luminouscuboids

A Double Eval Bar

Chung van Gog, 02 October 2022

In chess tournaments, the spectators can sometimes see the evaluation bar for a given position, based on raw engine analysis. But this analysis doesn’t take into consideration (a) time, (b) player ratings, and (c) human factors, such as how easy a position is to win or draw as a human. Certainly we can make a decent model that factors in all these factors and predicts the player’s chances at any given moment in time. Wouldn’t it be interesting to have the output of such a model superimposed on the objective evaluation, perhaps as a red line?