Corpus ID: 201049745

NORMALIZED ELO

@inproceedings{Bergh2019NORMALIZEDE,
  title={NORMALIZED ELO},
  author={M. D. Bergh},
  year={2019}
}
It is often discussed how one should compare the suitability for engine testing of opening books with different characteristics (balanced versus unbalanced, drawishness). Here we answer this question. A lot of inspiration came from Kai Laskos’ empirical results posted on talkchess as well as from his comments. Credit goes also to him for first noting that the trinomial model is insufficient for working with unbalanced openings.