The Shape of Time: How Openings Evolve from Blitz to Classical

@tpr thanks for the thorough answer, appreciate the effort and quality.
I agree with most of what you said but I have 3 remarks/questions:

"A blitz game by strong players is like a rapid game between medium players and like a classical game between weak players."
=> That was my prediction, and it was wrong. Can you double check your logic (maybe draw the charts that you'd expect to see) and confirm? If this sentence is true, then the blitz curve should be on the right of the classical one, but it's on the left.
There is a lot of certainty in you message. Is this your intuition or are there fundamental reasons that make these claims true. For example
Your explanation for the gaps in amplitudes of the curves is that in Blitz people make so many mistakes that they opening advantage doesn't matter so much. Didn't think of it this way, interesting

@tpr thanks for the thorough answer, appreciate the effort and quality. I agree with most of what you said but I have 3 remarks/questions: - "A blitz game by strong players is like a rapid game between medium players and like a classical game between weak players." => That was my prediction, and it was wrong. Can you double check your logic (maybe draw the charts that you'd expect to see) and confirm? If this sentence is true, then the blitz curve should be on the right of the classical one, but it's on the left. - There is a lot of certainty in you message. Is this your intuition or are there fundamental reasons that make these claims true. For example - Your explanation for the gaps in amplitudes of the curves is that in Blitz people make so many mistakes that they opening advantage doesn't matter so much. Didn't think of it this way, interesting

LKama

@ixdap interesting point but when I look like the distributions of ranks they look very similar between time controls:
https://lichess.org/stat/rating/distribution/rapid

@ixdap interesting point but when I look like the distributions of ranks they look very similar between time controls: https://lichess.org/stat/rating/distribution/rapid

LKama

@Toscani glad you tried the script! It actually does make a local save of all the requested API call. Check the folder ".wickedlines"

If you request something that is present locally, it won't fetch it again unless you use the flag that forces the refresh.

Regarding your usage of the tool, I believe you can use it to find opportunities in your lines. Having stats on your past play doesn't automatically tell you how to play better. But studying common traps (with the hunt mode for example, or manually with the line mode) for your variation could.

At the end of the day it's mostly for exploring stuff.

Best,

@Toscani glad you tried the script! It actually does make a local save of all the requested API call. Check the folder ".wickedlines" If you request something that is present locally, it won't fetch it again unless you use the flag that forces the refresh. Regarding your usage of the tool, I believe you can use it to find opportunities in your lines. Having stats on your past play doesn't automatically tell you how to play better. But studying common traps (with the hunt mode for example, or manually with the line mode) for your variation could. At the end of the day it's mostly for exploring stuff. Best,

tpr edited

"A blitz game by strong players is like a rapid game between medium players and like a classical game between weak players."
*Here is a 5+0 blitz game between strong players.
https://www.chessgames.com/perl/chessgame?gid=1044673
I would be happy to play a classical game like that.

"in Blitz people make so many mistakes that they opening advantage doesn't matter so much"

Here is a Carlsen 3+2 blitz game from the blitz World Championship:
https://lichess.org/evYvYqyF#1

Your colleague blogger @jk_282 has done some research "Looking at the Quality of Rapid and Blitz Games"

"A blitz game by strong players is like a rapid game between medium players and like a classical game between weak players." *Here is a 5+0 blitz game between strong players. https://www.chessgames.com/perl/chessgame?gid=1044673 I would be happy to play a classical game like that. "in Blitz people make so many mistakes that they opening advantage doesn't matter so much" * Here is a Carlsen 3+2 blitz game from the blitz World Championship: https://lichess.org/evYvYqyF#1 Your colleague blogger @jk_282 has done some research "Looking at the Quality of Rapid and Blitz Games"

Toscani

I think the blog images was using lichess database, not OTB 2012 games. So, using an example from lichess databaase would have been more appropriate than using an old OTB game. The PGN provided lacks critical details like exact time controls, engine version, and depth of analysis, which complicates precise interpretation. Words like Blitz and Rapid mean little to without actual time controls used. The game seems to be from the 2012 World Blitz Championship, where time controls were typically 3 minutes + 2 seconds per move. Stockfish 2.3 was released on September 15, 2012. Just after that game was played. So I think the evals in the pgn could very well be from that time period. But I think over all it really does not matter all our comments and dissecting quotes because the suboptimal moves are give.

