Pai Gow Tiles - x-ray vision

What is the effect on the house edge for Pai Gow tiles if all players are able to see each other's tiles before setting their hand?
The more players there are, the fewer 'unknown' tiles will form the pool from which the banker's hand may be drawn.
I am very interested to know what the effect on house edge will be for all numbers of players.
Thanks in advance for any help possible.

Quote: Arran65

What is the effect on the house edge for Pai Gow tiles if all players are able to see each other's tiles before setting their hand?
The more players there are, the fewer 'unknown' tiles will form the pool from which the banker's hand may be drawn.
I am very interested to know what the effect on house edge will be for all numbers of players.
Thanks in advance for any help possible.

It's hard to quantify, but note that if all the players could see each other's hands, and the maximum number of spots were in play, everyone could figure out exactly what the dealer's hand was, and set their hands accordingly.

Let me know if you find a place that deals all the dominoes face-up.

Quote: Arran65

What is the effect on the house edge for Pai Gow tiles if all players are able to see each other's tiles before setting their hand?
The more players there are, the fewer 'unknown' tiles will form the pool from which the banker's hand may be drawn.
I am very interested to know what the effect on house edge will be for all numbers of players.
Thanks in advance for any help possible.

My uninformed opinion is that the edge would drop as more players with middle hands would set them to push rather than win unless the strong tiles were in non-banking hands. I don't know about the actual numbers though.

I have played at tables where there is tons of chatter in chinese, where I would guess that the last player might know almost all of the hands. I believe in theory you are not supposed to speak in a non english language at the table, but I have never seen it enforced. If you knew all the tiles, then you would know exactly what the dealer's tiles are, it would become a substantial EV game. I expect that someone with better ability than myself could find the exact change in EV if they had access to exact house rules. I would surmise it would give a positive EV of 10 -15 %.

Quote: Arran65

What is the effect on the house edge for Pai Gow tiles if all players are able to see each other's tiles before setting their hand?

Beats me, but it doesn't seem to upset the casino much. I don't think jabbering away in a foreign tongue is of much concern to them, I think its more jabbering away in a foreign tongue and winning! That doesn't seem to be happening all that often so apparently the casinos don't much care.

I think standard policy is to keep one player location "RESERVED" and not deal out all the tiles to all the players for this very reason. Best case, if you knew all the players' tiles, you could only figure out a pool of 8 remaining tiles from which the dealer's 4 tiles are selected. Knowing how one tile can radically change the value of a hand, I don't think this yields enough information to make any significant difference.

I suppose there might be a few circumstances: If there are a few teen or day in the pool, you might assume there is a greater than normal likelihood that the dealer will be making a gong or wong, with the possibility of a small low hand. So given the choice, you might want to make a 3-5 instead of a nothing-8.

JB has done some analysis on this. I'll direct his attention to this thread.

Quote: FleaStiff

Beats me, but it doesn't seem to upset the casino much. I don't think jabbering away in a foreign tongue is of much concern to them, I think its more jabbering away in a foreign tongue and winning! That doesn't seem to be happening all that often so apparently the casinos don't much care.

This is very true.....our players are always trying to find out where the teens and days are by talking in an asian language.....I don't enforce the no talking rule because they will just use hand signals and peeking to get their answer.....once in a blue moon the table will get on a run and the floor will enforce the no talking rule. Because these players are inherantly gamblers and not AP, the money always comes back into my tray.

Quote: PapaChubby

I think standard policy is to keep one player location "RESERVED" and not deal out all the tiles to all the players for this very reason. Best case, if you knew all the players' tiles, you could only figure out a pool of 8 remaining tiles from which the dealer's 4 tiles are selected.

I agree. I haven't been playing tiles as long as some here, but I've never seen 7 hands dealt to players, or a table with 7 spots. Six is the max.

I guess JB has not had time to look at this, but he told me his previous analysis showed the benefit of collusion to be quite small and not worth the bother.

Quote: Wizard

I guess JB has not had time to look at this, but he told me his previous analysis showed the benefit of collusion to be quite small and not worth the bother.

That depends on exactly who's doing the colluding. There was a pretty big scandal involving marked Pai Gow tiles that were actually coming from the manufacturer that way. See Bill Zender's presentation on it here:
http://www.lastresortconsulting.com/pages/portal/polarized/rt_01.html

I played 50 hands on WOO using house ways and not banking. Bad streak of tiles for the player. Won 13, Drew 17, Lost 20. Had I been able to see the dealer's tiles , there would have been 7 changes from loss to draw, no changes from draw to win. 14% change in result I think makes x ray vision or full collusion at a table valuable. I think you would need to know the exact 4 last tiles rather than 4 of 8 to acheive anywhere near this level of value.

Quote: SOOPOO

I played 50 hands on WOO using house ways and not banking. Bad streak of tiles for the player. Won 13, Drew 17, Lost 20. Had I been able to see the dealer's tiles , there would have been 7 changes from loss to draw, no changes from draw to win. 14% change in result I think makes x ray vision or full collusion at a table valuable. I think you would need to know the exact 4 last tiles rather than 4 of 8 to acheive anywhere near this level of value.

