If you have followed my posts very closely, you may have been able to piece together how the forecasts are generated. Here's all the information in one convenient place. It's only a few pages long and non-technical:
The careful reader may note that the forecast does not account for whether a player has White or Black. This should not be much of an issue, since in each tournament, the number of Whites and Blacks is either equal or nearly equal. Thus, any errors should cancel out.
Nevertheless, I have been working on a model that accounts for whether a player has White or Black in each game. The main motivation is not to make more accurate forecasts (any improvement will be too small to be noticed), but rather to test if the draw rate after 1.d4 is higher than it is after 1.e4. Currently, I am still refining the model and subjecting it to further testing.
No comments:
Post a Comment