In a 3+2 game I tend to play slower than masters which should be expected since their rating is near double mine. So I end up taking twice as much time to play. Move quality is affected by both evaluation (centipawn loss) and time spent.
https://lichess.org/insights/Toscani/movetime/winPercent/variant:blitz
https://lichess.org/insights/Yarebore/movetime/winPercent/variant:blitz

I don't envy the way Master's play their games. I can just imagine the agony of their defeat or their moves getting scrutinized by everyone. Thanks but no thanks. Even blogs and forum posts get that critic treatment.

I think the blog images was using lichess database, not OTB 2012 games. So, using an example from lichess databaase would have been more appropriate than using an old OTB game. The PGN provided lacks critical details like exact time controls, engine version, and depth of analysis, which complicates precise interpretation. Words like Blitz and Rapid mean little to without actual time controls used. The game seems to be from the 2012 World Blitz Championship, where time controls were typically 3 minutes + 2 seconds per move. Stockfish 2.3 was released on September 15, 2012. Just after that game was played. So I think the evals in the pgn could very well be from that time period. But I think over all it really does not matter all our comments and dissecting quotes because the suboptimal moves are give. In a 3+2 game I tend to play slower than masters which should be expected since their rating is near double mine. So I end up taking twice as much time to play. Move quality is affected by both evaluation (centipawn loss) and time spent. https://lichess.org/insights/Toscani/movetime/winPercent/variant:blitz https://lichess.org/insights/Yarebore/movetime/winPercent/variant:blitz I don't envy the way Master's play their games. I can just imagine the agony of their defeat or their moves getting scrutinized by everyone. Thanks but no thanks. Even blogs and forum posts get that critic treatment.

LKama

@tpr I agree with what you're showing, and the "chess" aspect of it. My only remark was about the expected shape of the curve.

Let's assume this sentence is completely true (and I think it's at least partially true):

"A blitz game by strong players is like a rapid game between medium players and like a classical game between weak players."

And that all performance distributions are bell curves (just an example). Would you expect the blitz bell curve to be shifted left or right from the classical bell curve? -> This is the thing I believe we disagree on. It's minor of course, but in one case the results I found are surprising and in the other they are expected. And I believe they are surprising.

@tpr I agree with what you're showing, and the "chess" aspect of it. My only remark was about the expected shape of the curve. Let's assume this sentence is completely true (and I think it's at least partially true): "A blitz game by strong players is like a rapid game between medium players and like a classical game between weak players." And that all performance distributions are bell curves (just an example). Would you expect the blitz bell curve to be shifted left or right from the classical bell curve? -> This is the thing I believe we disagree on. It's minor of course, but in one case the results I found are surprising and in the other they are expected. And I believe they are surprising.

True_Romance

Thanks for the interesting analysis!
I have a couple of thoughts regarding the discrepancies between blitz and rapid, and classical rating gains. IMO, there is no clear connection between time and how good you play given the objective evaluation of a position. It all depends on how easy it is to blunder for the attacker, how easy it is to blunder for the defender and what the cost of a blunder is. I can give a couple of FEN examples to clarify.
These are not some top computer lines, I made some deliberate inaccuracies to get there, but these positions can be easily reached in your games.

