elo.py finds the limiting distribution of Markov chains with state space 1000-2800 representing the elo score of a player P of skill level 1500 playing against a queue of identical opponents each with score and skill Elo_a in {1000, 1500, 2000}.
elo.png shows the limiting distributions all align at 2800, meaning that playing infinite games, Player P's elo score converges to 2800 even though their actual skill does not improve.
This is from the working branch