$20,000 ICUC - a report on wins

[Ozymandias]

who says the Marshall is a draw?

Hundreds of thousands of engine games, for example.

My first thought about your point with ironman was, that it probably shuffles up "only" the bottom part of the rankings. But I'm the one who should have known it better, because these results possibly have cost me some thousands of dollars, leaving me on the 3rd place only by a tie-break-system, that is influenced by any game played in the tournament. But the fact that my first-round result against a player that quits the tour immediatly afterwards has been counted for my SB-Rating with 0 points (despite my veto) would have been a knock-out for me in any case. At least one future thing became crystal clear: Nowadays tournaments become such close, that we need to pay much more attention to having reasonable (and FIDE-conform) Tie-Breakers, squeezing out as much randomness as possible.

Leaving out as much randomness as possible, out of the equation, should certainly be one of the two main concerns, towards any future (serious) Freestyle event; the other one being how to tackle the high draw rate problem. Still waiting to hear from Arno.

indeed, what I am seeing is at least 90% of Marshall lines are won for white(after Re5 Bd6 Re1 Qh4 g3 Qh3 Be3).

If you mean 11.Rxe5 Bd6, you're just hanging the Knight on d5.

[Ozymandias]

indeed, what I am seeing is at least 90% of Marshall lines are won for white(after Re5 Bd6 Re1 Qh4 g3 Qh3 Be3).

I just updated my DB for the 27th Engine Masters, and looking at the playchess and IC games, from the past year (more that half a million) I still see no change in the trend: The Marshall Attack is the preferred draw weapon of choice, for black.

1.e4 is still played 77% of the time, but it isn't met by anything other than e5, 3 times out of 4
2. Nf3 Nc6 are ubiquitous, but then comes
3. Bc4 (79%),instead of the classical Bb5 (which has fallen to just above a 20% popularity), and not because of the Berlin as some have suggested, (it has a 20% support and good draw rates), but a6 clearly dominates here with a 79% probability of play. Then of course, comes the line
4. Ba4 Nf6
5. O-O (actually d3 already attracts 23% of the players, for all the good it will do) Be7
6. Re1 b5
7. Bb3 O-O and now white just doesn't know how to proceed
8. c3 maintains the fidelity of 70% of the players, but only because none of the anti-Marshall schemes, has proven successful, and once we have d5 on the board, (less than 5% of the time is anything else tried), you have a 91% draw rate. Maybe you also misread that number and thought it was whites's score? The rest is almost forced
9. exd5 Nxd5
10. Nxe5 Nxe5
11. Rxe5 c6
12. d4 (d3 is also played nowadays regularly, this isn't a move you can just skip and assume d4 will be played by white) Bd6
13. Re1 Qh4
14. g3 Qh3 and the move you give as a refutation, though still the main line, it's only used 68% of the time, with a 91% draw rate and 47.9% white score:
15. Be3 leaves white in a position that's far from desirable

Maybe your'e talking about private games that have a refutation much further down the line, but that's not what you said at all. The games made public still hold the Marshall as the culprit against e4, the data is there.

[Lyudmil Tsvetkov]

indeed, what I am seeing is at least 90% of Marshall lines are won for white(after Re5 Bd6 Re1 Qh4 g3 Qh3 Be3).

I just updated my DB for the 27th Engine Masters, and looking at the playchess and IC games, from the past year (more that half a million) I still see no change in the trend: The Marshall Attack is the preferred draw weapon of choice, for black.

1.e4 is still played 77% of the time, but it isn't met by anything other than e5, 3 times out of 4
2. Nf3 Nc6 are ubiquitous, but then comes
3. Bc4 (79%),instead of the classical Bb5 (which has fallen to just above a 20% popularity), and not because of the Berlin as some have suggested, (it has a 20% support and good draw rates), but a6 clearly dominates here with a 79% probability of play. Then of course, comes the line
4. Ba4 Nf6
5. O-O (actually d3 already attracts 23% of the players, for all the good it will do) Be7
6. Re1 b5
7. Bb3 O-O and now white just doesn't know how to proceed
8. c3 maintains the fidelity of 70% of the players, but only because none of the anti-Marshall schemes, has proven successful, and once we have d5 on the board, (less than 5% of the time is anything else tried), you have a 91% draw rate. Maybe you also misread that number and thought it was whites's score? The rest is almost forced
9. exd5 Nxd5
10. Nxe5 Nxe5
11. Rxe5 c6
12. d4 (d3 is also played nowadays regularly, this isn't a move you can just skip and assume d4 will be played by white) Bd6
13. Re1 Qh4
14. g3 Qh3 and the move you give as a refutation, though still the main line, it's only used 68% of the time, with a 91% draw rate and 47.9% white score:
15. Be3 leaves white in a position that's far from desirable

Maybe your'e talking about private games that have a refutation much further down the line, but that's not what you said at all. The games made public still hold the Marshall as the culprit against e4, the data is there.

could you please post the final position, very pressed for time now, yes, this is the main line, and it is almost certainly won for white, or at least the vast majority of subvariations.