https://lichess.org/analysis/fromPosition/r1b1k2r/pp1nbp1p/1q2p1pB/3pP3/6Q1/3B1N2/PP3PPP/R4RK1_b_kq_-_1_14
This position can be reached from the French defense. White has given away a pawn, but has massive atack on the kingside, black`s bishops are underdeveloped, it is quite easy to play for a win with white here. Just throw pieces on the kingside, use the weaknesses of the bishops, eventually you either mate or take back the material and get better position. Engine evaluation: ~+0.5, but it is much easier to play it with white. Black can defend, but almost any wrong move leads to a disaster.
https://lichess.org/analysis/standard/r1bq1b1r/pp2pkpp/2np4/2p5/4Q3/2P2N2/PP1P1PPP/RNB1K2R_b_KQ_-_0_7#13
This position can be reached from the Sicilian. Material is even, black has lost castling rights, but nonetheless, it is ok for them. White has some initiative, but black has lots of moves to defend, although it is not easy to win for black here, too. Engine eval: ~0.0
https://lichess.org/analysis/standard/r3k2r/1pp1npp1/p1pb1q2/4p2p/4P1b1/3PBN1P/PPPN1PP1/R2Q1RK1_w_kq_-_5_10#18
This can be reached from the Ruy Lopez. I have to say, I dont understand this position too well, so my following analysis is somewhat flawed. Personally, I hate playing these positions as white. Black has initiative, white pieces are cramped, you cant take the bishop usually, but you have to calculate almost on every move if you can take it or not, because sometimes you can and it is the only move. It is just annoying to play against. So, it is not easy to play for white, any wrong move and you get checkmated. Black has some space for error, but objectively they are worse, so they should play somewhat risky, attacking the kingside. Engine eval: ~+0.5.
https://lichess.org/analysis/standard/r2q1rk1/1bpp1ppp/p1nb1n2/1p2p3/4P3/1B1PBN2/PPPN1PPP/R2Q1RK1_w_-_-_5_10
Another position from the Ruy Lopez. Well, good luck winning it for any side. Although the objective eval is ~+0.5, I doubt it is easy to win for any side. You can still make a lot of dumb moves for both sides and keep the position equal. I guess some really good players can have an edge here for white, but it is really difficult to convert.
https://lichess.org/analysis/standard/r1b1kb1r/pp2pppp/2np4/q2P4/8/2P2N2/P3BPPP/R1BQK2R_b_KQkq_-_0_9#17
Another position emerging from the Sicilian. Quite difficult for me to understand for both sides. Black has an extra pawn, but thiere pieces are underdeveloped, although there is no immediate danger to their king. White has somewhat better pieces, but they have lost a pawn and have somewhat weak pawn (in my opinion, objectively it is fine, i guess) on the c-file. Both sides can play for win and have some room for error. Engine eval: ~+0.5

So, to sum up: It doesn’t matter a lot which opening you choose, who is the attacker and who is the defender. The thing that matters, IMO, is how easy it is to blunder and how disastrous your blunders are (i.e., how easily your opponent can capitalize on them). The easier it is to blunder for opponent and the easier the game is for you, the better for blitz.

Now, your analysis doesn’t seem to capture that, as far as I understand. Since you analyze too broad of openings, you can get so many different positions from e4–c5 that they will have nothing in common in terms of game plan, winning chances, initiative, etc., although still objectively playable. So, probably you would like to do the thing you have already mentioned — analyze the positions from the emerging midgame. I guess it will give you a clearer picture.

