Backgammon Galaxy grandmaster Marc Olsen, author of "From Basics to Badass", "Pure Strategy" and "Cube Like a Boss" breaks down how he categorizes backgammon positions.
Would love to see video tutorials on each of the categories you have outlined in this video. What should we be doing in each situation, how should we be thinking - the do’s and don’t s of each. That would be fantastic and is really what I have been looking for which nobody else covers. You also have such a great style of explaining, would love to see you do this.
Awesome video! Thanks! EDIT: One thought I'm having as I begin compiling a database myself after watching this video, is that there are certain categories of error that aren't dependent on game type, but rather on thought process. Such as, "Wrong move at this score," or "Didn't consider the point switch," etc. So I'm making those categories as well. Another category I've got going is, "Don't know why this was wrong." Lol. : ) Anyway, these videos are super helpful for beginner / intermediate players like me, and I really appreciate everything you are doing for the game! Cheers.
Just read about "Stack and Straggler" today in BG Bootcamp in the Cubes and Races (Pips and Rolls) section. pg. 141-142. - Walter Trice. Great video, thank you Marc! Being able to catagorize positions over the board will help my brain make sense of the madness and form prudent strategies.
Hi Marc, Thanks for yet another great video. I’ve already watched it twice! I found it incredibly valuable. I have recently committed to building a library of positions and have become obsessed with categorisation strategies, so this came at the perfect time for me. I like the way you carefully define the characteristics of each. Many authors leave the categories a little vague, but you’ve distinguished each very nicely. I expect you’re familiar with the term MECE - mutually exclusive, collectively exhaustive. Personally, I don’t really like the “Middle game” category because it’s a bit open-ended. I call these positions “Evolving Games” with subcategories “Two Way Forward Game” (that can go either toward blitz or prime), “Proto Back Game” (similar to your Early Back Game category, acknowledging that the situation is still open to shift to Holding or Anchor) and “Post-Blitz Turnaround”. I think it’s important to view each position type from the perspective of both players - and sometimes this is overlooked. For example, your “Blitz” example is pure because while the opponent has a threatening prime, you’ve escaped it for now; if you hadn’t this would be more correctly “Blitz vs Prime”. Many people blur decisions/moves and positions. So they’ll add categories such as “Pay Now or Pay Late”, “Run or Stay” etc to the mix. This is a mistake. I believe the big position categories should be defined without reference to the dice rolled, the cube characteristics or the match score. You seem to agree with this general approach. (Another video would be great on any Decision/Move categorisation principles you may have - hint, hint :-) Some other category ideas for your consideration: I think your “Bearoff Against Contact” closed board variant versus checker on the bar example is important enough for its own category: “Post-Ace Point Game” (or “Closed vs Bar”) Your ‘goalkeeper’ variant of the Holding Game I term the “Anchor and Guard” game. I have a few categories you didn’t mention in my backgame hierarchy: eg “Super-Backgame” (when a very large number of checkers are back) and “Triple+ Backgame” (when 3+ points are made in the opponent’s home). By the way, I think Walter Trice coined the “Stack and Straggler” term. Whatever the case, straggler is an English word which essentially means “one who has been left behind the main group”. Thanks!
Stack and Straggler is a term I first heard used by Walter Trice when he wrote about the subject on the Gammonvillage. It's prototype position would be post-late hit where one side has many checkers off and got hit. The other side contained the checker and started bearing off, you open up, they re-enter and that's the straggler. One side usually has the stack and the straggler vs a normal looking bear off position. I'm 50% sure.
Another excellent lesson. I was a bit surprised that you use screen shots instead of xg files. When I review xg files I can do much more with the position, like changing the score, looking at variations and even replacing the position with a similar position that is a better reference for the future. My biggest problem with this type of study is allocation of time. It takes a lot of time to identify the error, catagorize it, save it, and later review it. With all you do, it amazes me you have the time to do this!
