Openings vs. Ratings

again, what does theoretical opening exactly mean.. without examples first. with some characterisitcs so that i could myself in front of such an opening, understand that, well it seems i recognize here the characteristics of a theoretical opening.

I.e. not a list of openings to define theoretical.. or not only.. i am bad with names to start with. but not with well defined concepts.

Very interesting -- is the code available somewhere? I'd be very interested in one specific position: in the Accelerated Dragon at the crux where they choose between Nc3 and c4 (Maroczy!).

With the Sicilian Defense: Accelerated Dragon ... with all speeds and ratings, it's Nc3 that is common.
lichess.org/opening/Sicilian_Defense_Accelerated_Dragon/e4_c5_Nf3_Nc6_d4_cxd4_Nxd4_g6
lichess.org/analysis/pgn/e4_c5_Nf3_Nc6_d4_cxd4_Nxd4_g6

Rating / Move (all speeds) ... c4 was not the most preferred move, but it was in the list. So I counted to see how deep in the list of preferred moves it was in classical games.

Rated 600 The move choice is Nxc6 ... Class G player, ... c4 was #7 in the list of preferred moves;
800 (Unknown) ... Class F, ... c4 was maybe #7;
1000 Nxc6 ... Class E (C5), ... #6
1200 (Unknown) ... Class D (C4), ... #6;

1400 Nc3 ... Class C (C3), ... #5;
1600 Nc3 ... Class B (C2), ... #4;
1800 Nc3 ... Class A (C1), ... #4 (I would have assumed it reaching #3 position);
2000 Nc3 ... Expert (CM), ... #2;
2200 Nc3 ... Master (NM) ... #2;
2400 Nc3 ... Grandmaster (GM) ... c4 is preferred choice #1 position, if in rapid or classical games, but not in blitz and faster time controls. @giziti

It took some time, but I was curious too. It was done manually by clicking at the ratings on the link above.
The lichess database preferred move was Nc3, but the master database was c4. I guess the master database is more GM level moves.

I just spent a solid 10 mins trying to figure out which kind of sick prep is behind 1. d4 Nf6. 2. c4 e6 3. Bh6 and why so many high rated players would play this until I saw g3 had the same color. Time to get some coffee

Was nobody else maddeningly curious what the scale on the x-axis was? I can kind of work out where I am roughly based on the fact I very rarely get "treated" to 1. e3.
BookBuilder, which Nate has talked about before, is a great implementation of this stuff. Just need to rebuild the opening prep every two hundred rating points or so, easy.

@Sachington said in #25:
> Was nobody else maddeningly curious what the scale on the x-axis was? I can kind of work out where I am roughly based on the fact I very rarely get "treated" to 1. e3.
> BookBuilder, which Nate has talked about before, is a great implementation of this stuff. Just need to rebuild the opening prep every two hundred rating points or so, easy.

Yes, as a former scientist I found the lack of an x-axis a bit jarring. You get an idea (assuming each plot is scaled the same) based on the prose, but I'd have loved the ratings on the x-axis.

I think there's also a big difference by region. For example in our city top trainers used to teach and play themselves Averbakh against King's Indian and 4.f3 against the Nimzo so in local tournaments those moves were disproportionally popular. Now everybody has engines and databases, not having to rely upon someone's teaching, but still I notice when I play on Lichess I get a different set of variants played against me most commonly then in local OTB tournaments.

@giziti said in #22:
> Very interesting -- is the code available somewhere? I'd be very interested in one specific position: in the Accelerated Dragon at the crux where they choose between Nc3 and c4 (Maroczy!).

similar presentation, using R as data analysis platform. did not look. but probably using lichess API from there.

@melgrove said in #14:
> I did this very same project a few years ago, it looks like we have similar results! I'm curious about your methodology, what I found interesting is that the results differ widely by time control. See here: github.com/melgrove/chess-openings

@bubbleboy said in #26:
> Yes, as a former scientist I found the lack of an x-axis a bit jarring.

in contextes where development has priority over analysis or research, the data analysis presentations habits gets lost pretty fast.... as we should all know what everyone is talking about anyway... (ranting)..

i hope that open source code spirity would also make it to data sharing. but data analysis is not used as centrally as in scientific contexts. more as suggestion for a new development idea... but chess could be viewed with scientific eyes, as if it was a natural phenomenon.. where every quantitiy shown might be trusted enough to have own axix label, and dimension units...

points could have some legend about their dispersion, some sense of amount of data behind a line....

I think it goes with predicting the background of audience.. if one is used to talk about same things with same background people over and over, the amounts of effort put into completeness of data result presentation loses its necessity.. communication goes faster...

internet forum means unknown audience characteristics. so we project what we can, depens on how many different walks of thinking we have encountered in life.. me i babble forever, having had to deal with many backgrounds....