Großartig erklärt, wie jedes Mal. Danke für den Content und die Mühe dahinter. Mehr als 1000 Leute schauen dir beim "Training" zu. Das ist doch ein würdiges Publikum für einen Weltklassespieler. Ausnahmetalent
The cube blunder in game 1 reminded me of when you passed the no double vs Mochy in UBC. You were also trapped behind a strong prime but had counter play. Both positions were deceptive.
Hi Dirk, Great content! I have a question: how can I calculate an expected winning percentage of a 4PR player vs a 10PR player in a money game? is there any mathematical formula to calculate the edge a stronger player has over a weaker one?
Hi, unfortunately there is no such formula because PR always normalizes cube level to 1. That means you could even have a negative expected value in a session where you have the lower PR. For more detailed explanations I refer to the Theory of Backgammon.
There's a table floating around. As Dirk said, it's not exact but it gives you a rough idea. You can find it under "Backgammon MWC estimator by pr difference and match length". I also heard of the approximation of MWC=50%+sqrt(N)×∆PR, although I find both of these way too optimistic for the stronger player. In any case for a difference of PR of 6 and a 9-pointer, we'd get 68% (66.5% in the spreadsheet I mentioned). I think that's too high but it's probably a decent approximation.
In particular, Joseph Heled showed that the difference in skill (Elo) doesn't increase with sqrt(N) but much more slowly. Obviously long matches favor the stronger player but it's nowhere near as fast as a square root.
(sorry for the triple reply) A money game is a weird situation because the expected value is kind of hard to define (the expected value in $ is probably not infinite or anything, but it will have annoyingly thick tails). I'd suggest that DMP or a match of length N is an easier avenue to study.
Why doesn't Galaxy have a clock setting to correspond to major tournaments ? (Like 12/2 = 12 seconds per move and 2 minutes reserve, times the match length). Galaxy has choices of 10/1 (standard- Too fast) and 15/3 (casual- Too slow)
10/1 is kind of standard for online play. The logic being that you have a pip count already and you don't need to roll the dice. It's the most popular time control online (including on the WBIF which is the most popular online series). Although some tournaments do have 90 seconds/point with 10 seconds/move. PS I'm not a fan of Galaxy or anything lol, I never play there. But this is probably the reason.
Does anybody else wonder why Galaxy programmers don't do simple fixes (like a volume control and a toggle to turn off that annoying checker hockey feature they forced on us lately)?
Großartig erklärt, wie jedes Mal. Danke für den Content und die Mühe dahinter. Mehr als 1000 Leute schauen dir beim "Training" zu. Das ist doch ein würdiges Publikum für einen Weltklassespieler. Ausnahmetalent
The cube blunder in game 1 reminded me of when you passed the no double vs Mochy in UBC. You were also trapped behind a strong prime but had counter play. Both positions were deceptive.
Great work as ever. Thank you.
Can´t remember winning or losing a backgammon against just one checker on the bar without the board ever being closed,
Hi Dirk, Great content!
I have a question: how can I calculate an expected winning percentage of a 4PR player vs a 10PR player in a money game?
is there any mathematical formula to calculate the edge a stronger player has over a weaker one?
Hi, unfortunately there is no such formula because PR always normalizes cube level to 1. That means you could even have a negative expected value in a session where you have the lower PR. For more detailed explanations I refer to the Theory of Backgammon.
There's a table floating around. As Dirk said, it's not exact but it gives you a rough idea. You can find it under "Backgammon MWC estimator by pr difference and match length". I also heard of the approximation of MWC=50%+sqrt(N)×∆PR, although I find both of these way too optimistic for the stronger player.
In any case for a difference of PR of 6 and a 9-pointer, we'd get 68% (66.5% in the spreadsheet I mentioned). I think that's too high but it's probably a decent approximation.
In particular, Joseph Heled showed that the difference in skill (Elo) doesn't increase with sqrt(N) but much more slowly. Obviously long matches favor the stronger player but it's nowhere near as fast as a square root.
(sorry for the triple reply)
A money game is a weird situation because the expected value is kind of hard to define (the expected value in $ is probably not infinite or anything, but it will have annoyingly thick tails). I'd suggest that DMP or a match of length N is an easier avenue to study.
@@tgwnn thanks a lot for your answer, it is really informative ✌🤟🤟
Why doesn't Galaxy have a clock setting to correspond to major tournaments ? (Like 12/2 = 12 seconds per move and 2 minutes reserve, times the match length). Galaxy has choices of 10/1 (standard- Too fast) and 15/3 (casual- Too slow)
10/1 is kind of standard for online play. The logic being that you have a pip count already and you don't need to roll the dice. It's the most popular time control online (including on the WBIF which is the most popular online series). Although some tournaments do have 90 seconds/point with 10 seconds/move.
PS I'm not a fan of Galaxy or anything lol, I never play there. But this is probably the reason.
Does anybody else wonder why Galaxy programmers don't do simple fixes (like a volume control and a toggle to turn off that annoying checker hockey feature they forced on us lately)?