Posted on

3-pointers – Should be easy, right?

3-pointers - Should be easy, right?

Short matches are fun and they don’t require a lot of time to play. At the time of writing I’ve played 139 3-point matches on Galaxy this year (I’m playing as “Hallberg”). In this blog I’ll go through some of the things these matches brought to light.

Catching up

Since last blog a lot of stuff happened. I went to the 7th Merit Open in Northern Cyprus with my girlfriend to play the tournament. The results were mediocre but the tournament and surroundings were awesome. We immediately booked a room for the fall tournament in 2020.

Late 2019 I started in a new job managing IT-outsourcing – now from a customer perspective. Most of my career in IT-outsourcing has been delivering IT-services and I thought it would be interesting to experience receiving them. This meant less time for hobbies where my favorite pastime was playing strategy games on my tablet. Instead I chose to spend the little free time I had on playing 3-point matches online.

Playing online

My two main purposes of playing online is to

  • get input for where to put my effort when learning from my mistakes (keywords: Strategical errors, technical errors, time management)
  • practice regularly to an extend where I’ll never be rusty when going to a real life tournament (keywords: Flow, intuition, speed, repetition).

There are many different speed settings and I chose to go with 1 minute per point in the time bank plus 10 seconds per move (“Normal” speed). I think that this mimics the speed of making decisions in a real life match in the best way.

I typically accept challenges from anyone but I rarely get challenges from higher rated players. There seem to be only one solution to that problem and that is to become higher rated myself. Recently I changed my policy of playing everyone to only play against players within 250 rating points. It removes a lot of the variance of being run over in a simple game by a significantly lower rated player loosing 20+ rating points, standing to gain only 6 points when winning. Below is a graph showing how much you win or lose with a certain rating difference in 3-point matches on Galaxy.

Won or lost rating in 3p matches based on rating difference

I also plotted my opponents’ PR in the below table and they are following a nice trend. Looking at the 1400-1500 groups we see that they are playing significantly below expectation which I believe is due to new inexperienced players starting at 1500 not having found their level yet. The same goes for the 1500-1600 group where good players are on their way up to find their right level. The bump at the 2000-2100 level is due to a player losing on time getting a huge PR penalty.

Average PR in 3p matches for my opponents

Thinking about the 3-point match

Since I started playing backgammon back in 1995 I’ve gone through some phases regarding match and money game. In the early years I preferred match play for the sake of all the adjustments. Later on I realized that for focusing on understanding the strategy of the game money game is much better because you remove the “noise” of having to adjust for match score. In the later years where my activity level has been very low I only played a few games of money game with XG on my phone. In the few real life matches I played over those years I got so many huge mistakes because I forgot to adjust for match score.

The 3-point match is interesting because you don’t play like money game in any of the match scores. This forces me to think and be aware of the score all the time. You have it all: Double Match point, Gammon Go and Gammon Save. Naturally this will show if you are capable of making the proper adjustments. I have not succeeded yet.

Before I stared I expected to play around 5.0 in PR i 3-point matches. Most likely a little below. After 139 matches I’ve played on average 5.3. Split up in checker play and cube action I played 4.6 in checker play decisions alone and 10.5 in cube actions alone. A good rule of thumb is that cube is typically double of checker play. I’ll continue to play 3-point matches until I reach an average PR of 4.5 over 100 matches. Then I’ll switch to 5-point matches.

Strategical errors

Typically these errors are quite big and sometimes huge blunders. Most of the improvement in PR should come from these.

Priming games are difficult in general. It often takes courage to play the right moves and they can be very difficult to spot under time pressure. One common theme is that I typically underestimate how difficult it is to escape a prime. This leads to a lot of mistakes where I

  • take a cube where it’s much harder to escape than I thought.
  • do not double because I believe that my opponent will easily run away.
  • do not activate my rear checkers when starting to get primed.
  • do not play hard enough for a prime myself.

Example 1: Priming game (late to the party)

Position 1: White on roll, 0-1/3

General observations

  • In 0-1/3 you double very aggressively especially if there are gammon chances
  • A little bit down in the race
  • Both having two checkers back
  • Both having 5-primes
  • White has better timing

I viewed the position as having an advantage but not decisive. Black is closer to the edge of the prime and I’m only 9 pips behind. Not a lot of timing. What I clearly underestimated was how much time I actually have to move my three outfield checkers around waiting for Black’s position to fall apart.