SF reaches scores over 100cps, true much later, for almost all of them.

stats are usually meaningful, but sometimes not.

what if those players have been following the same or similar line, due to its high exposure/being in vogue?

you know, many lines have been considered sound for quite some time, and then suddenly a new unexpected move changes everything.

[Ozymandias]

could you please post the final position, very pressed for time now, yes, this is the main line, and it is almost certainly won for white, or at least the vast majority of subvariations.

SF reaches scores over 100cps, true much later, for almost all of them.

stats are usually meaningful, but sometimes not.

what if those players have been following the same or similar line, due to its high exposure/being in vogue?

you know, many lines have been considered sound for quite some time, and then suddenly a new unexpected move changes everything.

[pgn][Event "?"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]

1. e4 e5 2. Nf3 Nc6 3. Bb5 a6 4. Ba4 Nf6 5. O-O Be7 6. Re1 b5 7. Bb3 O-O
8. c3 d5 9. exd5 Nxd5 10. Nxe5 Nxe5 11. Rxe5 c6 12. d4 Bd6 13. Re1 Qh4 14.
g3 Qh3 15. Be3 *[/pgn]
I think I see where the dilemma resides. You're talking about evaluation, while I talk about statistics. I had a similar exchange with Thomas in private, and it simply cannot be resolved.

Just as an example, I reached a 6-men draw a few minutes ago, in an IC tour, and my opponent didn't have 6-men TBs, so his SF gave an eval of +1.46. If this happens just one piece shy of the TBs at hand, can you imagine how off they can be, with more pieces on the board?

[Lyudmil Tsvetkov]

could you please post the final position, very pressed for time now, yes, this is the main line, and it is almost certainly won for white, or at least the vast majority of subvariations.

SF reaches scores over 100cps, true much later, for almost all of them.

stats are usually meaningful, but sometimes not.

what if those players have been following the same or similar line, due to its high exposure/being in vogue?

you know, many lines have been considered sound for quite some time, and then suddenly a new unexpected move changes everything.

[pgn][Event "?"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]

1. e4 e5 2. Nf3 Nc6 3. Bb5 a6 4. Ba4 Nf6 5. O-O Be7 6. Re1 b5 7. Bb3 O-O
8. c3 d5 9. exd5 Nxd5 10. Nxe5 Nxe5 11. Rxe5 c6 12. d4 Bd6 13. Re1 Qh4 14.
g3 Qh3 15. Be3 *[/pgn]
I think I see where the dilemma resides. You're talking about evaluation, while I talk about statistics. I had a similar exchange with Thomas in private, and it simply cannot be resolved.

Just as an example, I reached a 6-men draw a few minutes ago, in an IC tour, and my opponent didn't have 6-men TBs, so his SF gave an eval of +1.46. If this happens just one piece shy of the TBs at hand, can you imagine how off they can be, with more pieces on the board?

I am following the game simultaneously with watching SF scores, and mostly they seem 100cps convertible positions.

that is prone to errors too, of course, everyone and everything prunes some moves and evaluates at times below par, so mistakes are unavoidable.

but what I am seeing is mostly white wins.

[Ozymandias]

what I am seeing is mostly white wins.

I honestly can't wait, for you to have the time to share with us some details, about those games you mention. Things like TC, CPU, number of threads, syzygy probe depth and limit, starting position... in the meantime, what we saw, was the 27th EMT, a tournament that ended, for the first time (IIRC), with a 90%+ draw rate. And those are unfiltered numbers, I'm not taking into account the couple of players who were happy to lose games, or the many book loses that took place, throughout the event. I will only mention a couple of games from the guy who "won" the event.

