It depends on the length your games had (match length and game length).
For example for 50 1-point/single games of a short 20 moves per game(that means i've taken the best case so your case would have even less probability) the probability of not having a single double is:
(5/6)^(50·20) = 6.5 E-78 %
That means a 100% FLAW on dice generator!!!!!
But i guess you are overreacting and that "50" and "NO double at all" is not correct....
PS: To give a feeling of what 6.5 E-78 % is:
6.5 E-78 % = 0.000000000....another 65 zeros....00065 %
I just played through over 50 backgammon games without even 1 double roll. Am I just unlucky or is the random dice generator not doing such a good job?
alanback: That's much what GoldToken said. There's no need to protect the player from a catastrophic blunder, in fact slightly unfair to the opponent. But they disable the cube when it's dead for both players which, in my mind, contradicts that argument. Even if it isn't an established rule I'd still implement dead cube = no doubling.
grenv: As far as I know there is no rule against doubling a dead cube. Since by definition it cannot give the doubling player an advantage, I don't see why there should be a rule against it. The only harm that I can see is that an inattentive player might drop, but rules are generally not imposed to protect players from their own sloppiness.
I believe that the number of games required to "stabilize" the ELO formula is quite high, around 500 games. (On FIBS, the formula stabilizes at 400 experience points, I believe. Experience points are the total of all your match lengths, so a 7 point match counts as 7 experience points regardless of the final match score.) Until you reach that number of games (or matches, here), your ELO adjustments will be increased by a factor that gradually declines to zero (or 1, I suppose, since it's a multiplier) as you gain experience. The great majority of players don't have that much experience. It's difficult to speculate about the effect this has on average ratings, but it must have some effect. I think the most important consideration is that successful players tend to stay and play more often than less successful players. (This is a correlation, not necessarily cause and effect.) Thus, a player who enters the pool and initially loses a lot of points may go away, or even just change his identity, leaving the low rating in the pool of averages. I believe this tends to drag down the overall average. This is one reason I based my earlier speculations on the average of the top 50 established BKR. The drop-out element is unlikely to be represented in that group! (Putting aside those few players who enjoy initial success, fly to the top of the ratings, and stop playing to guard that inflated status).
Hrqls: For sure, it's crazy to double a dead cube. The question is whether it's against the rules to allow that option or is the cube disabled only because it's logical, and convenient for the player?
playBunny: it would be very bad to offer a double when you are just 1 away .. you gain nothing from it while your opponent might if he is more than 1 away.
Is it against the rules to double when the cube is dead? I know that Brainking, Vog, IYT and TrueMoneyGames disable the cube when it's dead. So does GnuBg and, I suspect, Snowie. But is that a matter of being sensible or is it a requirement? I've just discovered at GoldToken that at 1-away I'm still be asked whether I want to double or roll the dice.
playBunny: i dont have anymore data either :)
all we have seems to be the site for the chess formula which is used (but i think a bit altered) and our games data
also a few months ago the rating formula was changed so the data from previous games in my memory might have been with the old formula
(but i think the old formula favoured more games played much more by a higher rating)
pentejr is right, i have noticed that when playing a series of separate games against a similarly rated opponent, if we end up splitting, we both gain a few points in the end. i don't really have a problem with this as it rewards people for playing more and not sitting on a rating like I..er..I mean..some people do in chess.
As for Loop/Pahtum, that's easy. Pahtum seems to have an insane advantage for the starting player (maybe it's even a forced win with perfect play?), so playing both colors will shave the ratings down since if there's no or not enough skill involved, the cream will have trouble rising to the top. on the other hand, while loop does confer an advantage to white, like chess, it is not insurmountable and one can play a great game from either side.
Hrqls: I don't believe that's the case. There's one formula for the first 25 matches and another for match 26 onwards. Thus at 26, 100 or 10,000 it should make no difference. Achieving stability is a different issue. 25 matches in a luck-based game will often be insufficient to bring a player to their "true" rating but with the current formula that's not accounted for, given that it assumes that there's no luck. The proper Backgammon formula has a diminishing "bonus" factor which lasts until the 399th experience point.
