multiple players, pairwise games.

[cg78ithaca]

I have a question about how to use the code. Let's say we have three players A, B, and C, and all pairwise game results are known as a timeseries. That is AB, AC, and BC. I have run the code three times, once for AB to get a set of ratings for A and B; then for AC to get an independent set of ratings for A and C, and similarly for BC. This was done by defining e.g. A as black and B as white in the AB scenario, inputting the game results just as described in the README.md documentation file.

Clearly, since it is based on independent datasets, the ratings for A from the AB run have nothing to do with the ratings for A from the AC run. As I understand it, the rating rA(t) should be game-independent, so a reflection of A's performance in ALL the games where it participates (in this case, AB and AC).

How would I run/define the code for this specific scenario, i.e. three players, three pairwise games?

Cristian