I put an emphasis on the term "win", because the definition is really peculiar: "to get possession of by effort or fortune". I'm on the side of effort, while Arno considers fortune to be enough, as his words reflect in this small exchange:



Round 1
[pgn][Event "27th Engine Masters, Sunday August 27, 13 Rounds, 12m+1s, 16.00 ST"]
[Site "Infinity Chess"]
[Date "2017.8.27"]
[Round "1"]
[White "Soloman"]
[Black "KhalidOmar"]
[Result "1-0"]
[ECO "C77"]
[WhiteElo "2494"]
[BlackElo "2426"]
[WhiteEngine "stockfish_17082014_x64_modern"]
[BlackEngine "SF Eman 0.9 x64 BW-CTG"]
[EventType "Swiss"]
[WhiteTeam ""]
[BlackTeam ""]
[PlyCount "205"]
[TimeControl "720+1"]

1.e4 {[%emt 0][%B]} e5 { [%eval -0.05/1][%emt 3]} 2.Nf3 {[%emt 0][%B]} Nc6 { [%eval -0.05/1][%emt 0]} 3.Bb5 {[%emt 0][%B]} a6 { [%eval -0.05/1][%emt 0]} 4.Ba4 {[%emt 0][%B]} Nf6 { [%eval -0.05/1][%emt 0]} 5.d3 {[%emt 0][%B]} Bc5 { [%eval -0.05/1][%emt 0]} 6.c3 {[%emt 0][%B]} b5 { [%eval -0.05/1][%emt 0]} 7.Bb3 {[%emt 0][%B]} d6 { [%eval -0.05/1][%emt 0]} 8.Bg5 {[%emt 0][%B]} h6 { [%eval -0.05/1][%emt 0]} 9.Bh4 {[%emt 0][%B]} Bb6 { [%eval -0.05/1][%emt 0]} 10.a4 {[%emt 0][%B]} g5 { [%eval -0.05/1][%emt 1]} 11.Bg3 {[%emt 0][%B]} b4 { [%eval -0.05/1][%emt 0]} 12.Nbd2 {[%emt 0][%B]} Rb8 { [%eval -0.05/1][%emt 0]} 13.O-O {[%emt 0][%B]} O-O { [%eval -0.05/1][%emt 0]} 14.h3 {[%emt 0][%B]} Be6 { [%eval -0.05/1][%emt 0]} 15.Bc4 { [%eval 0.34/28][%emt 26]} Qc8 { [%eval -0.05/1][%emt 0]} 16.Re1 { [%eval 0.40/30][%emt 42]} Re8 { [%eval -0.05/1][%emt 0]} 17.Rc1 { [%eval 0.40/29][%emt 28]}[/pgn]

This wasn't a blatant book loss, but the out of book position gave white a clear advantage, that just kept climbing. Statistics aren't definitive, but a 62.5% for withe, coupled with a 0.40 evaluation by Soloman's SF at move 17 (depth 29), are enough to figure out when the game was lost.

And now for the wrapped gift. Round 4
[pgn][Event "27th Engine Masters, Sunday August 27, 13 Rounds, 12m+1s, 16.00 ST"]
[Site "Infinity Chess"]
[Date "2017.8.27"]
[Round "4"]
[White "Soloman"]
[Black "Paul"]
[Result "1-0"]
[ECO "C77"]
[WhiteElo "2507"]
[BlackElo "2522"]
[WhiteEngine "stockfish_17082014_x64_modern"]
[BlackEngine "stockfish_17082609_x64_modern"]
[EventType "Swiss"]
[WhiteTeam ""]
[BlackTeam ""]
[PlyCount "167"]
[TimeControl "720+1"]

1.e4 {[%emt 0][%B]} e5 {[%emt 1][%B]} 2.Nf3 {[%emt 1][%B]} Nc6 {[%emt 0][%B]} 3.Bb5 {[%emt 0][%B]} a6 {[%emt 0][%B]} 4.Ba4 {[%emt 0][%B]} Nf6 {[%emt 1][%B]} 5.d3 {[%emt 1][%B]} Bc5 {[%emt 0][%B]} 6.c3 {[%emt 0][%B]} b5 {[%emt 0][%B]} 7.Bb3 {[%emt 0][%B]} d6 {[%emt 1][%B]} 8.Bg5 {[%emt 1][%B]} h6 {[%emt 1][%B]} 9.Bh4 {[%emt 0][%B]} O-O {[%emt 0][%B]} 10.O-O {[%emt 0][%B]} Bb6 {[%emt 0][%B]} 11.a4 {[%emt 1][%B]} g5 {[%emt 1][%B]} 12.Nxg5 {[%emt 1][%B]} hxg5 {[%emt 0][%B]} 13.Bxg5 {[%emt 0][%B]} Na5 {[%emt 0][%B]} 14.Bc2 {[%emt 0][%B]}[/pgn]

Paul's game is almost a joke. Black exits the book at move 13, leaving white's book move with almost an 82% score in favour. This is one of those cases, were stats clearly crunch engine evaluation. Paul's SF was still giving a 0.00 score at move 18, depth 30. 3 moves later, it was admiring defeat with a much more realistic 1.55 for white, at depth 29.

