen).

>>Try having all the players play basic strategy. If they do
>>not show equal results based on seat position, something is wrong.
I can try it, but it will take a little time. However, by disabling the card count system, I think every player will show the expected return of about -0.5 percent or so.

>>third base will yield a small advantage, because they
>>see more cards dealt before they make their play.
Okay, I'll tell you the results. First base is best (+1.5%). Worst is third base or next to third (+0.6% or so). All players win, but some win more than others. It's surprising because it's counter-intuitive. Or maybe not.

All the players' bets are determined by the ace-adjusted true count before the deal. All the players make the same bet. During the course of play of each round beginning with a fairly high count, the average tendency is for the the running count to rebound downward toward the norm. By the time the last player gets to act, the temporized true count will likely be lower than his large bet justifies. Compare this to when the first player acts: Most of the cards dealt are unseen, except for his own, his neighbor's, the dealer's upcard and any snappers that have been turned over. The unseen cards can be effectively ignored, the players behind him have not depleted the deck with their draws, and therefore his temporized count is usually closer to the original betting count.

Compare this to the cases with a really negative betting count. While the running count will usually rebound upward during the course of the round, the net effect is minimized because everyone has the minimum bet out.

I am suggesting that these results depend on the cards being dealt face-down in a single or double-deck game, and would probably be very different in a multi-deck game dealt face-up.

But this is merely guess and conjecture on my part. I will add a screen dump module to the program and will check the program's execution. Again, this might take a little time.

Thanks for the response. I'll try to tell you more when I know more.

"All the players' bets are determined by the ace-adjusted true count before the deal. All the players make the same bet." -- Yes, they all have the same information available to them at that point, they should be making the same bet.

"During the course of play of each round beginning with a fairly high count, the average tendency is for the the running count to rebound downward toward the norm." -- Yes and no, the running count does have a tendency to go down, but it will be at a rate equal to the rate of cards depleted -- the true count tendency is to remain the same. If the true count has changed, then 3rd base has the advantage, he can make play variations with the most information.

"By the time the last player gets to act, the temporized true count will likely be lower than his large bet justifies." -- Since first base has bet an equal amount, he has just as large unjustified bet out, when it comes time for the dealer to draw.

Compare this to when the first player acts: Most of the cards dealt are unseen, except for his own, his neighbor's, the dealer's upcard and any snappers that have been turned over. The unseen cards can be effectively ignored, the players behind him have not depleted the deck with their draws, and therefore his temporized count is usually closer to the original betting count. -- Exactly. 1st and 3rd base place their bets with the same amount of information, they are the correct amount to make at that time. They each have an equal chance to get the same type of cards indicated for that count. (With three players -- 1st base's hand consists of the 1st + 5th card, 3rd base's hand will be 3rd + 7th -- no reason these cards will be any worse than 19th and 24th, or 6th and 11th, or whatever).

The only difference that position causes is that first base is making his play based on the count and N cards left in the deck, 3rd base gets to see hit cards (H) and busted hands (B) before he makes his play, so it is based on the count and N - (H + B) left in the deck. 3rd base gets better pen and is making more accurate play decisions.

I have some results, but only for pure basic strategy.

I did not set out to compare results of various seat positions, but just kept track of the results of five computerized "players" separately, and then noticed there was a difference after a run of ~200 million rounds (i.e., ~1 billion player hands).

After noticing the difference (not a great difference, but larger than SOTSOD alleges is valid), I then ran tests with 3, 4 and 6 players at the table.

There is a consistent pattern, but the problem is that you're looking at a moving target. IOW, the best seat when you are playing with 2 other players is not the best seat when playing with 3, 4 or 5 other players.

In no case was first base the best seat. If you want more info, email me.

If you show a difference due to seat position for players playing strict basic strategy -- there is something wrong with your program.

Possibly your random number generator isn't exactly random enough.

I shuffle a deck of cards so they are in 1 of the 52! possible ways a deck of cards can be arranged. Now which seat will have the advantage playing against this card arrangement?

And if this seat position does have an advantage (obviously one position or the other will probably do better), why does it consistently have an advantage with 1 of the other 52! possible ways the cards could be arranged?

