Stats about Castling

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Stats about Castling

    One of my students, an adult new to chess but I would estimate about 1200, wanted to know the stats about castling.

    I explained to him the importance of castling, how the centre determines if you should be first or second to castle, why you castle, etc. but he wanted something more concrete and I am not sure how to go about finding the following info:

    What are the percentages when:

    - The centre is open and White/Black is the only one to castle?
    - Same as above but the centre is blocked?
    - The centre is open but White/Black castle first?
    - Same as above but the centre is blocked?
    - The centre is open but no one castles?
    - Same as above but the centre is blocked?

    This does have the advantage of quantifying to some extent how important it is, for example, to be the only one to castle when the centre is open or how advantageous it is to delay castling when the centre is blocked.
    "Tom is a well known racist, and like most of them he won't admit it, possibly even to himself." - Ed Seedhouse, October 4, 2020.

  • #2
    Re: Stats about Castling

    When White plays e2-e4 he practically always castles first. And if Black delays castle usually he gets troubles.
    When White plays d2-d4 (Nf3, c4 etc.) very often Black castles first, 'cause White tries to develop his queen's side first in order to put pressure on the centre. And white king stays in the centre much more often than black one
    Something like that. Have no idea how to calculate percentage.

    Comment


    • #3
      Re: Stats about Castling

      Hi Tom, not sure how do all the percentages, but I am sure there is some way to do it in chessbase. The closest thing I can think of is, you can do a position search in to see what the results are, when one side castled before the other.

      I did a search where I put the Black King on e8, the White King on g1, and the White Rook on f1 in my database, and I got the following numbers:

      When White Castles Before Black

      White - 38% 386582 games

      Draw - 33% 323773 games

      Black - 29% 299687 games


      When Black Castles Before White

      White 38% 324543 games

      Draw 32% 269575 games

      Black 30% 257857 games



      This doesn't take into account some of your criteria, like if the centre was blocked, or if one side even castled at all but it could give you some idea.

      Comment


      • #4
        Re: Stats about Castling

        IMHO, it is like doctrines - develop pieces, capture the center, no useless pawn moves on a&h, no early queens moves, etc. They are good to follow to a certain level (including rating). Though there are tons of chess variations were nobody follows those rules.

        Comment


        • #5
          Re: Stats about Castling

          Hi Tom

          I have ChessBase 8.0

          Getting the winning/drawing percentages you wish for would require [edit: averaging] the percentages found in each of [edit: a number of groups of] many cases, using the 'position search' search mask feature the way Liam described. The cases for blocked centres would include at least the most common four (W:Pd4,Pe5;B:Pd5,e6 and W:Pd5,Pe4;B:Pd6,Pe5 and W:Pd3,Pe4;B:Pd4,Pe5 and W:Pd4,Pe3;B:Pd5,Pe4) plus any far less common cases you might conceive of. Lets's say for the sake of argument just the most common cases are included. For open pawn centres, make a list in similar fashion, including semi-open cases if you wish (e.g. W:Pd4,Pe4;B:Pd6,Pe5), which can turn into either closed or open centres. [edit: note that if you search using the CB 8.0 search mask feature 'Material', you can search for 'Blocked'/'Not Blocked', but the position of one side's pawns is considered blocked if only all of that side's pawns are blocked :(]

          I think you'll find you can easily generate a dozen (or more) cases based on central pawn positions alone (I've assumed no doubled etc. pawns). Multiply this by all of the (3x3=9) cases of the positions of the kings, as Liam has described above. You can specify for how many moves minimum in a given game(s) that the position(s) of given piece(s) haven't changed (you also can do this as far as material balance goes), if you wish to see whether a player hasn't castled for such a long period of time (or whether the central pawn structure has stayed the same for such a time, if you wish).