It was a huge blunder not to double and an even bigger mistake if Black took the cube! So a huge D/P at the match score. From 0-0/3 it is still a huge double but now also a clear take.

Example 2: Priming game (defending by attacking)

Position 2: White to move 64, 0-0/3

General observations

  • 0-0/3 you double early and pass early
  • Up 30 in the race
  • Both anchored up
  • Black owns the cube
  • Seems like an equal game

With the roll 64 it is clear that you have to jump the prime. There are two options. Running all the way with 21/11 or playing 21/15 6/2*.

My thought process was something like this. The advantage of running all the way is that you run a small risk of being hit and you are ready to jump the prime next time. By hitting on the deuce point I will give both deuces and threes to hit and even potentially send another checker behind the prime. Should be clear?! I ran and got a huge blunder (-0.141) for that.

After thinking about it more I got to the conclusion that I highly overvalued my chances of jumping the prime next time if I get the chance. By giving Black the whole roll it is much easier to attack me while making a point (or jump my prime) at the same time. Hitting on the deuce point prevents Black from launching a really strong attack against me.

Example 3: Priming game (over-adjusting)

In the below example the score is 0-0/3. White is on the bar and thinking about a cube.

Position 3: White on roll, 0-0/3

General observations

  • In 0-0/3 you double early and pass early
  • 48 pips down
  • Black not at the edge of the prime
  • 5-prime is very strong
  • White has bad numbers from the bar.

My thought process was something like this. If I enter and jump out with one checker it will be very difficult for Black to take. Only 33, 31 and 22 are bad. The 9 dancing numbers are of course not great either. If black rolls an ace it would have to be played 23/22 not building the board at the same time. The race is not that important because I’ll most likely get the outfield control along with my 5-prime.

I doubled and my opponent passed. Most likely I would have accepted the cube myself. It looks like a reasonable take. Things still need to go well for White before Black is in trouble.

To my big surprise this turned out to be a huge (-0.135) no-double/take. But why?

Thinking a bit more about it I realized that the dancing numbers are actually really bad too. It gives Black time to either roll an ace or build a point in the board gaining even more time on me. It is also more difficult for me to move freely against a stronger board. Look at the next 4 variations of position 1 for how much better you need to be to cube.

Position 3a: White on roll, 0-0/3. ND/T.
Position 3b: White on roll, 0-0/3. ND/T.
Position 3c: White on roll, 0-0/3. Small double. Huge take.
Position 3d: White on roll, 0-0/3. Double. Huge pass.

Contact value when slightly behind in the race. The rule of thumb is that you should go for more contact when you are behind and less contact when you are ahead in the race. This can lead to big mistakes when you are close in the race. I undervalue contact value in those situations. Often I break contact too early when not needed and sometimes I wait too long to run to maintain my racing chances.

Example 4: Contact value (slightly behind in the race)

Position 4: White to move 44, 0-0/3

General observations

  • 0-0/3 you double early and pass early
  • 1 pip behind in the race after the move

In general White should not break contact being behind in the race after the move. When the race is this close I don’t like to end up being stuck on the opponents bar point potentially being squeezed of it or have to bury checkers to avoid running. For that reason I chose to play 18/14(2), 13/9(2). This is almost a blunder (-0.067). The natural looking 13/5(2) is of course the best.

Technical errors

In general these errors are not big but there should be very few of them. Unfortunately seeing these moves quickly is one of the things that disappears when not in shape.

I will not go through them in this blog (maybe in a later blog) but here are the three categories where I drop equity that should not happen:

  • Duplication
  • Minimizing shots
  • Tempo hitting

If you liked this blog please give it a thumbs up and share. Until next time, take care.

Posted on

Preparations for a big tournament

Finally, I managed to find time for a real tournament. My eyes are set on the 7th Merit Open Int Backgammon Championship held in Cyprus early November. Based on peoples previous experiences with that tournament it will be a nice experience.

This blog is about how I prepare for a larger tournament, especially when out of shape (in more than one way). The main goal is to be ready for the tournament to a degree where I believe that I could win it! How do I plan to get to that point?

I’m very structure and process oriented. It may seem boring and tedious but I find comfort in having everything in a system. This way of thinking has brought me many practical challenges because I’m heavily biased towards consistency rather than what works. If it does not make sense to me I won’t do it even though I objectively understand that it is right.

Then the question is: Can you put Backgammon into a system? To a very high degree, yes!

