If you're not a compulsive/addictive personality, there's no point playing backgammon. Phil Simborg.

I took backgammon a little more seriously in last couple of months, training at the laptop. Not intended to became a pro obv, but I have a new goal: moving to 7-7.5 Performance Rating.

Now I played a lot of cash games with XG and I am around 9 PR but I decided to move to match play because online play is always based on that format.

So I played a lot of 3 point match or 5 point match, and after a good sample I am around 9-9.3 PR (With checker play around 8-8.5 PR, but a not so good cube play).

I followed Robertie's schedule, he kindly gave to me a year ago when I started posting here on 2+2: I have a huge file of positions, explained and categorized into various themes (blitzes, prime vs prime etc) I studied deeper "Backgammon boot camp" by Trice and now I start having some benefits. Thank to that I moved to 11 PR at 9 PR, good.

But, as I said, I'm still stuck in the 9-9.3 PR range and I cannot see a way to improve further. I still try to cancel my errors studing them as above but When the sample gets bigger my PR curve always stabilize in the 9 range.

Things goes better in 3 point matches: due to some trivial cube play I perform as my checker play PR and I go to 8.9 8.7 PR but it's still 9 PR more or less.

What to do and what to study to get 7-7.5 PR? Take into account that I cannot follow a local club because the nearest to my city is at 100 Km and players aren't top players.

Thank for any advice.

I quit.

I consider myself a student of the game and past wins are just that - in the past. Oliver Cromwell said "When I stop getting better, I stop being good". So here is what I have done to improve lately.
1) Join the USBGF and watch Phil's videos. There are about 200 of them exclusive to USBGF members (and I am one of the non -American subscribers) You can get a taste for them on facebook https://www.facebook.com/media/set/?...3680398&type=2
2) read any thing you can. If money is tight then the absolute essential books are 501 Essential Backgammon problems, Modern Backgammon both by Bill Robertie. The Backgammon Encylopedia by Kit Woolsey, Backgammon Bootcamp by Walter Trice and the very underrated Boards, Blots and Double shots by Norm Wiggins. Mary Hickey's What's your Game Plan is also great.
3) Read all of Stick Rice's articles on Gammon Village. Then reread.
4) Read all the Problem of the week at this forum
5) Take some lessons with Phil once you have done the above. A cheaper alternative is mindgames.com which is good.
6) More and more I am focusing on what my game plan is. Sounds corny but it has really helped.
7) Develop a friend as good or better than you and talk about why you make moves as you play. I play every week with another good player online who lives 700 miles away. We talk about our moves. I am the better cube handler and he is the better checker player. We have both got better and we have been doing this for 5 years.
8) Try to so some thing Backgammon related every day.

And I have some serious time challenges, but still work these things in.

At a certain point, you probably need to do something like a meta-analysis of your errors. You've clearly got systematic flaws in your thinking. But do you even know what they are?

Where are your cube errors? Are they errors coming out of holding games? Priming battles? Are you dropping too much? Taking too much? Forgetting to cube too much?

Are you making mistakes because you're failing to take into account the match score when you make your moves?

Then you need a plan. Gather up positions where you've made errors. Categorize them. Analyze them. Make variants of them and run them through XG. Try to learn to sniff out the distinctions.

Beyond that, you can get into technical things. Not just pip-counting, but adjusted pip-counts and bear-off pip-counting. Learn how to compute your doubling window. Learn how to compute match equity. Learn to estimate your win/lose/gammon win/gammon lose probabilities by memorizing a bunch of reference positions.

My PR is around 9, but I've basically stopped trying to learn. I don't have the time to invest. But if I had the time, that would be the sort of thing I'd do.

I wonder what a 9 PR would translate to in a chess elo rating terms. My best guess would be a 9.0 player would be 1900-2000 FIDE while a 7.0 player would be 2100 to 2200 FIDE.

nonono not even close. 9PR is like 1700 max. There really is a major dropoff from playing above 5 to 4 and lower.

As for the original question, I would stop playing against XG and play against real people. It will be more experience related to real world playing.

I respectfully disagree, I play against live players once every two weeks (a DMP tournament for money with a 50% win rate so far), online once a week sometimes, twice (and I run the games/matches through XG) but play XG 6 times a week. I usually play a 7 point match against XG every day but sometime focus on play from positions playing. Live playing gets you use to the unexpected ie an open 21 played 24/22 6/5 or opening 41 played 24/20 6/5 (both of which has happened to me) but to learn from your mistakes play against XG.

There's some truth to this, though it's not directly tied to improving your PR (though it might do so indirectly). Playing against real players will put you in positions that you will NEVER see playing XG alone. It will give you food for thought which may help you to learn to analyze positions better because you'll have a larger set of positions in your experiences.

But if you run those matches through XG, I'd expect your PR to be a little bit worse than when you play against XG.

I have heard the argument about XG not getting into unusual checker positions before, but I don't really buy it. All sorts of checker positions arise all the time, both against humans and against the bot. And when a human makes gross errors in checker play, that usually just makes the game easier.

Nack Ballard has proven time and again that he can come up with a well-played (i.e., bot-like) move sequence for any position you offer. I have seen him produced 25-roll sequences more than once. It's a fun diversion called "retrograde backgammon." (On a good day, my limit is about 10 rolls!)

What I do agree with is that cube play is much different against mere mortals. As others have pointed out, the fact that almost every cube offered by the bot will be a correct cube simplifies the decision tree for takes. David Rockwell explained it best. In a non-volatile situation where the bot has not just rolled a joker (or you an anti-joker), the bot will probably have gotten in its cube before losing its market.

You should be thinking take.