The situation became so surreal, that at a point, some of the higher ranked players, still without a single "win" in their pockets, started asking for friends to give them the full point. Whether they were being serious or not, that's something you can always hide behind a smiley, but the bottom line is, what credibility do these events hold anymore? Let's not forget that Soloman was tied first in the ICUC, despite a lackluster performance (17th place!).

At one point, people will start to ask themselves just how much luck can simply be luck, and not something else entirely.

[jefk]

12. d4 (d3 is also played nowadays regularly, this isn't a move you can just skip and assume d4 will be played by white)

agree that 12.d4 can be equalized by Black, Tsvetkov probably
is running behind most GM theory with his SF analysis.
But 12.d3! still is a strong move, better White chances i think
then when going against a Berlin Queen's exchange, because
usually you go against the Berlin it's bishop pair (Black) but
with double pawn handicap.
With the Marshall Black is a whole pawn behind, but often
can gets the bishop pair against White in the endgame.

Very hard to win for White, i agree, and thats why i
prefer 1.d4 instead of 1.e4 again, with some new analysis
ofcourse trying to maintain an advantage, which i will
not reveil here right now; look later at my correspondence
games i would suggest

[Lyudmil Tsvetkov]

what I am seeing is mostly white wins.

I honestly can't wait, for you to have the time to share with us some details, about those games you mention. Things like TC, CPU, number of threads, syzygy probe depth and limit, starting position... in the meantime, what we saw, was the 27th EMT, a tournament that ended, for the first time (IIRC), with a 90%+ draw rate. And those are unfiltered numbers, I'm not taking into account the couple of players who were happy to lose games, or the many book loses that took place, throughout the event. I will only mention a couple of games from the guy who "won" the event.

I put an emphasis on the term "win", because the definition is really peculiar: "to get possession of by effort or fortune". I'm on the side of effort, while Arno considers fortune to be enough, as his words reflect in this small exchange:



Round 1
[pgn][Event "27th Engine Masters, Sunday August 27, 13 Rounds, 12m+1s, 16.00 ST"]
[Site "Infinity Chess"]
[Date "2017.8.27"]
[Round "1"]
[White "Soloman"]
[Black "KhalidOmar"]
[Result "1-0"]
[ECO "C77"]
[WhiteElo "2494"]
[BlackElo "2426"]
[WhiteEngine "stockfish_17082014_x64_modern"]
[BlackEngine "SF Eman 0.9 x64 BW-CTG"]
[EventType "Swiss"]
[WhiteTeam ""]
[BlackTeam ""]
[PlyCount "205"]
[TimeControl "720+1"]

1.e4 {[%emt 0][%B]} e5 { [%eval -0.05/1][%emt 3]} 2.Nf3 {[%emt 0][%B]} Nc6 { [%eval -0.05/1][%emt 0]} 3.Bb5 {[%emt 0][%B]} a6 { [%eval -0.05/1][%emt 0]} 4.Ba4 {[%emt 0][%B]} Nf6 { [%eval -0.05/1][%emt 0]} 5.d3 {[%emt 0][%B]} Bc5 { [%eval -0.05/1][%emt 0]} 6.c3 {[%emt 0][%B]} b5 { [%eval -0.05/1][%emt 0]} 7.Bb3 {[%emt 0][%B]} d6 { [%eval -0.05/1][%emt 0]} 8.Bg5 {[%emt 0][%B]} h6 { [%eval -0.05/1][%emt 0]} 9.Bh4 {[%emt 0][%B]} Bb6 { [%eval -0.05/1][%emt 0]} 10.a4 {[%emt 0][%B]} g5 { [%eval -0.05/1][%emt 1]} 11.Bg3 {[%emt 0][%B]} b4 { [%eval -0.05/1][%emt 0]} 12.Nbd2 {[%emt 0][%B]} Rb8 { [%eval -0.05/1][%emt 0]} 13.O-O {[%emt 0][%B]} O-O { [%eval -0.05/1][%emt 0]} 14.h3 {[%emt 0][%B]} Be6 { [%eval -0.05/1][%emt 0]} 15.Bc4 { [%eval 0.34/28][%emt 26]} Qc8 { [%eval -0.05/1][%emt 0]} 16.Re1 { [%eval 0.40/30][%emt 42]} Re8 { [%eval -0.05/1][%emt 0]} 17.Rc1 { [%eval 0.40/29][%emt 28]}[/pgn]

This wasn't a blatant book loss, but the out of book position gave white a clear advantage, that just kept climbing. Statistics aren't definitive, but a 62.5% for withe, coupled with a 0.40 evaluation by Soloman's SF at move 17 (depth 29), are enough to figure out when the game was lost.

