List of 3-ply openings, transitions and statistical results

Current status: 7 openings with 60000+ master games. Queen's Gambit and Anglo-Indian: King's Knight are bad for black.

Professor Igor,

I don't doubt your mathematics or statistics.   I have read a good number of comments and posts whereby you state that 1. e4 is suboptimal and that 1. d4 or 1. c4  are clearly superior to 1. e4.  I have also read your comments where 1. ... e5 in response to 1. e4 is suboptimal to the Sicilian.

Yet, SuperGrandmasters seemingly don't put much stock in statistical outcomes.  For example, in the first round of today's Sinquefield Cup there were 5 games.  4 of them started out with 1. e4.  And in all four of them, the response was 1.... e5!   And these are SuperGMs too!  And almost all of them deliberately chose to go down "suboptimal" paths?  

Don't you think that seems odd?  I mean, when they have statistical data and analysis available that shows that their choices are "suboptimal"?  Why do you think they choose "suboptimal" openings?

SeniorPatzer: This comparison answers your question:

https://www.chess.com/forum/view/chess-openings/best-first-move-statistical-comparisons-from-2200-to-2700

1. e4, suboptimal for 2200+ players, becomes excellent at 2600+ level. Also, the statistical evaluations don't reflect the whole truth. Hilbert, one of most famous mathematicians, said that when our horizon of consciousness narrows down to a single point, we obtain the so-called viewpoint.

This statistical analysis is just pure garbage. Using results of games were mistakes could have been made deep into game to "prove" what is best on the first move. There is no conceptual basis on what trying to do, unlike use of statistics in physics and elsewhere.

GM's use software for opening work, typically 10+ moves into a line, to calculate concrete lines. They would be at a competetive  disadvantage in certain concrete positions if did not. The numbers presented in these threads are no use to anybody in winning chess games.

It's clearly not intended to serve for the GM's preparations. But it reflects correct things: the fact that 1. c4, 1. d4, 1. e4, 1. Nf3 and 1. g3 are good moves etc. It can be applied to calculate evaluations of concrete lines too when there are enough master games. I just started from the very beginning. It could/should be combined with the engine analysis. I don't see any problem.

Actually, I decided to change my presentation and go deeper into main lines:

https://www.chess.com/forum/view/chess-openings/indian-game-main-lines-transitions-and-statistical-evaluations

instead of considering it generally ply-by-ply.

It's my impression that currently 1: ...Nf6 is currently the premier defense to 1: d4. Not necessarily the best, just what people in 2017 play. People are going to make lectures about games in 2017 and go "1: d4 Nf6 of course, because in the 2010s Nf6 was forced." Like how we joke that the king's gambit was forced in the 1850s.

But how much does theory change? I would be interested in that. Where are there some over-time graphs of opening popularity? I feel like there's no way it has changed much since 2000 where computer opening preparation got serious, but maybe the style has changed quite a bit even if the best moves haven't. 

Phoenyx75: Welcome to our virtual team with Differntiation2.

xman720: I saw an online database that gives statistics by time periods. I didn't use it so far.

I can't find it now but I've found a database with engine evaluations where U can filter by Elo from 1800 to 2900. 

https://chess-db.com/public/explorer.jsp?interactive=true&avelo=2700

Well, it's still a beta-version with some flaws. For example, there is no 1. g4 as the first move.

Current status: 8 openings with 50K+ games. I've simplified the presentation.

Current status: 10 openings with 40K+ games. Pirc: Main #10 is the first white-sharped opening.

Pnoenyx75:

1. "s" in "sev" stands for "statistical" in order to distinguish it from engine evaluations.
2. yes
3. they are ordered by the number of master games starting with the most popular openings
4. PSCC is my system in order to classify openings by pawn structures, cf. here https://www.chess.com/forum/view/chess-openings/pawn-structure-classification-codes

Phoenyx75: Ah yeah, right, the numbers of master games are from the local Explorer's database. Well, I could use smaller numbers from chesstempo but it doesn't change the popularity order. So I decided to keep Explorer's numbers.

Right, sh1=W/D and sh2=B/D where W is the number of white wins, B the number of black wins and D the number of draws.

Current status: 12 openings with 30K+ master games.

18 openings with 20K+ games.

I've added links to my threads describing main lines of Indian and Sicilian games.

The link to #3 King's Knight games is added too.

I started to add approximative engine evaluations (lichess and ChessBase).