So I do agree that practicing against a bot won't give you all the training you need for doubling decisions.

Another critical issue is pressure. Against a human, whether in a tournament or for stakes, the pressure is much higher than when you practice at home with a bot. Learning to deal with that pressure requires putting yourself in the situation as often as possible.

Mike

I agree that human errors make the game "easier" as in "easier to win matches." But it doesn't mean that your PR will be the same. You will often have several options, all of which are objectively "good" but only one of them is the best. And you can pick the wrong one and still have a big gain in equity. And your PR can still get worse.

Yes, but if it requires that type of sequence to get there, the chances of you seeing it by playing against XG is very small.

Another thing about live play is getting used to the physical mechanics of a clock (if you're playing in a clocked tournament, and that seems to be the direction that things are moving). It also adds pressure on you if you're used to a very meandering pace of thinking.

I'll second this. I used to play against XG but use the hint feature with moves. That resulted in very normal games. When I was playing live I'd then find myself in situations completely alien to me.

The same is happening again now that I've started playing live online matches. Playing only against XG I don't think does you any favours. A mix seems preferable.

I have read carefully your comments and I found something very interesting, so thank you very much.

Second I want to clarify my postion, in order to get more and more precise advice. I keep me humble, because I think it's the only way to improve. (

If you have the time, read until end please. )

But I think I'm a good intermediate (I divide BG players into these cathegories: clueless beginner, beginner, weak intermediate, strong intermediate, weak advanced, strong advanced, world class, elite, ) and I want, as I said, to became at least stable "weak advanced".

I can spend no more than an hour per day to backgammon, and I have read and reread and studied (for real) carefully 501 problems, Advanced BG both volume and BG boot camp. I mean truly studied, not fast read or read at the park.

I have built a huge database of reference position by myself and it was a huge work, like preparing an university exam.

My problems remain cube action, and playing position with many men back (3+) either my men or opponent's. I usually make good take/pass decision, but the few times I'm wrong I am wrong by a lot (as the position of the other thread.) And it's frustrating because I play say 10 games at World class cube play, even 5 points match, and SBANG a -0.4 cube blunder in a difficult position and I jump to advanced level and sbang another and it eventually stabilizes in the 12-15 range.

I usually make no more than two/three blunders on average around -0.1 on checker play in a average game, and the times I make -0.2 blunder came almost always from "many men back".

This two part of the game are not so much covered by books or online articles and I try by myself but it seems that I can't get them.

You mix all this with small errors due to several sub-optimal move (say 0.04-0.06 green errors, very very hard to discriminate from the optimal one) and sometimes a tendency to double a little later than the optimal point and zac... 9 PR. Oh boy...

Post a few of these positions with 3 or more men back. Chances are you're missing a couple of key concepts that we can clear up.

The size of the database isn't the important thing. Finding good representative positions is important. It's still work, though. And there's no avoiding that.

Presumably, the times you're wrong by a lot are happening not because of careless play, but because of wrong thinking. Go study those situations and find out what is going wrong in your thinking.

Next time you play and you find yourself at the beginning of one of those positions, save it. Then play several games starting from that point. See if you can't discover patterns in your errors. It's easier to find the patterns if you reduce the noise. And you can reduce some of the noise by starting closer to the problem area than starting at the beginning of a game every time.

But that's the whole point. Give Nack any screwed up position that you think only a human could create, and he can find a well-played sequence that gets to the same position. That's a pretty good argument that anything you see against a human, you might also see against a bot.

I suppose you might argue that when only one side (the human) makes mistakes, you can fall into a position that would seldom arise when two humans play each other. I am skeptical, however, that that is true. Any messed up position that arises when a human plays a bot is just another messed up position in my mind. I don't see why it could not occur between two humans just as easily as when a human plays a bot.

Cube action, and in particular, the take decision, is decidedly different against a bot than it is against a human. Whatever differences exist in the kinds of checker positions you might face are much, much smaller.

Mike

Again, it's not that it can't happen, it's the frequency with which those positions happen. Think about it like this:

Imagine a 3 roll opening sequence in which one or more players makes a sub-optimal play. Playing against a bot, you will never see this position after 3 rolls, but maybe it's possible to get there with some 40-roll sequence.

Then by sheer probability, you won't have any realistic expectation of seeing that particular position when playing the bot, but it can certainly come up when humans play.

The mix of positions you'll get by playing against XG is very different from the mix you'll get playing against a weak human player.

With all due respect to Nack, plenty of these positions could never be reached by a well-played sequence of moves. Here's an obvious example:

Beginner: 52. 13/6
Good player: 31: 8/5 6/5

Beginner: 43: 13/6

I'll wait for a well-played sequence that gets to this position.

One reason for playing lots of games against humans is precisely to get some experience in playing these poorly-played situations.

Try some lessons.

Do you play on gridgammon? If so, I may be able to give you some freebies.

Here's the problem I have - I have seen a lot of people tell you you need to be able to calculate your Live and Dead take points and gammon values over the board. If you can do this great I cannot so I do not try to do it. Especially with clocks becoming more common.

So what I have done is memorise my Dead and live Take points for every score in a 9 point match for a 2 and 4 cube. Also my GV's for a 2 and 4 cube. I go for an hour walk every day and go over them and some other stuff then. If you know them then calculating your GAT (gammon adjusted take points ) is faster. Start with just the 5 point match take points and GV's for 2 and 4 cubes then expand it to 7 then 9 point matches.


Also get hold of kit Woolsey's article on the 5 point match which is very practical. Also underrated is "Cubes and Gammons at the end of a Match" by Antonio Ortega for good advice though it is quite mathmatical in places.