But that's only my understanding from having read (over-complicated) explanations of the Chess formula and from a limited set of observations. The code actually in use at BrainKing is a mystery. Do you have any further data about the points awarded to matched players at different experience levels?
playBunny: when you have less games played your bkr will rise or drop more steeply
it takes quite some games to become a bit stable ... the 25 games which are needed to get an established rating isnt really enough ...
you can see that mjost clearly when playing opponents with roughly the same rating ... you will both win or gain the same amount of rating ... but this will be slightly more for a player who has played less games.
when i was around 100 game i think it was 13 points for me, now its around 6 points i think
(your +2 and -13 have an average change of 7 (absolute) .. your opponents -3 and +23 have an avera change of 13 .. this is probably due to the number of games played for each opponent
playBunny: But that is because one player has fewer games played. This will only continue to happen if new players are introduced, and the low initial rating will somewhat compensate for the increase... or something like that.
pentejr: It depends on who the winner is. The higher rated winner gets less and loses more. Vice versa for the lower rated player. But the two are not balanced and that makes your point substantially correct.
For instance with me, at 2430, against a player at 2150.
I'd go up 2 for a win or drop 13 for a loss. He would lose 3 or gain 23.
Thus the ratings pool loses 1 if I win and gains 10 if I lose.
Given that I'm never going to win 10 : 1 against any player, the rating pool will gradually increase in value from me playing this opponent. The same will happen with other pairings. This supports Alan's theory that the rise in Backgammon and Hypergammon is a result of the number of games played.
I have noticed several times that the winner gains more points than the loser loses. Every time this happens, more ratings points get "fed" into the system, so to speak, so that there should be a direct correlation between average rating and overall number of games played, in that game.
Even at that, the average backgammon rating cited below strikes me as high. A rating of 2219 would rank you #14 out of 902 established backgammon players right now, and I am playing #451 (at the moment), so essentially, the median rating, and that individual is rated 2012. You're saying the average backgammon rating is 200 points higher than the median? There aren't that many people that are over 2200.
Are you counting every ridiculously high provisional rating into that average? That will skew your figures. Better to average just established ratings or to weight each average according to number of games played. Not that I would take that much time figuring all this out, but that would tell you more about the big picture.
when only a few (relatively) games are played, the real top players havent been able to gain a real advantage in their rating yet as the opponents they did beat are just in the mid section of the ratings ... sub top players will have won about the same number of games, maybe a bit less, against opponents out of the mid section of the ratings .. as everyone is at that section when the game comes online first
when a game has been around longer then some people are around 2000 ... and the top players have been able to beat those 2000 players quite a few times raising their own rating ..
the difference between pahtum and loop chess probably comes from the fact that pahtum gives a bigger chance to win with white than with black .. while this isnt true (or at least much less) with loop chess .. therefore pahtum will have quite some draws (especially with the 2 games stairs) or an equal number of wins and losses between the same players .. in loop chess a real good player can win much more often than a real good player would be able to win with pahtum .. this will lead to a higher rating for the top player in loop chess
BIG BAD WOLF: There does appear to be a correlation between average games completed and average ratings. I compiled these for the top 50 players in each variant. Here are the results (Variant/average rating/average completed games):
Yes but there is indeed something strange about ratings:
Take for example Pahtum game and Loop Chess. Both have around 12000 completed games with Pahtum having even a little more games finished, but at Pahtum the rated players above 2000 are very much less than that of Loop Chess.....Can you or anyone exaplain this anomaly.....?
alanback: Have certain games/variants been here longer than others? More 2000+ rated players occur when there are more rated players. Look at the numbers for Espionage and it's variants. There are very few rated players, so no one is rated 2000+!
I am wondering what explains the different patterns of ratings in the various backgammon variants. I am puzzled why there are so many more 2000+ ratings in Hyper and regular Backgammon than there are in Crowded, Nack and Race. Is it possibly just a function of the number of games completed? There are probably many more completed games in regular backgammon because it is the standard and more popular than any of the variants. There are more complete games of Hyper because the games are short; it may also be more popular than the other variants. Does anyone have another explanation?
Question. Im white.. rolled a 2 and a 1` I want to use the 2 die so that I dont leave my self open, but it is making me move where I dont want to.. Is this a glitch?