Although very close, this position worths our attention as the "cubelessly" best choice 21/9(2) becomes second due to the higher gammon percentage of the other candidate 21/15(2), 10/4(2). Most probably next turn it will be too good/pass (or double/pass at worst) so it worths (for white) going for the gammon with the relative safety of black's future pass.

Tanaka and Michi, having not only a big racing lead but also an good distribution, should have played on, going for the gammon. As mentioned earlier in the vast majority of the next sequences it will be a pass for black and white have some gammonish jokers next turn (e.g. 66,55,33).

Against a weak player doubling could be correct as he could possibly make a huge blunder by taking the cube.

Kostas: "We were thinking of doubling here as they had a big advantage, however we didn't see any market loosers so we decided to wait for a roll. On the other hand I believe that there are many players that would pass this cube"

Excellent cube here from Haris and Kostas. Take is clear as 2point anchors are very strong (Kleitsas, once, took a much "harder" cube by just owning the 23pt and was validated by XG at my suprise) , provided that there is enough "timming" which can be improved in that case.

This is the play that most people would choose (including myself) but blacks have a strong board with many hitting numbers and any hit would make it easier for them to run their checkers home. Bar/21, 13/7 would be correct if whites had 2 spares on their 13pt but now 8/2 (although a bit ugly) is best. One benefit from staying to the 21pt (which is rather safe) is that next turn whites may advance their anchor to the 21pt and better control the outfield, making it difficult for blacks to break their 15pt and run their checkers home.

Now playing all the way (Bar/18) is better as it is the play that gives white some chances to make an advanced anchor next turn, or possibly improve their timming.

Tanaka and Michi played the correct 13/7 which is superior to 8/2 as it creates better structure for making the critical 5 and 4 points next turn(s). Bar/17 now is clearly wrong as it increases blacks' gammons by a lot, does not improve whites' timming (which is already good). Finally keeping a spare checker on the 23 point could prove be very helpful in the future on hitting a long shot without having to break the anchor.

Looks like a "Pay now or Pay later" problem. Charis and Kostas have seen the very clever play of bringing two checkers down (13/9, 13/7). This way they cleared the "liability" of the 13pt with minimal risk. White can only hit with 6 rolls next turn, while having only an 1point board and a blot on their own 7 point. Blacks will have many gammonish return shots if being hit, especially if whites hit with 61 or 62.

Blacks are so much ahead in the race that blocking whites' sixes can only harm themselves. XG says that they should play the equally safe 11/9, 10/9 which minimizes future contact.
Kostas: "We played it automatically, not even thinked about the other option"

The difference betweem the 5/1 4/off(3) and the alternative of 5/1(4) is huge. Three more checkers off make a big difference impacting both gammons and single wins. In addition, by this play blacks will be much less likely to leave a blot in the next turn.

Kostas: "Before the roll Haris said that if whites roll a 44 again, then he and Kostas would not have anything to worry about. Whites did roll 44 (going down to 6,7% winning rate) so it will be interesting to see what comes next.

Winning the game, although possible, is not top priority for whites here. Minimizing gammon losses is more important and 6/3 better serves that goal.

After some unfortunate rolls from Charis&Kostas, and some good from Tanaka&-Michi, the latter pair now has as top priority to win the game and not save the gammon. 12/11 7/3 not only saves a crossover in the critical race that will follow, but also keeps a checker back on the 11 pt so as to provide a better coverage of blacks' blot, in case they enter next turn.

Now whites are huge favorite to win the game and correctly redouble to 4. If blacks enter (except with 55 or 66) next turn whites will still be the favorite as they will have less crossovers to bear off. If blacks dance, then whites will have lost they market by a lot. Another factor also favoring doubling is the fact that blacks' take is not clear and blacks will probably have to make some kind of calculations which will cost them time and energy or they may just make a mistake.

It is black that has to think of the take and it is a difficult task to do so. According to a classic reference position (also included in Backgammon Encyclopedia of Kit Woolsey) where blacks have 5 checkers off, one checker on the bar and their remaining checkers well placed (e.g. on the 2 and 1 point) and whites have a perfect full board (The three spare checkers on his 6,5 and 4 points) then blacks have around 22% winning chances which give them a marginal take in money games. Here, blacks' spare checkers are not well placed (blacks will probably have some costly wastage during bear off). On the other hand although whites have one checker off, blacks have 6 mens off, which favours blacks compared to the reference position (one white checker off in a full board with excellent distribution would not make a big difference for whites). I would estimate that those two differences sum up to some kind of an improvement for whites (lets assume that it becomes from 22% of the reference position a 20% wins for blacks).

Of course we have not taken into account the already open 6 and 5 points of whites which form a big change, as blacks will be favorite to enter next turn. I think it is obvious that this last difference (although not easy to calculate directly) is more than enough to make it a clear take for money games let alone at this score.

Correct decision by both sides on a difficult position in a clocked match played at around 4 o'clock in the morning after a series of other long matches.

Kostas: "It took them 4 minutes in a clocked match to think about the redouble, so I didn't think it could be a bluff, despite our gaps in our 2 and 4 points. In addition, we noticed that if their next roll had an 1 or 2, they would not take 2 off and that would cost them a lot as the would not improve their crossover number. Haris was not sure if it is a take or not, but decided to gamble and thus we took the cube."

p.s. Some very strong players often decide very fast in similar positions, as they have a very strong "feel" of such positions which come from experience (e.g. performed similar calculations many time in the past, have similar reference positions, or just have much experience in similar positions)

As stated in the previous comment whites would have lost their market by a lot, had they not doubled.

Blacks are trailing 7-away 2-away and they need to be alert for gammonish plays and for turning the cube earlier than normal (if they have many gammon wins). According to XG's database 13/9, 13/8 is just slightly better than the standard split.

Similarly, whites should avoid gammonish positions and the running play is clearly best.

Running didn't work for white's and now they have to play the real game! Hitting with 6/5 is clear.