I think these conceptual categories of game 'types' is really useful, as you can begin to develop "ways of handling" various scenarios. As some others have suggested or asked about, though, categorizing by "decision type" is an equally valid approach to organizing a database of positions for study. So for instance, instead of an "Early Backgame" category I have one called "Backgame Commitment" which is for positions where typically I have a choice between giving up one of my anchors to "go forward" or making a blotty play in the hopes of improving my timing. Another category is "Backgame: Hit More Checkers?" where you are playing against the backgame. I also keep broad categories where I dump tough positions to return to later: "Backgame Techniques - As Leader" and "Backgame Techniques - As Trailer." Then when I have time, maybe I really try to look at "Trap Plays," where you try to force a checker off an opponent's anchor, I root through that folder for examples of that particular thing. Other "Decision Type" categories are things like "Slot or Not?," "Split or Not?, "Leave the anchor?" etc. Great stuff, Marc, fun to think about, so thanks for sharing your positional framework with us!
Albert Steg I think the way to do it is to organise flexibly with a tag hierarchy, so each position can have more than one attribute. That way, you can group on Position Type, Decision, Concept etc. Positions are complex and rarely may be reduced to one thing! Oddly, I don’t see much evidence that top players are using modern database strategies in managing their position collections. Most seem to be using old-school folders and word documents. Terje’s Backgammon Studio has the right idea (although I’m not always comfortable with the rules the engine uses for auto-categorisation...) I know you’re a FileMaker Pro guy (I’m a massive fan of Backgammon Taskmaster!!), so I expect you have a sophisticated personal setup in this regard. Any tips you’d be willing to share for those of us trying to build a framework from scratch?
@@ajitterbug Yeah, I use Filemaker rather than a folder system, so I'm able to characterize ('tag') individual positions in multiple ways (Match/Money; cube/checker; game type; early/middle/late). But what I found pretty quickly was that I wasn't really finding much use for all these tags afterwards because the positions they unite don't really have that much in common. So I moved to the equivalent of a 'folder' type system, where I create a 'parent' record (Topic), which can be broadly defined (Racing Cubes) or more narrowly defined ("Roll vs. Roll" bear off cubes or even "3-Roll vs. 2-Roll Cubes) -- then the 'child' records (Positions) are variations on this big theme. So, I tag the 'Topic' record with these various categories, rather than the individual positions, which saves a lot of time. For the Positions, I store a picture of the position and the XG result separately, so I can quiz myself, then have fields for 'Solution' and 'Comments'. But the Topics/Positions structure is pretty much just the same as having Folders/Positions. Glad you're enjoying the Taskmaster files -- they are an example of a single narrowly defined 'Topic' and its child 'Positions' wrapped up in a lesson format. Filemaker Pro, incidentally, is not difficult to learn -- anyone could buy the software and have a simple functioning database with all your positions stored in a single table (like a folder) and various fields for characterizing them. Learning how to work with a 'Relational' (parent/child) structure is a little more involved, but nothing like rocket science.
@Albert Steg Thanks! Your experience is greatly appreciated. I like very much the idea of a ‘parent’ record for grouping concepts and related positions (or just providing a collection structure for problems gleaned from a particular author or source). I still cling to the notion that a tag hierarchy might yield some fruit if implemented in a very structured fashion (probably because I haven’t tried it yet as you have!) Certainly, Backgammon Studio seems to be producing some interesting quizzes with this strategy. Some other ideas: I have been impressed with how modern backgammon authors such as Storer, Ballard and Bagai show illustrative offshoot ‘variants’ in addition to a main position - I’d like to create a structure for that. And Steve Sax once wrote a brilliant piece for GammonVillage called “The Equity Continuum” suggesting a mechanism to group reference positions sequentially by cube decision (all they way from “no double beaver” to “too good”), which I’d love to see implemented. David Presser and Igor Erovenko also wrote up some interesting ideas for grouping reference positions in types A and B (in PrimeTime BG) - I’d love to create a structure for this too! (I could go on...I have many, many overly ambitious thoughts on this!) I have a background in database design and software development, but as a Mac user, I hadn’t found the perfect database environment for personal tinkering. (In the old days, I used Microsoft Access for this type of thing - not an option on the Mac.) I think you’ve now pushed me over the edge toward FileMaker Pro - I had been on the fence mostly due to the very steep price tag... Pre-NY 2020 resolution: I will dive in and build a personal position database! (Now that it’s public, I’ll have to stop procrastinating!)
