AtO meaning 'according to opponent' of course. While this deserves its own proper tome that nobody will ever write I wanted to give what I find a good example from one of my matches this weekend. I played a lot of matches this weekend, maybe half of them were recorded. I just finished entering the second one and going into the match I didn't know my opponent so I work on the assumption that I'm 3 PR better. This is quickly defined in most backgammon matches. You can tell not only by what moves a player makes but what moves they consider, how long they consider them, how confidently they make their moves, how quickly they find plays, etc...
My PR for the match unfixed or 'unadjusted' if you prefer was 3.5. My actual PR after I went in and debugged the choices XG doesn't understand was 2.5. (it should actually be ever so slightly lower in my opinion too as I should get credit for the decisions where I adjusted properly but XG isn't able to take that into account) I will list/show them all shortly and you can have your own thoughts on the matter. This is something I still think the players that are considered the best in the world fail miserably to execute in their matches. I maintain it is extremely hard to play consistently super low PRs (under a 3) if you're properly adjusting.
As I've said before, I slot all my opening non natural aces against almost all opponents and by and large, regardless of score. So the first two decisions that XG dinged me for were simple opening moves. I slotted with an opening [51] leading 7a 9a costing me .020 and I slotted with an opening [41] leading 4a 9a costing me .030. Now you may scoff at a .020 here or a .030 here but they add up. Especially if you're trying to grind out one of the lowest PRs in the world.
I think there are far fewer opportunities to tailor your checker player according to opponent than there is your cube decisions. There are certain checker situations that arise outside of the early early game but they are somewhat rare. Cube decisions you can be doubling too early because the opponent is likely to drop. You can double early because the opponent has shown they are hopeless in playing out the position. You can double late because they're a taker (normally a 'gammon ignorer'). You can (re)double late because it's a simple position to play out, you know they're taking, it's not a huge double, and you don't want to elevate the size of the cube and significantly shorten the match. You can play on instead of cash because you know even if they get back in the game they'll be so relieved to not be getting gammoned they'll pass anyway. And on, and on, and on...
Here we have position #1
                        |
| ||||
| XGID=aBaB-bCB-B--cA---cBe--A---:0:0:1:00:5:0:0:9:10 | |||||
| Blue on roll, cube action? | |||||
| Analyzed in XG Roller++ | No double | Double/Take |
| Player Winning Chances: | 63.68% (G:31.08% B:0.51%) | 63.81% (G:31.51% B:0.53%) |
| Opponent Winning Chances: | 36.32% (G:7.72% B:0.33%) | 36.19% (G:7.73% B:0.31%) |
| Cubeless Equities | +0.605 | +1.200 |
| Cubeful Equities | ||
| No double: | +0.590 | |
| Double/Take: | +0.344 (-0.246) | |
| Double/Pass: | +1.000 (+0.410) | |
| Best Cube action: No double / Take | ||
| Percentage of wrong pass needed to make the double decision right: 27.3% | ||
eXtreme Gammon Version: 2.10, MET: Kazaross XG2
It is important to be in the flow of the game sometimes too. When you look at a static position here on the forums there's nothing other than checking the score, analyzing the position, and coming to your conclusion. In the flow of this game my opponent kept getting hit and had just fanned two times on a three point board while I just tossed double sixes to do a lot of clean up and bring things around the board. I doubled instantly. He passed just as quickly. .250 error XG? Fuck you son!
Here is a second example from the same match.
         |
| ||||
| XGID=aDDBC----A------A---ccbbd-:1:1:1:00:3:0:0:9:10 | |||||
| Blue on roll, cube action? | |||||
| Analyzed in Rollout | No redouble | Redouble/Take |
| Player Winning Chances: | 76.48% (G:0.34% B:0.00%) | 76.66% (G:0.28% B:0.00%) |
| Opponent Winning Chances: | 23.52% (G:0.31% B:0.00%) | 23.34% (G:0.49% B:0.01%) |
| Cubeless Equities | +0.530 | +1.030 |
| Cubeful Equities | ||
| No redouble: | +0.742 (-0.002) | ±0.004 (+0.738..+0.747) |
| Redouble/Take: | +0.744 | ±0.007 (+0.736..+0.751) |
| Redouble/Pass: | +1.000 (+0.256) | |
| Best Cube action: Redouble / Take | ||
| Rollout details | ||
| 1296 Games rolled with Variance Reduction. Dice Seed: 23977590 Moves: 3-ply, cube decisions: XG Roller | ||
| Double Decision confidence: | 65.1% | |
| Take Decision confidence: | 100.0% | |
| Duration: 7.5 seconds | ||
eXtreme Gammon Version: 2.10, MET: Kazaross XG2
Now here you'll notice it actually is a double. I took some time over this one OtB and came to the conclusion that it was a no double I thought I won about 75% of the time and didn't think that was quite enough. I certainly don't want to jack the cube up to 4 and then possibly 8 if my opponent has any hint of getting back in the game so why did I double!?
I wasn't positive on this one like I was the previous cube. The previous cube I was certain that I was getting a pass. No doubt, like Federer being the GOAT or Neilkaz wearing a Hawaiian shirt at the next tournament he attends. Here however I guesstimated he would pass 50% of the time. I admit it is a wild, made up figure based on nothing more than gut feel. We have to realize based on the level of player what they likely do and don't understand. Who has a formula for this? You don't. Random player A doesn't. How are they supposed to make an informed decision based on anything other than experience or guessing? Their experience with this type of position isn't likely to be high. If it was there wouldn't be such a big PR difference between us. So working off all of that I decided to double and my opponent passed.
Now, I got 'lucky' that it was a borderline cube but even if it wasn't what do I care? How big of a ND would it have to be assuming my opponent did pass 50% of the time? Pretty damn large.
I love PR. You have to be able to play at one of the lowest PRs in the world to be considered among the best players in the world but I think we've went too far lately. There are too many PR whores. While I'm a PR whore to the max it has its time and place.
Stick