Guiding principles for improvement

  • Strengthen basic understanding of the game
  • Work on specific problems in your game (learn/understand new concepts)
  • Generalize specific problems and solve the specific problem with the generalized model
  • Get feedback through analysis of your own games and training session with other players

My main issue: Thinking vs. knowing

Thinking is easy: Requires effort here and now
Knowing is hard: Requires planning, hard work and discipline

I’m lazy. Not in the sense of lying on the couch in front of the TV all day, but in the sense of relying on thinking rather than knowing. One of my strengths is that I’m very good at focusing a lot of energy into solving a specific problem here and now. The flip side is that I never really appreciated what hard work can achieve because a focused effort with minimal preparation would get me through almost any challenge. This is basically the story of my backgammon life too. Let’s take a closer look at the current situation in relation to backgammon.

Past experience tells me that trying to patch up the holes in my understanding of the game will fail (trying to think instead of knowing). It is too short sighted and require way too much energy when sitting at the board trying to make it work. Generally speaking I tend to spend too much time pondering over positions while playing a tournament (trying to think instead of knowing) . This usually brings me into time trouble resulting in even more mistakes.

When you are playing a tournament you should focus on utilizing what you already know and when you are in a training setting you should focus on getting to know more.

Clearly, I’m breaking my own rule but I intend to do better. By identifying the types of positions where I spend a lot of time I can go back and analyse those situations in depth to actually understand them. This will save me a lot of time in future situations. Later I’ll go more into exactly what type of problems I need to focus on.

Basic understanding of the game

There are a lot of things that we know about backgammon – a lot of things where we have a clear answer. This knowledge has been put into many good books. Basically everyone could sit down and learn these things and become an expert player. Of course it requires a significant effort but you don’t have to think a single original thought yourself.

Here is my recommended reading list for basic knowledge (sorted: Easy to hard):

  • Backgammon From basics to badass, Marc B. Olsen
  • Opening Concepts, Michihito Kageyama and Roland Herrera
  • Endgame Technique, Michihito Kageyama and Roland Herrera
  • Backgammon Boot Camp, Walter Trice
  • Backgammon Pure Strategy, Marc B. Olsen
  • Modern Backgammon, Bill Robertie

In my own preparations I’ll read quickly through Michi’s two books (448 pages) and spend some time studying cube action in Pure Strategy (179 pages) and Boot Camp (81 pages).

Specific problems in my game

To be as efficient as possible I’ll address the specific issues that I’m actually facing in my matches. After a match I’ll go through the whole match move by move and try to formulate what is going on in the position. When in doubt I test my hypothesis by changing features of the position to see the effect and then restate my understanding of the position.

Recent matches indicate that I struggle heavily with cube action and safe vs. bold positions. In future blogs a lot of examples will be analyzed.

Generalizing a problem

I learn through conceptualizing things. When faced with a specific problem

  1. I try to conceptualize it
  2. Solve the conceptualized problem
  3. Use the solution on the specific problem

This way I am solving a class of problems instead of a single instance.

Feedback

Most of the time I’m trying to update my conception of the game but it is a challenge doing this only with a computer. For this preparation I’ve asked two Grandmasters (Karsten Bredahl and Marc B. Olsen) to have a few session with me based on the themes that I have been working on.

The next blogs will be centered around the things I’ll learn and work on. If we are lucky there is a chance that one of the training sessions with a Grandmaster will be recorded for publication.

Posted on

Learning from mistakes 4: Conflicting concepts

Yesterday I played a 3-point match where everything went wrong. Somehow I managed to use the wrong concepts in a lot of positions. It is rather painful to go through the game step-by-step but that is in fact the most important part of improving.

The steps are

  1. Understand why you came to the wrong conclusion.
  2. Understand how you come to the right conclusion.
  3. Review the positions multiple times until it becomes natural to you what the right concepts are.

In the following part I have selected two of the positions and go through step one and two.

Example 1

Score: 0-0/3 – Cube centered. White to move 43.

I looked at the following concepts/rules of thumb:

  • Don’t split when your opponent made the 2-point early in the game. By splitting you give the opponent a chance to attack you and get value of the 2-point.
  • Don’t be afraid of being blitzed when the attacker has 8 or less checkers in the zone .
  • Don’t be afraid to split when your opponent has striped the 8-point (only two checkers on the 8-point).
  • Unstacking the heavy mid-point is good.
  • Bringing more builders into play.