          In short, there are a lot of cases to search for in a database (and then to appropriately [edit: take averages of] the percentages found [edit: for desired groups of the cases]). If I am correct that this way of doing it is unavoidable, I'm afraid your adult student, probably without knowing it, has sent you on a bit of a fool's errand. :)
          Last edited by Kevin Pacey; Sunday, 26th February, 2012, 09:01 AM.
          Anything that can go wrong will go wrong.
          Murphy's law, by Edward A. Murphy Jr., USAF, Aerospace Engineer

          Comment


          • #6
            Re: Stats about Castling

            Megabase 2012 in the strategy tab as displayed under Chessbase 10 has a key for king position though the uncastled options look for games where the uncastled side lost. You might be able to change this key to get all the games in the database and thus be able to generate winning percentages.
            Last edited by Vlad Drkulec; Sunday, 26th February, 2012, 07:18 PM.

            Comment


            • #7
              Re: Stats about Castling

              Hi Tom,

              I've often been puzzled as to why better quantified data hasn't been assembled on topics like this.

              My own idea, which I have for several years, is for chess databases to include a 'Castling' categorization for all games, similar to an ECO code.

              This could be abbreviated something like this: with White listed first and Black second, there would be nine possibilities:
              a) White castles K and Black castles K (K/K);
              b) White castles K and Black doesn't castle (K/N);
              c) White castles K and Black castles Q (K/Q);
              d) White castles Q and Black castles K (Q/K);
              e) White castles Q and Black doesn't castle (Q/N);
              f) White castles Q and Black castles Q (Q/Q);
              g) White doesn't castle and Black castles K (N/K);
              h) White doesn't castle and Black doesn't castle (N/N);
              i) White doesn't castle and Black castles Q (N/Q).

              This system could then be applied within each ECO code, to give a significant partition function on most effective play, as one more determining factor. For example, if White is scoring 65 per cent in one particular 'castling properties' sector, that is a pretty good indication the variation isn't working very well.

              Cheers,
              Frank Dixon
              Kingston
              :)

              Comment


              • #8
                Re: Stats about Castling

                I normally castle where the theory articles on the openings and variations I am playing indicate I should castle. Also, Bator has given good general guidelines.

                I have a sign that says "Remember to castle". I keep it beside the sign that says "The dog eats every day". :)
                Gary Ruben
                CC - IA and SIM

                Comment


                • #9
                  Re: Stats about Castling

                  You could even add in who castled first rather easily.

                  Actually it wouldn't be too hard to write a program that could analyse PGN files and come up with this data. It might take a while to actually DO the analysis...
                  Christopher Mallon
                  FIDE Arbiter

                  Comment


                  • #10
                    Re: Stats about Castling

                    Incorporating Chris Mallon's helpful suggestion (thanks Chris!), we can, for example, use a '*' to designate which player has castled first. This means there will be two further potential sub-branches, referring back to my first post.
                    :)

                    Comment


                    • #11
                      Re: Stats about Castling

                      Originally posted by Tom O'Donnell View Post
                      One of my students, an adult new to chess but I would estimate about 1200, wanted to know the stats about castling.

                      I explained to him the importance of castling, how the centre determines if you should be first or second to castle, why you castle, etc. but he wanted something more concrete and I am not sure how to go about finding the following info:

                      What are the percentages when:

                      - The centre is open and White/Black is the only one to castle?
                      - Same as above but the centre is blocked?
                      - The centre is open but White/Black castle first?
                      - Same as above but the centre is blocked?
                      - The centre is open but no one castles?
                      - Same as above but the centre is blocked?

                      This does have the advantage of quantifying to some extent how important it is, for example, to be the only one to castle when the centre is open or how advantageous it is to delay castling when the centre is blocked.
                      Ask him if he wants to be on the wrong side of a brilliancy award. Then tell him that chess enables castling kingside and queenside for a reason, why would a game have such a unique rule if it's not frequently used?
                      Shameless self-promotion on display here
                      http://www.youtube.com/user/Barkyducky?feature=mhee

                      Comment


                      • #12
                        Re: Stats about Castling

                        Originally posted by Bindi Cheng View Post
                        Ask him if he wants to be on the wrong side of a brilliancy award.
                        hehehehe

                        Comment

                        Working...
                        X