And now for the wrapped gift. Round 4
[pgn][Event "27th Engine Masters, Sunday August 27, 13 Rounds, 12m+1s, 16.00 ST"]
[Site "Infinity Chess"]
[Date "2017.8.27"]
[Round "4"]
[White "Soloman"]
[Black "Paul"]
[Result "1-0"]
[ECO "C77"]
[WhiteElo "2507"]
[BlackElo "2522"]
[WhiteEngine "stockfish_17082014_x64_modern"]
[BlackEngine "stockfish_17082609_x64_modern"]
[EventType "Swiss"]
[WhiteTeam ""]
[BlackTeam ""]
[PlyCount "167"]
[TimeControl "720+1"]

1.e4 {[%emt 0][%B]} e5 {[%emt 1][%B]} 2.Nf3 {[%emt 1][%B]} Nc6 {[%emt 0][%B]} 3.Bb5 {[%emt 0][%B]} a6 {[%emt 0][%B]} 4.Ba4 {[%emt 0][%B]} Nf6 {[%emt 1][%B]} 5.d3 {[%emt 1][%B]} Bc5 {[%emt 0][%B]} 6.c3 {[%emt 0][%B]} b5 {[%emt 0][%B]} 7.Bb3 {[%emt 0][%B]} d6 {[%emt 1][%B]} 8.Bg5 {[%emt 1][%B]} h6 {[%emt 1][%B]} 9.Bh4 {[%emt 0][%B]} O-O {[%emt 0][%B]} 10.O-O {[%emt 0][%B]} Bb6 {[%emt 0][%B]} 11.a4 {[%emt 1][%B]} g5 {[%emt 1][%B]} 12.Nxg5 {[%emt 1][%B]} hxg5 {[%emt 0][%B]} 13.Bxg5 {[%emt 0][%B]} Na5 {[%emt 0][%B]} 14.Bc2 {[%emt 0][%B]}[/pgn]

Paul's game is almost a joke. Black exits the book at move 13, leaving white's book move with almost an 82% score in favour. This is one of those cases, were stats clearly crunch engine evaluation. Paul's SF was still giving a 0.00 score at move 18, depth 30. 3 moves later, it was admiring defeat with a much more realistic 1.55 for white, at depth 29.

The situation became so surreal, that at a point, some of the higher ranked players, still without a single "win" in their pockets, started asking for friends to give them the full point. Whether they were being serious or not, that's something you can always hide behind a smiley, but the bottom line is, what credibility do these events hold anymore? Let's not forget that Soloman was tied first in the ICUC, despite a lackluster performance (17th place!).

At one point, people will start to ask themselves just how much luck can simply be luck, and not something else entirely.

well, g5 in both games is simply very bad; black king remains unsheltered, f5 square difficult to defend, etc.

regarding the Marschall, I can not share anything, as there are endless lines to analyse, but one thing is certain: there is not a single sound black gambit, and the Marschall is a gambit.

after Be3, black has no satisfactory reply, one way or another.

interesting, how those players manage to draw it?

[Lyudmil Tsvetkov]

12. d4 (d3 is also played nowadays regularly, this isn't a move you can just skip and assume d4 will be played by white)

agree that 12.d4 can be equalized by Black, Tsvetkov probably
is running behind most GM theory with his SF analysis.
But 12.d3! still is a strong move, better White chances i think
then when going against a Berlin Queen's exchange, because
usually you go against the Berlin it's bishop pair (Black) but
with double pawn handicap.
With the Marshall Black is a whole pawn behind, but often
can gets the bishop pair against White in the endgame.

Very hard to win for White, i agree, and thats why i
prefer 1.d4 instead of 1.e4 again, with some new analysis
ofcourse trying to maintain an advantage, which i will
not reveil here right now; look later at my correspondence
games i would suggest

in the Ruy Lopez, best black defense by far is 3...f5, the Schliemann, black fully equalises on move 15 or so, 3...a6 and 3...Nf6, the Berlin, are much inferior.

time to update your database.

[jefk]

regarding the Marschall, I can not share anything, as there are endless lines to analyse, but one thing is certain: there is not a single sound black gambit, and the Marschall is a gambit.
after Be3, black has no satisfactory reply, one way or another.

simply with 15...Be6! and so on =

'no sound Black gambit' ? hmm well i have to think about that
but indeed i prefer gambits with White only indeed;
yet your statement is in disagreement with your
remark that 3...f5 the Schliemann/Jaenisch gambit in the RL
is the best Black defense. It's a *gambit* and although it's
sharp/double-edged i don't think it's good in computer chess
(for Black); if White has a good book