@@ajitterbug if you have background in database modeling, you'll find Filemaker super-easy to grasp. On the 'variant' approach, one way is to give your Positions a Number field, with letter appended to indicate variants -- 121, 122A, 122B, 122C, 124. Another would be to create a third level to the Relational structure -- 'Grandchild' records. The latter I think would be a little unwieldy for building the layout/interface. Another is simply to give each 'cluster' of variants its own Topic, and use the Topic Title to distinguish them. This will lead to a whole lot more 'Topic' records, but that isn't necessarily a problem -- you can always 'spin off' subsets of your Topics into separate database files -- which is essentially what I'm, doing with the taskmaster lessons. My full Taskmaster file has around 3500 positions as children to about 180 Topics.
@@ajitterbug The Sax suggestion is a great example. I would give each of these "Continuums" its own Topic, descriptively titled like "Cube Continuum: 33 Blitz" -- then the first 5 or however many positions would be the examples you describe. Then when this sort of situation arises in actual play, worthy of study or note, you just keep adding it to the topic, so long as you maintain interest in it. A nice effect of this approach is that eventually you feel like you've pretty well mastered a type of position, and you no longer feel a need to add further examples to the Topic.
hey Marc thanks this is really useful. As an intermediate player would I start by putting mainly blunders into these categories? Or any error? Or just interesting situations?
Question for Marc. Let’s say you have a position that has a particular character, say Prime vs Prime. Now you roll a 66 and have the chance to switch game plans. Do you file the position under PvP (before roll) or Bearin vs Deep Anchor (after roll)? I am currently going through the excellent Hickey/Storer Game Plan book and am struggling frequently with these types of position categorisation dilemmas! As I said in my previous comment, my gut feel is to ignore roll, match score and cube ownership, focusing on structural board features only when categorising. The category name therefore only reflects a PROBABLE assessment of the type of strategy needed assuming game plan remains constant... What’s your take on this?
I think in your discussion of Blitz around 5:30 wouldn't the salient point be more that your opponent is anchorless than that he's on the bar? Your opponent can be on the bar and you can have 12 in the zone, but if he's anchored on your 3-point, it's not generally a blitz (unless maybe he's got blots all over the place and you're trying to close out a bunch of them). On the other hand if he has two blots in your home board and no anchor and you have 11 in the zone, with a couple points made, that qualifies as a blitz, surely?
Hey I want to purchase the program that you are doing this on. On my Galaxy account, I can't seem to purchase these features or upgrade my account, it says it's still "under construction." Is it no longer available or am I missing something?
So this is off topic but how come I lose 27 points when I lose to someone 100 points lower than me but only win 11 when I beat someone 200 points over me? Why is it setup that you lose more points than win?
Marc, your dog sounds like he has heard enough about deep anchor games lol.
Would love to see video tutorials on each of the categories you have outlined in this video. What should we be doing in each situation, how should we be thinking - the do’s and don’t s of each. That would be fantastic and is really what I have been looking for which nobody else covers. You also have such a great style of explaining, would love to see you do this.
I think one of the most difficult things about starting a database is thinking about what categories to make - super useful structure you've provided!
Awesome video! Thanks!
