What to do with openings stats ?

Well, this is a general question about openings.

Even if we try to concentrate on specific openings, it nevers goes the way we like it, and we go into many many variations. I noticed that examining my stats from about 2 years of "15 0" games on ICC, i though i was doing more or less the same things but in fact it goes into lots of branches.

Here is my stat file -->
http://zwim.free.fr/utils/openings_stats.txt

This is quite hard to get through this kind of data, and generally i prefer to go through a tree explorer available on some PGN database tools, but it gives me a quick idea of what i am doing most, and where i win and loose.

But there is no major killing opening nor a one where i really sucks, in my games, i find out that it seems quite balanced between win and loose, with some small exceptions here and there.

So my question is, how to choose the opening that better suits me and then practice it in more advanced way ?
- Do i have to choose the one i practice most : sicilian
- Or the one with better results : seems to be ruy lopez berlin defense
- Or some other with only a few variations i have not bad results with so i can lead the game in the direction i want : for example the hungarian defense
- Or even do i have to choose any and just practice it

well, the stats wil be more useful if you will have some rating of the opponents attached to them too. For example, if in a particular opening/variation you make 60% against opponents rated in average with 100 points higher than you then you can safely say that you are quite strong in that particular opening and you maybe should focus more on it.

It does make sense, so i have written a script of mine with the following principles.

If I win using one opening i credit it :
+1 point if opponent is not more than 100 elo above me
+2 points if opponent was 100 elo above me
+3 points if opponent was 200 elo above me

If I loose then it is :
-1 point if opponent is not less than 100 elo below me
-2 points if opponent was 100 elo below me
-3 points if opponent was 200 elo below me

And if i draw :
-1 point if opponent was 200 elo below me
-0.5 point if opponent was 100 elo below me
0 point if opponent is in 100 elo range with me
+0.5 point if opponent was 100 elo above me
+1 point if opponent was 200 elo above me

Finally i sum up the points that gives me how i score using some particular opening.

Here are the results, and they are much more significant than my previous file.

http://zwim.free.fr/utils/openings_5.txt

I should conclude this :

Try to avoid these openings at all cost and learn more about not to fall in their traps the future.
-20 : QGD Slav defense: exchange variation
-13 : Queen's gambit accepted
-12 : English: symmetrical variation
-12 : Sicilian: Pelikan (Lasker/Sveshnikov) variation
-11 : King's Indian defense, 3.Nc3

Be happy with these 6 others, and try to study them for even sharper middlegame.
12 : Ruy Lopez: Berlin defense
14 : Scandinavian (center counter) defense
15 : Queen's pawn game
18 : King's pawn game
19 : Philidor's defense
42 : Sicilian defense

i assume that you split the results by colours, right? I mean, a statistic for white and one for black. for black, an ideea will be to choose the best 2 openings against e4 and the best 2 against d4 and focus on them. for white is the other way around: choose the worst openings for you and work to improve (assuming that you are already established yourself as a e4/d4/c4 player; if not, just pick the first move with which you have the best results)

Just pick all the games you have played within an opening and calculate your ELO for that subset of games. That is probably the best way to find out how well you are doing.

Well, it is a variant for what i've done, but i don't want to complicate the script with ELO calculations.
And i find out that ChessDB is not calculating what i want. Find me a tool that does correctly the job and i will use it (if it's free...).

Anyway i'm already satisfied with the results, they highlights a few openings of mine, so i now know where to go.

Yes Randy, i've split up black and white before calculating the results. But even though it's not so important, because when you know an opening for one color you are also able to take advantage of it with opposite color when your oponent uses it, since you already know where the strongness and weakness is.

For example i'm now defending quite well with classical sicilian and because of this, i punish blacks when they deviate from classical line, since i know i was defeated in such a way when i deviated myself from this line as black before.