How the Clock impacts the Outcome of Blitz Games

Looking at scores depending on time advantage

In the past, I have looked at how the outcome of classical games is affected by the evaluation of the position. I didn’t consider blitz games, as I’ve no clear idea of how the clock impacts the result of the games. So now I want to look further into that.

In this post, I’ll only look at OTB 3+2 games between players rated above 2500. The games are mainly from the Blitz World Championships from 2021-2024.

How evaluation impacts the score

To start things off, I looked at how the evaluation impacts the score in blitz games.

Score of blitz games depending on the evaluation

As one would expect, the score follows a sigmoid curve, where the score slowly improves for much worse positions and then increases rapidly when the evaluation is closer to equal, before slowing down again in winning positions. However, the scores plateau at around 90%. One reason for this could be the impact of the clock on the outcome of the games.

How time impacts the score

As the score is highly dependent on the evaluation, I decided to first separate the positions based on it.

I chose three classes, equal positions, which have an evaluation between -0.5 and 0.5, better positions with an evaluation between 0.5 and 1.5, and much better positions, where the evaluation is between 1.5 and 3. Note that I didn’t look at worse positions, as these are just better positions for the other player.

Plotting the score based on the time advantage for each of these classes gives the following graph.

Score of blitz games depending on the time advantage

There is a clear upward trend in the score as the time advantage increases. This is especially noticeable in equal positions, where the score is below 30% when the player is 120 seconds down in time and climbs to about 70% in positions with a time advantage of 120 seconds.

The trend is also there in better positions, but not as strongly as for the equal positions. One reason may be that a 2-second increment is often more than enough for 2500+ rated players to convert better positions. There are also fewer positions that are better or much better, so the data is noisier. This is particularly clear for situations with severe time disadvantages.

Final remarks

I looked at a couple of different stats involving time, the absolute time advantage, the relative time advantage, and also the time left. The clearest picture was just the absolute time advantage, so I just included that plot.

Another interesting angle to look at the impact of time is to see if the time advantage allows a more accurate prediction of results than the evaluation alone.

The impact of the clock may also be stronger for lower rated players, especially in good positions. In the future, I may look at online games from Lichess, which also has the advantage that there are enough games to separate them into more rating bands.

Let me know if you have any other ideas to look at the impact of the clock on blitz games.