EDIT: One thought I'm having as I begin compiling a database myself after watching this video, is that there are certain categories of error that aren't dependent on game type, but rather on thought process. Such as, "Wrong move at this score," or "Didn't consider the point switch," etc. So I'm making those categories as well. Another category I've got going is, "Don't know why this was wrong." Lol. : ) Anyway, these videos are super helpful for beginner / intermediate players like me, and I really appreciate everything you are doing for the game! Cheers.
Thanks a lot Sandy :)
Just read about "Stack and Straggler" today in BG Bootcamp in the Cubes and Races (Pips and Rolls) section. pg. 141-142. - Walter Trice.
Great video, thank you Marc! Being able to catagorize positions over the board will help my brain make sense of the madness and form prudent strategies.
Great video! Makes "Cube like a Boss" even more valuable. I will re-read it over X-mas holidays.
Well done Marc, nice reinforcement of the PBRC model.
great stuff Marc. Thanks for all the info.
hi thanks, i look forward to all your tutorials, luv from gridg
Really usefull videos thnx Marc. Btw just ordered your new book thnx for all the work you do for backgammon.
Hi Marc,
Thanks for yet another great video. I’ve already watched it twice! I found it incredibly valuable.
I have recently committed to building a library of positions and have become obsessed with categorisation strategies, so this came at the perfect time for me.
I like the way you carefully define the characteristics of each. Many authors leave the categories a little vague, but you’ve distinguished each very nicely. I expect you’re familiar with the term MECE - mutually exclusive, collectively exhaustive.
Personally, I don’t really like the “Middle game” category because it’s a bit open-ended. I call these positions “Evolving Games” with subcategories “Two Way Forward Game” (that can go either toward blitz or prime), “Proto Back Game” (similar to your Early Back Game category, acknowledging that the situation is still open to shift to Holding or Anchor) and “Post-Blitz Turnaround”.
I think it’s important to view each position type from the perspective of both players - and sometimes this is overlooked. For example, your “Blitz” example is pure because while the opponent has a threatening prime, you’ve escaped it for now; if you hadn’t this would be more correctly “Blitz vs Prime”.
Many people blur decisions/moves and positions. So they’ll add categories such as “Pay Now or Pay Late”, “Run or Stay” etc to the mix. This is a mistake. I believe the big position categories should be defined without reference to the dice rolled, the cube characteristics or the match score. You seem to agree with this general approach. (Another video would be great on any Decision/Move categorisation principles you may have - hint, hint :-)
Some other category ideas for your consideration:
I think your “Bearoff Against Contact” closed board variant versus checker on the bar example is important enough for its own category: “Post-Ace Point Game” (or “Closed vs Bar”)
Your ‘goalkeeper’ variant of the Holding Game I term the “Anchor and Guard” game.
I have a few categories you didn’t mention in my backgame hierarchy: eg “Super-Backgame” (when a very large number of checkers are back) and “Triple+ Backgame” (when 3+ points are made in the opponent’s home).
By the way, I think Walter Trice coined the “Stack and Straggler” term. Whatever the case, straggler is an English word which essentially means “one who has been left behind the main group”.
Thanks!
Brilliant instruction-very helpful.
Stack and Straggler is a term I first heard used by Walter Trice when he wrote about the subject on the Gammonvillage. It's prototype position would be post-late hit where one side has many checkers off and got hit. The other side contained the checker and started bearing off, you open up, they re-enter and that's the straggler. One side usually has the stack and the straggler vs a normal looking bear off position. I'm 50% sure.
Great video! Very pragmatic and helpful.
Another excellent lesson. I was a bit surprised that you use screen shots instead of xg files. When I review xg files I can do much more with the position, like changing the score, looking at variations and even replacing the position with a similar position that is a better reference for the future. My biggest problem with this type of study is allocation of time. It takes a lot of time to identify the error, catagorize it, save it, and later review it. With all you do, it amazes me you have the time to do this!
Thanks a lot phil!
Excellent resource. Thanks!
