At what rating does the Evans gambit fall off?

After losing to Spassky in the King's Gambit in 1960, Fischer wrote an article "A Bust to the King's Gambit" in which he claims that the King's Gambit is busted and loses by force.  If you plug in his analysis with modern engines, he evaluates the positions out of the opening as roughly equal (a far cry from "losing by force").  The irony is that a few years later he played a few games in the King's Gambit from the White side (using the Bc4 move order) and won all 3.  It would seem that even he did not believe his own claim that the KG was refuted.

That is not the argument.  The points being made are the following:  1) it is not dubious (meaning it does not put White in a worse position - as the position even after the pawn sacrifice is roughly equal), and 2) that it has been played even at the very top level (which goes to the OP's question).  Just like with the KG, Black can hold with proper play - which is the same with any sound opening.  No one would claim the Ruy Lopez is unsound, and with proper play from both sides it is a draw.  The fact that the Evan's Gambit gives up material (gaining space, time, and central control in return) and leads to a draw with proper play indicates that the opening is not unsound.

If the goal of the opening is to get to a playable middlegame, the Evan's is no better and no worse than any other opening.  You get to a position where White has good center control, more space, and easy development for the cost of a pawn.  If your goal in the opening is to squeeze out every advantage you can, then sure, most gambits are not going to work for you.

I agree with the last statement (not making it your main weapon), but it does not hurt to have it in your arsenal.  At the US Open several years ago, there were a LOT of Evan's Gambits being played in the 1200-2200 rating ranges, with many decisive results.  In fact, it was probably the second most played opening that tournament (the London sadly being the most popular - at one point when I was walking around, there were 2 whole tables of boards where every single game was the London).

Agreed.  However, if the assertion is that you cannot play the Evan's Gambit above a certain level because it becomes dubious (which is what the question was implying), that is incorrect.  It can be a lot of fun to play (at any level) and is one of the more sound gambits available - far more so than things like the Grob, Blackmar-Diemar Gambit, and Danish Gambit, just to name a few.

"If the goal of the opening is to get to a playable middlegame, the Evan's is no better and no worse than any other opening.  You get to a position where White has good center control, more space, and easy development for the cost of a pawn."  

No, you don't get a better center control, you don't get more space, and black gets at least as easy a development as white's in the Lasker line - which is precisely the reason I don't suggest playing the Evans' competitively.

"If your goal in the opening is to squeeze out every advantage you can, then sure, most gambits are not going to work for you."

I apologize for being the bearer of the sad news, but that's how consistent results are achieved at the master level.

"However, if the assertion is that you cannot play the Evan's Gambit above a certain level because it becomes dubious (which is what the question was implying), that is incorrect.  It can be a lot of fun to play (at any level) and is one of the more sound gambits available - far more so than things like the Grob, Blackmar-Diemar Gambit, and Danish Gambit, just to name a few."

I never said the Evans' is plain bad. On the contrary- I clearly stated that it can be a really fun opening. I myself have played quite many games with mixed results - but the games were always very entertaining.

White doesn't have more space there?  Really?  Perhaps you were taught to count space differently than I was. Using the Sierawan method, White pressures 7 squares on Black's side of the board, Black pressures 5 on White's.  Black has a pawn and has his dark-squared bishop outside the pawn chain (in effect, giving him a better version of a Philidor defense).

I guess someone should tell Magnus that.  Perhaps he would be more consistent if he strived for an opening advantage every time.

In all seriousness, striving for an opening edge is one way to play, it is not the only way.  Andrew Tang got his GM title playing the London the majority of the time as White.  There is no press for an opening edge there - just developing the pieces to get to a playable middlegame.  There are many other players who use equalish lines because they know them better than most people and are most comfortable in them.  The Evan's is another line that does the same thing.

Perhaps I inferred from your comment about "100% soundness" that you were implying if an opening isn't 100% sound you shouldn't use it as a main weapon.  Just pulling up my master database, and limiting it to 2010-2018 (I need to update it for the last few years), I count 279 master-level games (where both players were 2300+) in the Evan's with White having a +99=93-87 score. (35.5%, 31.2%, and 33.3% respectively).  In that time window, there were a grand total of 0 games played in the Lasker variation.  So, apparently masters simply haven't gotten the memo?

I agree with you that the Evan's should not be a main weapon.  I disagree with the assertion (implied by the OP) that it is dubious at a certain level.  The main problem you would have at the master level if it were your main weapon is you are too easy to prepare for, but that is the case for many non-dubious openings.

#25

"White doesn't have more space there?" ++ White has more space, but white is a pawn down. White cannot convert the space, but black can convert the pawn.

"striving for an opening edge is one way to play"
++ Chess is a draw. Objectively you cannot strive for an edge. However gambits strive to lose. You either overrun the unprepared opponent, or you end up in a lost endgame.

"Andrew Tang got his GM title playing the London the majority of the time as White."
++ The London does not lose a pawn.

"developing the pieces to get to a playable middlegame" ++ Yes, but a pawn down.