The program which produced those results uses a model of the ShuffleMaster machine, where the cards are dealt, picked up, and shuffled exactly as on an actual table (assuming the dealer never accidentally changes the proper pickup sequence).

The RNG is only used to do a "computer random shuffle" of both sets of cards in the ShuffleMaster machine every 300 shoes, as a way to approximate the casino changing cards every 24 hours.

The code is perfect, and the RNG is the "r250()" version widely accepted as being one of the very best available.

Since your assumptions rely on "randomness" and the ShuffleMaster shuffle, while very thorough, is completely deterministic, I think you are trying to compare apples to oranges. This is just more evidence that the cards do *not* come out of the shoe purely randomly.

Ergo, trying to measure the outcome using the "randomness" yardstick is simply incorrect and invalid.

Please describe this "ShuffleMaster shuffle" that you think you are doing.

The ShuffleMaster randomly selects a riffle size (sometimes it will shuffle every other card together 1 by 1, sometimes it will shuffle 2 by 2, every so often throw in a 3 card clump for one pile) as it goes thru the stack of cards. This random selection of 'grab' sizes or riffle procedure is duplicated twice. Once as the central stack of cards is split into seperate piles, and then again when the two piles are put back together.

Couple this with the 'horizontal' shuffle that takes place as the cards are played and hands are put into the discard tray in various orders due to busting, hitting, etc. and their is NOTHING deterministic about the cards on a ShuffleMaster table.

REPEAT ** NOTHING DETERMINISTIC ** REPEAT.

If the cards could be tracked in any coherent manner on a table with a ShuffleMaster -- the casinos would be broke. Do you think they actually pay more, in the form of high monthly lease fees to ShuffleMaster, for a machine that deals cards in a predictable manner?

BUT THAT IS A SIDE ISSUE -- HOW DO EVEN DETERMINISTICLY SHUFFLED CARDS FAVOR ONE PARTICULAR POSITION OVER ANOTHER?

Does someone cut the cards at the fantasy land casino where you play?
If the cut card is placed one card back 2nd base is getting the cards 1st base should of, 1 card forward and 1st base's 1st card is a card that shouldn't have even been played.

No matter how carefully you stack the deck, one random cut destroys it.

I actually had access to a 'real' casino ShuffleMaster machine for a time. I stuck cards into the machine numbered 1 to 312, and recorded the order of the cards coming out of the machine. Although the machine does sometimes 'zone' cards, the order and frequency of this happening is not consistent enough for any type of tracking to take place.

But, again, this beside the point. This non-random event is not enough to favor any one position at the table. Suppose I know that the first three cards in the discard tray will appear EXACTLY at positions 210,211,212 on the next deal. 1st base get T,5 hits and get a T, busts, and therefore these are the 1st three cards in the discard tray. I carefully count the cards as they are dealt, after using amazing eyeball skills to factor in the cut, and I now have 16 vs a dealer's T when it is my turn to hit with the 210th card. The procedure is random enough so I don't know if the cards are 5,T,T or T,T,5 or T,5,T.

Again, even with this knowledge, how does it favor one position over another, when all the players stick to strict basic strategy?

Although the machine does sometimes 'zone' cards, the order and frequency of this happening is not consistent enough for any type of tracking to take place.

That makes little sense. Either the thing is vulnerable to some type of tracking ie hypothetical computer-perfect tracking, or it ain't.

Although I agree with most of what you are saying, this line is definitely not true.

No matter how carefully you stack the deck, one random cut destroys it.

Yes, check here:

If the cards could be tracked in any coherent manner on a table with a ShuffleMaster -- the casinos would be broke. Do you think they actually pay more, in the form of high monthly lease fees to ShuffleMaster, for a machine that deals cards in a predictable manner?

Regardless of the merits or otherwise of the particular method in question, it should be perfectly obvious that the existence of some legitimate advantage play method is not going to beggar the casinos. Card counting and shuffle-tracking has not ruined the casino industry.

Cards that go into the machine on the top deck (usually the plug, if they just slap it on top of the discards before inserting into the machine) have a greater chance than probability would suggest to show up in the third deck from the top from the removed stack.

Have at it.