Based on these ideas I have three options:

  1. 24/20 13/10 (going for better anchor)
  2. 24/21 13/9 (going for less shots)
  3. 13/10 13/9 (going for more builders)

Choosing between option 1 and 2 heavily relies on the amount of good rolls Black can roll. Black will hit the checker on the 10-point with both 54 and 63 where only 44 and 62 will hit on the 9-point (22 will make the 4-point).

On top of that Black is more happy to hit loose on the 5-point with aces because they are not especially useful anywhere else. We can conclude that 24/21 13/9 is better than 24/20 13/10.

The above conclusion points to a concept that I forgot in my analysis done over the board: Efficiency of opponents next roll.

Now, choosing between option 2 and 3 is difficult. Looking at the concepts option 3 seems to meet most of the criteria. Option 2 win more games where option 3 win more gammons. It turns out that they are basically equally good with a very small bias towards option 3.

Example 2

Score: 0-0/3 – Cube centered. White to move 32.

Concepts that I took into account:

  • Pip-count: Down by 18 before the roll
  • Risk of being hit
  • Relative board strength
  • Builders for attacking
  • Unstacking the mid-point (13-point)
  • Leave 3 checkers on the mid-point

Again I considered three options:

  1. 13/11 13/10 (going for prime)
  2. 13/10 6/4 (going for attack)
  3. 13/8 (going for safety)

Analyzing option 3 is easy. We are behind in the race and it is not super risky to be hit because of the blot on Blacks 4-point. These factors are enough to dismiss option 3.

The next question is should we attack or should we prime?

Pro attack (option 2):

  • Only one blot
  • 10 checkers in the zone for a quite strong blitz next roll
  • Behind in the race
  • Black has a blot in the home board

Con attack:

  • Difficult to attack only one checker
  • Black has better relative board strength
  • White is not lost in the race

Pro prime (option 1):

  • More checkers in the zone
  • More control
  • More possibilities to make point next roll
  • Behind in the race

Con prime:

  • Two blots (more risk)
  • Not leaving 3 checkers on the mid point

Taking all these considerations into account one point becomes clear. We do not have a position where we can win here and now. If we have to build up to an attack or a priming game the latter has way more long time potential. This takes a lot of power out of the attacking plan and leaves us with the priming option.

The missing piece of the puzzle was that the attack is too slow and it does not have the necessary power.

Good luck with your training. See you at the tables.

Posted on

Learning from mistakes 3: When the hand is faster than the mind

Today’s topic is centered around these kind of very annoying errors where the alarm bells go off too late – like one second after you made your move.

A nice vacation with time to read some (backgammon)books and relax has come and gone. Thankfully I only have one more week at work before I have another two weeks of vacation. This time I’ll go to England and play a BMAB tournament. These kinds of tournaments have focus on how well you can play according to the computer programs. This is more a battle of theoretical prowess than a battle to score the most points.

A lot of time has gone into preparations for this tournament. I have been studying some reference positions, gotten some inspiration from books and last but not least a lot of practice against other people. XG Masters did not go well but that was a wake up call for me to also get some experience playing against people.

I have played more than 100 matches at Backgammon Galaxy with a satisfactory result. The backgammon server is still in beta but overall it is working well. If you would like to sign up and try out a few games you can use the beta code “galaxystar” when signing up to gain access.

Example 1

In this example I’m playing the white checkers and rolled a nice 62. Without spending more then 2 seconds on it I moved 13/5*.

The score is 0-0/5 and I have already doubled and of course I’d like to win a gammon and why not go for the win (gammon) with such a strong roll? When looking a little closer it becomes clear that White has really good chances to win the game with very little risk by just running. White is ahead 18 pips before the roll. Black has a nice prime that White would do best not to get closed in behind by being hit on the 5-point.

Suddenly it seems very clear that the right move is not to hit and just play the safe 13/7 6/4.

Example 2

This one still hurts and I’m sure I’ll wake up in the middle of the night screaming thinking of this. The score is 0-0/3 and I rolled 51.

I start by hitting with 13/12*. Then I look for the 5. Playing either 13/8 or 22/17 sucks, so I make a quick decision to jump the prime while Black is on the bar.

Basically everything went wrong here. After my move Black can actually double from the bar and I have to pass it! Looking at the position is it clear that Black will soon have some issues with his timing (I’m down 27 pips in the race before the roll) where he can be forced to leave his anchor and most likely also leave a blot too. If I just play the simple and beautiful move 13/8 6/5 I’ll have all the counter-play I need.

Example 3

The next one is a bit tricky. Once more the score is 0-0/3 and I rolled a nice 44. But it was not quite clear what needed to be taken into consideration.