"There are many other players who use equalish lines because they know them better than most people and are most comfortable in them."
++ The Evans Gambit does not do that, it loses a pawn. 'A pawn is a pawn' - Fischer

"it can be a really fun opening" ++ That is for sure. 'It still makes for interesting chess' - Fischer.

"the games were always very entertaining." ++ Yes indeed.

"if an opening isn't 100% sound you shouldn't use it as a main weapon"
++ Would you play it if the stakes were $ 1 million?

"In that time window, there were a grand total of 0 games played in the Lasker variation."
++ That is because white avoids the Lasker Variation with 6 d4 d6 7 Qb3. White got the memo.

"I disagree with the assertion (implied by the OP) that it is dubious at a certain level."
++ It is dubious in correspondence and at top GM level. That is the expert opinion of Kasparov.

"if it were your main weapon is you are too easy to prepare for, but that is the case for many non-dubious openings." ++ In the non-dubious openings say Giuoco Pianissimo you get an even position, in the Evans Gambit you end up a pawn down with no compensation.

#27
Yes, Kasparov beat Anand and Piket with it, but that is mainly because of the surprise value.
They were obviously prepared for a Ruy Lopez and it is not easy to defend the Evans Gambit unprepared against a prepared opponent. Kasparov himself called the Evans Gambit dubious in an interview. So the Evans Gambit falls off at a rating of 2500.

Static vs Dynamic advantages.  They balance out, hence the reason the engine (both Stockfish and Leela) evaluate it to be roughly equal - in short, White has compensation for the pawn.

That is a gross over-simplification.  There are static advantages (structure, material) and there dynamic advantages (piece activity, space).  Often you trade one for the other.  In many gambits, you attempt to trade material for development.  The Evan's trades material for space.  Saying "you cannot convert your dynamic advantage but I can convert my static one" is asinine.  The whole point is the fight over which is more important.

Chess is a draw with best play.  Humans are not perfect, however.  Many sound openings for White attempt to create some imbalance that makes it difficult for Black to navigate (aka "fighting for an opening advantage").  Many others just try to get a playable position. 

The point was the London doesn't try to fight for an opening advantage, and despite giving up a pawn, both openings are roughly equal.

Oh, it doesn't?  So, the equal evaluation is a mirage?  The roughly equal results at the highest level is also a mirage?

A non-sequitur.  There are a lot of things I wouldn't play if $1M was on the line, including things like the London, Colle, Torre, Reti, KIA, or the Italian in general (assuming I needed a win).  If all I needed was at least a draw, some of those come back into play.

In other words, Black cannot just know the Lasker Variation to refute the Evan's, which was the assertion.

Correspondence is a whole different animal.  Centaur chess is not the same.  So, in that regard, I agree somewhat.  @pfren would be a better source to discuss that as he plays those types of games.

At the top level, it is not going to be any Super-GM's main weapon in classical chess.  However, almost all of them have it in their repertoire and bring it out (even in classical games) from time to time.  Interestingly enough, the only player I cannot find a single game in the database in the Evan's is Magnus (he's played the Black side several times, but I cannot find a single game where he played White).  But virtually all of the rest of the top 20 have games (some classical, some rapid/blitz) in the Evan's (as White), as do Kasparov, Fischer, Tal, Timman, Sveshnikov, etc.  If they can bring it out, even as a surprise weapon, against their contemporaries, certainly players significantly weaker playing others who are significantly weaker can use it.

Leela and Stockfish disagree with your assertion that you get no compensation.  They assert you get compensation, but nothing more.  You can take up that disagreement with our chess overlords

I'd agree with your point, but I don't think it's quite answering the question being asked.  If I understand the question about "falling off" correctly, what the OP means is that many club players score a fair amount better than 50% playing the Evans, whereas it's clear that grandmasters don't. For a crude metric, the Lichess database has their players scoring 55% with the Evans, but my database of master games gives White only 49%.  So the question is, at what point does that healthy plus score start to turn into a dead-equal score.  I'm curious too, but I'm guessing around 2000-2200.

#31
"I'm guessing around 2000-2200."
++ Perfectly playable below 2500.

Evans gambit is my fav opening

kasparov was even a huge fan and believer beating super GM's with it but honestly its hard to tell as you can debate he won just cause he is kasparov and not cause of the gambit but i beg to differ cause if the Evans has been a very solid attacking opening which many use and win a lot then i say its a pretty good opening for any level, even alphazero used it against SF so its good

#30
"Static vs Dynamic advantages." ++ The dynamic advantage is ephemeral.
If white cannot make good use of it, then it evaporates and the static disadvantage remains.

"White has compensation for the pawn." ++ Yes, but is it enough?

"The whole point is the fight over which is more important."
++ Prevalent grandmaster and ICCF opinion is that the dynamic compensation does not outweigh the pawn. That is also why the Evans Gambit is no longer played in ICCF correspondence as it was by Estrin and why GMs only play it for surprise value.

"The point was the London doesn't try to fight for an opening advantage" ++ The opening advantage is an illusion from the previous century. Chess is a draw. Black has more ways to err than white. The London is sound and develops pieces into play and aims to control the center.

