Drawing Rates in Chess Pt.1 - Time Matters!
Welcome all to my new blog!
I’ve been thinking of creating a blog for some time now, but it’s only been recently that I found the time to put one together. I had many ideas of what to write about, but I finally settled on a blog about the statistics of chess. I’ve done a little bit here and there in a group forum when a question arose, and I thought that I would put some of that into a blog.
Statistics is something that I have enjoyed, mainly due to my interest in numerical facts and figures which statistics basically fits hand into glove. Add into that my obsession for chess (hence the avatar), and voila: here is the result!
I shall start this with a question that I saw back around 6 months ago, and it went something like this:
“Is there a difference between the draw rates of Blitz, Rapid, and Classical time control games? And how do they change?”
Now the obvious answer is yes. The shorter the time control, the more decisive mistakes are made. Compare that to 5-7 hours Classical games, where play is much much more accurate and so draws are more common.
Look at the last two World Championship matches between Magnus Carlsen and Viswanathan Anand, where 14 of the 21 games between them were drawn. This leads to a 66.7% draw percentage between the two, but this is at the very pinnacle of chess, where the play is often very theoretical and prepared over 20 moves deep. Part of this is down to the Berlin Defence of the Ruy Lopez, that contributed to 4 of those draws (but three decisive games!).
At top level events but not World Championship level (we’re talking mainly about tournaments like Wijk Aan Zee, Shakmir Chess, Sinquefield Cup, etc.), you wouldn’t expect two thirds of the games to be drawn. You would also expect much fewer draws in Rapid and Blitz tournaments as well.
But what is the difference between different time controls?
To find this out, I looked at some of the top events for each time control category, settling on these three tournaments to use (with their average ratings):
- Qatar Masters Open 2014, Rating-Ø 2515
- World Rapid Championships 2014, Rating-Ø 2601
- World Blitz Championships 2014, Rating-Ø 2594
Do bear in mind that the average ratings for the Rapid and Blitz tournaments are the Rapid and Blitz ratings of the players, which tend to be higher (Carlsen hit 2948 in Blitz after that championship, much higher than his 2877 Classical rating at that time).
All three of these events were held at FIDE standard time controls (Qatar Masters held at 40/90’ + G/30’ + 30” inc., the World Rapid Championship at 15’+10”, and the World Blitz Championship at 3’+2”). Also, these were all Swiss events, and had some of the best players in the world playing in them (Carlsen, Anand, Kramnik, Giri and many others).
So, here are the data that I collated from those three events, and graphed accordingly:
As you can see, the proportion of draws varies hugely from Blitz to Classical. For Blitz, 18.5% of games are drawn, and this is at the very top level of chess! If these were 1500 or so players, or maybe even a sudden death time control (like G/5’), it could be even lower! Compare that to Classical, where at the master level (that tournament with the average player of GM strength), more games were drawn than won by White.
Taking the score for each game (1 for win, ½ for a draw), we get these average results for White and Black:
You can see that White scores best at Rapid, and Black scores best at Classical. So, why does the score for White increase from Blitz to Rapid, but then decreases again from Rapid to Classical? Maybe there is no reason at all, that regardless of the time control, White has the same advantage. There is a 0.965% difference between Rapid and Classical, it could be that it is too small to make an accurate judgement, 1% is about 8 games for each tournament.
Taking into account every game from those three tournaments, the average score for White is 55.009%, so it’s likely the time control does not increase nor decrease the advantage that White has. The time only has an effect on the likelihood of the game being drawn. And this is probably true for all players; you rarely see any drawn bullet games played online (although they do happen) but you would expect to draw many more standard or correspondence games.
And we might as well look at the most recent World Correspondence Chess Championship, held from 2011 through to 2014 as a 17-player round robin where the time control is 50 days for every 10 moves, also with 30 days ‘vacation’ per year. Engine use and endgame databases are also allowed in ICCF tournaments like this one. Nine of the 17 players went unbeaten, with 5 of them drawing all 16 of their games! And in those 136 games, there were 13 White wins, 3 Black wins and a whopping 120 draws!
But what about the skill of the player? Surely the player’s level of play affects their likelihood to draw, or to hold their advantage as White?
All to come in part 2...
Huge thanks to those people who offered to proof-read this article before I posted this!