Making the 4-point is quite clear. Should I go for the nice 5-prime with 6/4(2) or should I go for more control with 24/20(2). Both candidates look fine but there is an interesting point here. I would like to be able to double next roll. By playing 24/20(2) I’m almost certain that I have a nice double next roll. What I forgot was that I in most cases will play on for a gammon if I make the 2-point which is also clearly the strongest move.

Unfortunately I made the 20-point which is a huge mistake at this match score. In money game the two moves are equally good because you need to have doubled to win a gammon. In both situations you would end up doubling next roll with Black forced to pass.

Good luck with your training. See you at the tables.

Posted on

Learning from mistakes 2: Is it too good to double?

This problem is from yesterday’s XGM match against Ole Nielsen. We are in the 4th game where I’m leading 5-1 to 17 (5-1/17). I hold the cube at 2.

Position 1
Position 1

The pip-count (race) is also strongly in my favor having only 64 pips against Ole’s 81 pips.

It was not at all clear to me what the right cube action was. Let’s try to dig into a deeper analysis of the situation I was faced with.

Match score

When leading a match you want to protect your lead by lowering the variance which in most cases means playing more conservatively with the cube – especially high cubes. More variance will benefit the person trailing because it will be easier to become lucky and turn around the course of the match.

This means that I should be less inclined to double than in money game where match score does not play a role.

Context – no gammon, no prime to jump

Let’s start by looking at another situation, that could arise from our original position, where the cube action is clear.

Position 2
Position 2

White is not behind the prime and there is no way of catching the second checker. White will lose a few games by being hit from the bar but Black will also lose a few gammons. See the theory section for reference positions.

Conclusion 1: It is clear that White has no incentive to play on for the gammon in position 2 and the correct cube action is for White to double and Black to pass both at the current match score and in money game.

Theory

I have two reference positions that are nice to know.

Reference position 1
Reference position 1: White wins approximately 2% gammon

Reference position 2
Reference position 2: White wins approximately 5% gammon

Context – gammon, no prime to jump

Let’s take a look at a position where we can win more gammons but we don’t have the issue of being behind a prime.

Position 3
Position 3

It is clear that we are in a much better position than in position 2. The question is if we should double now or play on for the gammon.

If White catch a second checker there will be a significant amount of gammon in the position. If not, White will jump out and end up in one of two situations.

  1. If Black fans we have the same scenario as dealt with in “Context – no gammon, no prime to jump.”
  2. If Blacks enters the position it will be a clear double for White and depending on the distance between the checkers Black has either a take or pass. See the theory section for reference position.

In scenario two Black’s cube action is as shown below.

Position 4
Position 4: Where White’s checker has to be for Black to take or pass

In either case our cube decision is well defined if we do not double right away.

Conclusion 2: If we hit we play on for the gammon. If we miss we double and let Black decide what to do. By not doubling right away White gets a chance to try to win a gammon.

A good rule of thumb is that you need to win more gammons than you lose games to be able to play on for a gammon (the position is too good to double).

As seen in reference position 4 White will win less than 40% gammon if White hits the second checker (let’s say ~30% in position 3). It will happen ~1/3 of the time giving White close to 10% gammon chances. Another estimate is that white will win position 3 over 90% of the time.

Conclusion 3: The estimates above indicate that White will win ~10% gammon and lose fewer games than that. If White does not double there will be efficient cube action next roll. White should play on for the gammon in position 3. The decision is close though.

Theory

Rule of thumb: If you are in a position where you win if you run by your opponent or lose if you get hit, then

  • you can double if you are 11 away
  • your opponent should take if you are 9 or more away
  • your opponent should take or pass if you are 8 away depending on other factors

Reference position 3
Reference position 3: Black can either take or pass

If you have closed out two checkers you will win ~40% gammon.

Reference position 4
Reference position 4: White will win approximately 40% gammon

Analysis of the original position

Looking at position 1 it is now clear that we are not thinking about playing on for the gammon. If position 3 was just good enough to play on for a gammon then position 1 is definitely not good enough. The question is then if White can double at all.

The fact that our position is not a lot worse than in position 3 it still has to be a clear double and a huge pass for Black. This conclusion is so strong that the match related adjustments does not change anything.

After thoughts

If you made it this far consider leaving a comment with feedback on the style of this blog. If I can make improvements I’d like to hear about it, so I spend the time writing these blogs as efficiently as possible.

Good luck with your training. See you at the tables.