Glad it was helpful!
I think these conceptual categories of game 'types' is really useful, as you can begin to develop "ways of handling" various scenarios. As some others have suggested or asked about, though, categorizing by "decision type" is an equally valid approach to organizing a database of positions for study. So for instance, instead of an "Early Backgame" category I have one called "Backgame Commitment" which is for positions where typically I have a choice between giving up one of my anchors to "go forward" or making a blotty play in the hopes of improving my timing. Another category is "Backgame: Hit More Checkers?" where you are playing against the backgame. I also keep broad categories where I dump tough positions to return to later: "Backgame Techniques - As Leader" and "Backgame Techniques - As Trailer." Then when I have time, maybe I really try to look at "Trap Plays," where you try to force a checker off an opponent's anchor, I root through that folder for examples of that particular thing. Other "Decision Type" categories are things like "Slot or Not?," "Split or Not?, "Leave the anchor?" etc. Great stuff, Marc, fun to think about, so thanks for sharing your positional framework with us!
Albert Steg I think the way to do it is to organise flexibly with a tag hierarchy, so each position can have more than one attribute. That way, you can group on Position Type, Decision, Concept etc. Positions are complex and rarely may be reduced to one thing! Oddly, I don’t see much evidence that top players are using modern database strategies in managing their position collections. Most seem to be using old-school folders and word documents. Terje’s Backgammon Studio has the right idea (although I’m not always comfortable with the rules the engine uses for auto-categorisation...) I know you’re a FileMaker Pro guy (I’m a massive fan of Backgammon Taskmaster!!), so I expect you have a sophisticated personal setup in this regard. Any tips you’d be willing to share for those of us trying to build a framework from scratch?
@@ajitterbug Yeah, I use Filemaker rather than a folder system, so I'm able to characterize ('tag') individual positions in multiple ways (Match/Money; cube/checker; game type; early/middle/late). But what I found pretty quickly was that I wasn't really finding much use for all these tags afterwards because the positions they unite don't really have that much in common. So I moved to the equivalent of a 'folder' type system, where I create a 'parent' record (Topic), which can be broadly defined (Racing Cubes) or more narrowly defined ("Roll vs. Roll" bear off cubes or even "3-Roll vs. 2-Roll Cubes) -- then the 'child' records (Positions) are variations on this big theme. So, I tag the 'Topic' record with these various categories, rather than the individual positions, which saves a lot of time. For the Positions, I store a picture of the position and the XG result separately, so I can quiz myself, then have fields for 'Solution' and 'Comments'. But the Topics/Positions structure is pretty much just the same as having Folders/Positions. Glad you're enjoying the Taskmaster files -- they are an example of a single narrowly defined 'Topic' and its child 'Positions' wrapped up in a lesson format.
Filemaker Pro, incidentally, is not difficult to learn -- anyone could buy the software and have a simple functioning database with all your positions stored in a single table (like a folder) and various fields for characterizing them. Learning how to work with a 'Relational' (parent/child) structure is a little more involved, but nothing like rocket science.
@Albert Steg Thanks! Your experience is greatly appreciated. I like very much the idea of a ‘parent’ record for grouping concepts and related positions (or just providing a collection structure for problems gleaned from a particular author or source). I still cling to the notion that a tag hierarchy might yield some fruit if implemented in a very structured fashion (probably because I haven’t tried it yet as you have!) Certainly, Backgammon Studio seems to be producing some interesting quizzes with this strategy.
Some other ideas: I have been impressed with how modern backgammon authors such as Storer, Ballard and Bagai show illustrative offshoot ‘variants’ in addition to a main position - I’d like to create a structure for that. And Steve Sax once wrote a brilliant piece for GammonVillage called “The Equity Continuum” suggesting a mechanism to group reference positions sequentially by cube decision (all they way from “no double beaver” to “too good”), which I’d love to see implemented. David Presser and Igor Erovenko also wrote up some interesting ideas for grouping reference positions in types A and B (in PrimeTime BG) - I’d love to create a structure for this too! (I could go on...I have many, many overly ambitious thoughts on this!)