Even 4 of 8 might be pretty valuable. For example, let's say I can rule out a pair for the dealer. Now I will be much more inclined to keep a pair in a weak hand (like 1-bo), or split a pair aggressively (like Sky-Sky-6-6). Or more like what many collusive players actually do, I can try to find out if all the 2s and 12s are already in play, and adjust my hand accordingly (for instance, I might feel a lot better about a medium 9 if I knew that the 2s and 12s were gone).

Quote: mkl654321

Even 4 of 8 might be pretty valuable. For example, let's say I can rule out a pair for the dealer. Now I will be much more inclined to keep a pair in a weak hand (like 1-bo), or split a pair aggressively (like Sky-Sky-6-6). Or more like what many collusive players actually do, I can try to find out if all the 2s and 12s are already in play, and adjust my hand accordingly (for instance, I might feel a lot better about a medium 9 if I knew that the 2s and 12s were gone).

Agree- there would be some value, but not near the level of value if you knew the exact tiles.

If a player knows the dealer's 4 tiles (and therefore how the dealer would set them), assuming a standard house way, here are the player's advantages:

Dealer banker: 6.9252%
Player banker: 8.4681%
Average (alternating): 7.6967%

These figures come from the book Finding the Edge, which has multiple authors. The Pai Gow tiles section from which this came was written by John Gwynn.

What I studied in particular was the situation where, at a table with multiple players, before you finished setting your tiles, you could see one or more Teen and/or Day tiles in other players' hands, and whether that information would be valuable enough to use. In the end, it didn't reduce the house edge anywhere near enough to warrant pursuing.

The majority of the strategy changes for 4-tile hands where no pair, wong, gong, or high nine could be made were so scattered that they would be impossible to put into words, let alone memorize.

Two strategy changes dealing with pairs stood out to me as interesting:

1) If you have a pair of 11's with H4 and H8 as your other two tiles, you should split the 11's to make 5/9 if you see at least one Teen and/or Day in another player's hand. This avoids wasting the H4 in a 2-point low hand. It also makes a strong 9-point high hand which is more likely to win because the chances of the dealer making a high nine are reduced since you saw a Teen or Day in another player's hand.

2) If you have a pair of 5's, a high 4, and a Teen or Day, you should split the 5's if you see a Teen and Day among other players' hands. I found this particularly bizarre since you are improving the low hand by only 1 point while degrading the high hand by 3 points (instead of 6/Pair the best play is 7/9). Of course, this situation and opportunity would rarely, if ever, happen.

When all is said and done, this method of trying to find an advantage was simply not worth the effort.

Quote: JB

What I studied in particular was the situation where, at a table with multiple players, before you finished setting your tiles, you could see one or more Teen and/or Day tiles in other players' hands, and whether that information would be valuable enough to use. In the end, it didn't reduce the house edge anywhere near enough to warrant pursuing.

Thanks. Good post.

In "Pai Gow Withgout Tears," Bill Zender talks at length about the possibilities of advantage play at Pai Gow tiles by marking the tiles:

"The effect of marking only the Teen (12) and Day (2) dominos and using that knowledge to vary the playing strategy when the banker does possess those dominos, will not gain the player any advantage. For example, the "knowledge" that the banker holds the 12 gives the player a 1.7% advantage if he varies his strategy. That advantage is then crused by the overwhelming fact that possessing the 12 gives the banker an advantage of over 13%. The player can not expect to gain any advantage unless he marks all 12, 2, 9, 8, and 7 dominos. By marking those 14 dominos and varying the playing strategy accordingly, the player will only have an advantage of just over 1/2%. If the player were to mark all 32 dominos and play his hand optimally against the known banker hand, the player can expect to win 7% of every dollar wagered. A small reward for committing a felony."

Note that Zender is talking about marking tiles so that the player can directly know the banker's hand. This is obviously far more advantageous to the player than deducing the banker's hand from knowledge of what other players hold. Even at a full table, one hand is automatically dead (not exposed to any player), so that a player can never have perfect knowledge of the banker's hand simply by seeing every other player's hand, so Zender's numbers would have to be revised downward rather significantly, I would think, to adjust for that case.

Zender reports a related experiment he conducted with a tiles expert, in which they marked the tiles of the four highest ranked pairs (Gee Joon, 12's, 2's, and high 8's) and used permissible cuts and moves to try to direct the desired tiles into the a desired hand. In 30 rounds (playing against a total of 210 non-banking players), he and his confederate won over 140 bets and lost only 12. This is a very different kettle of fish than knowing what tiles the banker has after the hands have been distributed (and note that it also assumes that the dice roll can also be manipulated to ensure that the "stacked" stack ends up where you want it -- no mean feat), but is interesting to note.

I haven't tried to verify Zender's math, but he is a highly regarded gaming consultant who generally knows whereof he speaks.