Using this website’s rating calculator, I’ve found it interesting to calculate a “performance rating” on various openings that I play a lot. I suppose it doesn’t really tell me anything I shouldn’t already know, but when the last 6-12 games with one opening give me a performance rating of 1800, and the last 6-12 with another give me 2130, it’s hard to ignore. Couple lines I’ve used for years are destined for the scrap heap.
Anybody else goof around with this sort of thing?
Yes, when I was playing regularly I spent a lot of time (too much time, probably) analyzing my results with spreadsheets, etc.
At first I used pencil, paper and a calculator, but my latest efforts at analyzing my results have used Microsoft Access queries. Besides analyzing my opening results (inspired by Simon Webb’s “Chess for Tigers”, among other books) I’ve also attempted to analyze my performance in the middlegame and endgame. The latest version of my Games table has 41 fields:
GameID - a unique ID number for each game
TournamentID - index into a table of tournaments that I’ve played in
Round
Date
Result
Color
MyRating - my rating at the time the game was played
OppRating - opponent’s official rating at the time the game was played
OppEffRating - opponent’s effective rating, e.g. an estimated playing strength for unrated opponents
ECO - Encyclopedia of Chess Openings code for the opening
OpCategory - opening category, one of “1. e4”, “1. d4” or “Other”
Opening - opening name, e.g. “King’s Indian Defense”
Variation - opening variation, e.g. “Panno”
MyTac - my tactical mistakes (see below)
MyAtt - my attacking mistakes
MyDef - my defending mistakes
MyPos - my positional mistakes
MyOp - my opening mistakes
MyEnd - my endgame mistakes
MyTim - my time management mistakes
OppTac - opponent’s tactical mistakes
OppAtt - opponent’s attacking mistakes
OppDef - opponent’s defending mistakes
OppPos - opponent’s positional mistakes
OppOp - opponent’s opening mistakes
OppEnd - opponent’s endgame mistakes
OppTim - opponent’s time management mistakes
MySac - 1 if I sacrificed material, else 0
OppSac - 1 if opponent sacrificed material, else 0
MyIni - 1 if I had the initiative, else 0
OppIni - 1 if opponent had the initiative, else 0
MyWkP - 1 if I had pawn weaknesses, else 0
OppWkP - 1 if opponent had pawn weaknesses, else 0
MyCenP - number of center pawns I had (at the start of the middlegame)
OppCenP - number of center pawns opponent had
My4 - number of pawns I had on 4th rank and beyond
Opp4 - number of pawns opponent had on 4th rank and beyond
My5 - number of pawns I had on the 5th rank and beyond
Opp5 - number of pawns opponent had on the 5th rank and beyond
Open - 1 if there was at least one (fully) open file, else 0
HOpen - 1 if there was at least one half-open file, else 0
I created various queries to analyze the data in this table, e.g. performance rating with each opening or variation, using either official ratings or effective ratings for the opponents, rating change with each opening or variation (using an approximation with a K factor of 32), opening mistakes made by me and my opponents in each opening or variation. All of this can be broken down by time period, opponent’s playing strength, etc.
“Mistakes” are based on the concept that at any point in the game White’s and Black’s chances of winning can be expressed as a number from 0 to 10. 0 means completely winning for Black, 5 means equality, 10 means completely winning for White. I assume that White has a slight advantage at the start of the game: 6. If the opening ends in an equal position it’s now 5, so White lost 1 point and Black gained 1 point. I would record this as MyOp = 1 if I were White, OppOp = 1 if I were Black. If later on I made a blunder which gave my opponent an almost winning advantage, evaluating the position as 1 where I was White, I would record this as MyTac = 4 (tactical error resulting in a change in evaluation of 4 in my opponent’s favor). If the opponent maintained his advantage going into the endgame but I managed to draw the game I’d record it as OppEnd = 4 (opponent’s endgame mistake changing an evaluation of 1 into 5). This mistake analysis was inspired, in part, by “Catalog of Chess Mistakes” by Soltis.
My4, My5 etc. are an attempt to figure out what kinds of positions I’m good at playing. I didn’t get very far with it before I got busy with other things, like directing chess tournaments. Earlier on I used chapter and section numbers from “The Middle Game, Book One” by Euwe and Kramer.