Thanks for the interesting analysis! I have a couple of thoughts regarding the discrepancies between blitz and rapid, and classical rating gains. IMO, there is no clear connection between time and how good you play given the *objective* evaluation of a position. It all depends on how easy it is to blunder for the attacker, how easy it is to blunder for the defender and what the cost of a blunder is. I can give a couple of FEN examples to clarify. These are not some top computer lines, I made some deliberate inaccuracies to get there, but these positions can be easily reached in your games. 1. https://lichess.org/analysis/fromPosition/r1b1k2r/pp1nbp1p/1q2p1pB/3pP3/6Q1/3B1N2/PP3PPP/R4RK1_b_kq_-_1_14 This position can be reached from the French defense. White has given away a pawn, but has massive atack on the kingside, black`s bishops are underdeveloped, it is quite easy to play for a win with white here. Just throw pieces on the kingside, use the weaknesses of the bishops, eventually you either mate or take back the material and get better position. Engine evaluation: ~+0.5, but it is *much* easier to play it with white. Black can defend, but almost any wrong move leads to a disaster. 2. https://lichess.org/analysis/standard/r1bq1b1r/pp2pkpp/2np4/2p5/4Q3/2P2N2/PP1P1PPP/RNB1K2R_b_KQ_-_0_7#13 This position can be reached from the Sicilian. Material is even, black has lost castling rights, but nonetheless, it is ok for them. White has some initiative, but black has lots of moves to defend, although it is not easy to win for black here, too. Engine eval: ~0.0 3. https://lichess.org/analysis/standard/r3k2r/1pp1npp1/p1pb1q2/4p2p/4P1b1/3PBN1P/PPPN1PP1/R2Q1RK1_w_kq_-_5_10#18 This can be reached from the Ruy Lopez. I have to say, I dont understand this position too well, so my following analysis is somewhat flawed. Personally, I hate playing these positions as white. Black has initiative, white pieces are cramped, you cant take the bishop usually, but you have to calculate almost on every move if you can take it or not, because sometimes you can and it is the only move. It is just annoying to play against. So, it is not easy to play for white, any wrong move and you get checkmated. Black has some space for error, but objectively they are worse, so they should play somewhat risky, attacking the kingside. Engine eval: ~+0.5. 4. https://lichess.org/analysis/standard/r2q1rk1/1bpp1ppp/p1nb1n2/1p2p3/4P3/1B1PBN2/PPPN1PPP/R2Q1RK1_w_-_-_5_10 Another position from the Ruy Lopez. Well, good luck winning it for any side. Although the objective eval is ~+0.5, I doubt it is easy to win for any side. You can still make a lot of dumb moves for both sides and keep the position equal. I guess some really good players can have an edge here for white, but it is really difficult to convert. 5. https://lichess.org/analysis/standard/r1b1kb1r/pp2pppp/2np4/q2P4/8/2P2N2/P3BPPP/R1BQK2R_b_KQkq_-_0_9#17 Another position emerging from the Sicilian. Quite difficult for me to understand for both sides. Black has an extra pawn, but thiere pieces are underdeveloped, although there is no immediate danger to their king. White has somewhat better pieces, but they have lost a pawn and have somewhat weak pawn (in my opinion, objectively it is fine, i guess) on the c-file. Both sides can play for win and have some room for error. Engine eval: ~+0.5 So, to sum up: It doesn’t matter a lot which opening you choose, who is the attacker and who is the defender. The thing that matters, IMO, is how easy it is to blunder and how disastrous your blunders are (i.e., how easily your opponent can capitalize on them). The easier it is to blunder for opponent and the easier the game is for you, the better for blitz. Now, your analysis doesn’t seem to capture that, as far as I understand. Since you analyze too broad of openings, you can get so many different positions from e4–c5 that they will have nothing in common in terms of game plan, winning chances, initiative, etc., although still objectively playable. So, probably you would like to do the thing you have already mentioned — analyze the positions from the emerging midgame. I guess it will give you a clearer picture.

JestemTomkiem edited

You compare ratings between time controls like they had an objective meaning, I don't think they do. Different players are in pools for rapid and blitz, so anyone's blitz rating will usually be lower than their rapid rating. Chessgoals has a rating comparison table that shows ratings jumping by up to 200 points with every change of format to a slower one on lichess.

Also, is there a way to filter the games by both players' rating deviation? I imagine not (based on your other comments) but that could get rid of some artifacts.

Nice project! :)

You compare ratings between time controls like they had an objective meaning, I don't think they do. Different players are in pools for rapid and blitz, so anyone's blitz rating will usually be lower than their rapid rating. Chessgoals has a rating comparison table that shows ratings jumping by up to 200 points with every change of format to a slower one on lichess. Also, is there a way to filter the games by both players' rating deviation? I imagine not (based on your other comments) but that could get rid of some artifacts. Nice project! :)

Your network blocks the Lichess assets!