"despite giving up a pawn, both openings are roughly equal" ++ London > Evans
Many top grandmasters have regularly played the London. The Evans Gambit is only for surprise.

"So, the equal evaluation is a mirage?" ++ It is not even equal, it is minus something.

"The roughly equal results at the highest level is also a mirage?"
++ It is no longer played at the highest level except as a surprise. It is no longer played in ICCF.

"There are a lot of things I wouldn't play if $1M was on the line, including things like the London, Colle, Torre, Reti, KIA, or the Italian" ++ You win: you get 1 million; you lose: you pay 1 million, or get 1 million debt; you draw: no payment. What would you play? Evans?

"Black cannot just know the Lasker Variation to refute the Evan's" ++ Yes, the Anti-Lasker as well

"the only player I cannot find a single game in the database in the Evan's is Magnus"
++ So he must have reached the same conclusion that the Evans is dubious at his level

"some rapid/blitz" ++ In rapid / blitz anything goes

"players significantly weaker playing others who are significantly weaker can use it" ++ < 2500

"Leela and Stockfish disagree with your assertion that you get no compensation."
++ There is compensation, but not enough and it evaporates as the game progresses.

#34
"kasparov was even a huge fan and believer"
++ No, he was not. He played it twice in serious games: against Anand and Piket and he won both, mainly because of the surprise value for the unprepared opponent.
In an interview he said it is dubious and he does not recommend to play it.

@tygxc I wanted to say all of that as a reply to post #23 but I was too lazy and complacent (not something I'm proud of but still), so thanks for taking the time.

#38
OK so it is played in ICCF > 1900, but no longer at the highest level like in ICCF WC etc.
Evans is fun, even in correspondence, so some of the > 1900 players may play it for fun.

Well, yes, I agree. Did you read the rest of my post? I explicitly said I wasn’t saying it was unplayable at a higher level. I was trying to place the spot where white ceases to have a statistical advantage. 

#41
Nevertheless, in ICCF nobody plays Evans any more in say the World Championship, like Estrin did in his time. In over the board play Grandmasters venture an Evans Gambit from time to time for surprise value, but there is no surprise value in ICCF. Some may still play it in less important ICCF events for the fun of it.

You have moved your goalposts so far it is hilarious.  The Torre has only been played in 1 game in a WC match in the last 15 years (probably more, but that is enough to make the point).  Does that mean it is unsound?  The Colle hasn't been played in a WC match in that same time period.  Nor the KIA.  The London may have been in a single game or two (but none that I am seeing in my database).  Hell, even the Italian hasn't been played at the WC much in recent years.

What gets played in a WC match does not indicate other openings are unsound.  It is an indication the players have found something they think is interesting and may lead to a win.  The Berlin did not have a good reputation for almost 100 years before Kramnik brought it out against Kasparov.  Now everyone plays it.

And yes, if White doesn't make use of a dynamic advantage (either with an attack or converting it to a static one), it evaporates.  That is the very definition of a dynamic advantage.  Thus, White does get compensation and it is up to him to do something with it.

#43

"Torre", "Colle", "KIA", "London" ++ None of these sacrifice a pawn

 "Italian" Was played in Nepo-Carlsen and interesting enough Carlsen did not play 3...Bc5 allowing the Evans, but instead 3...Nf6, but Nepo did not grab the pawn with 4 Ng5, but played the tame 4 d3 is a must win situation.

"hasn't been played at the WC much in recent years."
++ We were discussing ICCF world championship, i.e. correspondence

"Berlin" ++ Yes, and Scotch too, but no loss of a pawn.

"White does get compensation and it is up to him to do something with it."
++ Yes, but can he after 1 e4 e5 2 Nf3 Nc6 3 Bc4 Bc5 4 b4 Bxb4 5 c3 Ba5 6 d4 d6 7 Qb3 Qd7

You keep saying this like a pawn sacrifice ends the game.

The above position is reached from a sideline of the French Defense, Closed Tarrasch Variation.  Notice Black is up a pawn.  Your assertion is that this means the line is dubious - in fact, it is, for BLACK.  Black is worse here.  Saying "it sacrificed a pawn" as if that ends the discussion is asinine and ignorant.

1) I said it hadn't been played much, not at all.

2)  Comparing Centaur Chess to Human Chess is asinine.  Even in the former, the Evan's draws more often than anything else.  The original question asked when it became dubious for human games (i.e. it doesn't) and you start making the case using centaur chess.  As @pfren already pointed out, you are wrong there as well.

As already stated, material is not everything.

Yes, this was a rapid game, but it was a rapid game between 2 members of the 2800-club.  White didn't seem to fall apart after sacrificing the pawn.

So what about at classical time controls?

Seems those players are well above 2500.

And here is an interesting example where a lower rated GM beat a GM with a rating almost 100 points higher than him in the Evan's:

I'm not saying White is better after Black accepts the gambit, but he certainly is not worse (which is your implication).  And the results at the master level and in recent ICCF games coincide with that.