I really went crazy with this stuff. A lot of it isn’t relevant to analyzing results with openings, but some of it could be useful:
Breaking down results by variation instead of just by opening
Looking at the evaluation of the position at the end of the opening and not just the final result
The ability to look at results over time and against various classes of opponent
Geez. I’m still trying to figure out Fool’s Mate.
The list for analyzing mistakes is interesting. Why do you think that it is accurate or gives the real reason for the error? Most mistakes occur for different reasons than the player thinks. Often a seemingly innocuous move foreshadows the mistake to come. In monster movies, it is not the encounter with the beast that is the mistake. Nor is it opening the wrong door and finding the monster. It is the turning down the pretty lane edged with flowers where the empty house sits instead of taking another path. We are often tantalized this way by an interesting tactic when we know or sense that it can’t be that easy.
I don’t think it’s necessarily accurate, but it might at least be better than random. If I really wanted to improve, of course, I should pay for lessons from a stronger player instead of trying to analyze my own games.
I used to keep track of my results with different openings, but eventually realized it had as much (or more) to do with who my opponents were as with the openings themselves. For example, I had (and still do have) a pretty good score as White in the Ruy Lopez – not because I play the Ruy Lopez especially well, but because lower-rated players play 1…e5 much more often than higher-rated players do (at least within the group of people I have played). So all it really tells me is that I perform better against lower-rated players. I could have figured that out with no statistics at all.
That’s a reason to look at performance rating instead of raw score. I like looking at the sum of estimated rating gains or losses. For example, if an 1800 player has a 1700 performance rating on the Black side of the French Defense and a 1500 performance rating on the Black side of Bird’s Opening, it looks like he should spend time studying Bird’s Opening, but looking at rating points lost might show 200 points lost with the French over a period of time vs. 50 points lost with Bird’s, since he probably plays the French more frequently.
Another factor, though, is that the loss of a game might have nothing to do with the opening. This will probably average out over a large number of games, but it’s reason for looking at the evaluation at the end of the opening instead of, or in addition to, the final result.
First, you mus’ find yourself a fool!
A lopsided plus score against lower rated players who use 1 … e5 need not result in a great Performance Rating, unless your results exceed what one would expect from the rating difference. Maybe you scored 6-4 when the ratings say you ought to have gotten 7.5-2.5.
I’m vulnerable to a different form of self deception: when I have one good result against a strong player with a particular opening, I’ve tended to let that result overshadow several poor results later.
Calculating OPR for openings makes it harder to fool myself in this way.
“The Blue Book of Winning Chess Charts” gave out percentage scores of the success of various openings in tournament play. IIRC, an opening like the Englund Gambit had a high score for Black based on very few game results. Just because an opening scores 3-0 does not necessarily mean that it is a good opening. Surprise value may net a point here or there, but it may not be a good idea to go back to that well too often.
Right. And then there’s the similar situation where a variation does very well until the correct response is worked out and virtually refutes it. After that, the line is seldom played, but the prior successes skew the statistics. Concrete analysis trumps statistics and performance ratings.
Oops! Never mind – I somehow got the impression we were talking about raw score rather than performance rating (even though the OP specified performance rating). I should learn how to read better.
I still have some reservations about the usefulness of all this, though. To get much out of it, you would have to play a variety of openings, and most people don’t – at least not voluntarily. Most of the variety in my openings has to do with how other people respond to the few things that I play (mostly 1.e4 with White, mostly Sicilian or King’s Indian with Black). Of course, there are numerous defenses to 1.e4, and numerous variations of the Sicilian and KID, but those are the other guy’s choices, not mine.
In my case I have played a fairly wide selection of openings over the years, so I do have at least some basis for comparison. Even if you always play the same openings it might be helpful to know which openings and variations you have the worst results with so you know which ones to study. If you do really badly in some opening you might decide to try something new. For example, if you have good results as Black with the Sicilian but bad results with the King’s Indian you might try playing the Nimzo-Indian or the Benko Gambit.
Granted, though, I probably spent too much time analyzing my results as opposed to just playing chess and studying between tournaments.