Minimizing the impact of Randomness to the Strength Rating Revision in Chess

The Elo rating system is used since 1970 in chess by which the players can be evaluated based on their formal games. This system is based on the win or loss probabilities described by a well fitted logistic curve. In this context, the mathematical expression of strength rating revise of a player, uses a constant development coefficient, the so called K-factor, having individual values depending on the strength level player's category. In the present work, a fairest strength rating revision is introduced, aiming to minimize the impact of the randomness. A variable development factor included ties in the games is used, based on the fact that large uncertainty, arising from randomness, is present mainly in evenly strength rating players and decreases gradually for greater strength rating differences. The function used for realizing the variable strength rating revise is based on the notion of information entropy. The proposed strength rating revision can boost the rapidly improving or talented chess players against others whose the rating increases mainly due to randomness around the pre-rating equivalence region.