I have a background in database design and software development, but as a Mac user, I hadn’t found the perfect database environment for personal tinkering. (In the old days, I used Microsoft Access for this type of thing - not an option on the Mac.) I think you’ve now pushed me over the edge toward FileMaker Pro - I had been on the fence mostly due to the very steep price tag... Pre-NY 2020 resolution: I will dive in and build a personal position database! (Now that it’s public, I’ll have to stop procrastinating!)
@@ajitterbug if you have background in database modeling, you'll find Filemaker super-easy to grasp. On the 'variant' approach, one way is to give your Positions a Number field, with letter appended to indicate variants -- 121, 122A, 122B, 122C, 124. Another would be to create a third level to the Relational structure -- 'Grandchild' records. The latter I think would be a little unwieldy for building the layout/interface. Another is simply to give each 'cluster' of variants its own Topic, and use the Topic Title to distinguish them. This will lead to a whole lot more 'Topic' records, but that isn't necessarily a problem -- you can always 'spin off' subsets of your Topics into separate database files -- which is essentially what I'm, doing with the taskmaster lessons. My full Taskmaster file has around 3500 positions as children to about 180 Topics.
@@ajitterbug The Sax suggestion is a great example. I would give each of these "Continuums" its own Topic, descriptively titled like "Cube Continuum: 33 Blitz" -- then the first 5 or however many positions would be the examples you describe. Then when this sort of situation arises in actual play, worthy of study or note, you just keep adding it to the topic, so long as you maintain interest in it. A nice effect of this approach is that eventually you feel like you've pretty well mastered a type of position, and you no longer feel a need to add further examples to the Topic.
THANK YOU!
Nice video
You got stack and straggler right. :) Straggler is someone that's falling/left behind, not keeping up.
eftersläpare på svenska
hey Marc thanks this is really useful. As an intermediate player would I start by putting mainly blunders into these categories? Or any error? Or just interesting situations?
Question for Marc. Let’s say you have a position that has a particular character, say Prime vs Prime. Now you roll a 66 and have the chance to switch game plans. Do you file the position under PvP (before roll) or Bearin vs Deep Anchor (after roll)? I am currently going through the excellent Hickey/Storer Game Plan book and am struggling frequently with these types of position categorisation dilemmas! As I said in my previous comment, my gut feel is to ignore roll, match score and cube ownership, focusing on structural board features only when categorising. The category name therefore only reflects a PROBABLE assessment of the type of strategy needed assuming game plan remains constant... What’s your take on this?
I think in your discussion of Blitz around 5:30 wouldn't the salient point be more that your opponent is anchorless than that he's on the bar? Your opponent can be on the bar and you can have 12 in the zone, but if he's anchored on your 3-point, it's not generally a blitz (unless maybe he's got blots all over the place and you're trying to close out a bunch of them). On the other hand if he has two blots in your home board and no anchor and you have 11 in the zone, with a couple points made, that qualifies as a blitz, surely?
Blunderdatabase in galaxy would be usefull for this
It is coming soon!!
Tack för alla tips. Har du kategoriseringsorden översatt på danska?
Desværre ikke.
Hi, can you please define "contact value"? Thanks
The probability of hitting a turn-around shot and winning the game from there. For instance in a holding game, one man back or backgame.
👍❤️🎲🎲💪🍻
Hey I want to purchase the program that you are doing this on. On my Galaxy account, I can't seem to purchase these features or upgrade my account, it says it's still "under construction." Is it no longer available or am I missing something?
So this is off topic but how come I lose 27 points when I lose to someone 100 points lower than me but only win 11 when I beat someone 200 points over me? Why is it setup that you lose more points than win?
Dog disagrees
woof woof
Boys?