4

I'm looking to implement a Glicko 2 library in my application. However the two libraries I'm considering are different. Firstly, they produce slightly different results for the same dataset (the default one given in the Glicko paper).

Second, and what I'm interested in, is how they calculate the results. Library 1 calculates each match one-by-one. That is, n-function calls to calculateMatch(). Library 2 takes the entire set of data (all matches and players) and seemingly calculates it in one batch (although I'm not 100% on this).

Does it make a difference if you calculate the results one by one or as one batch?

double-beep
  • 4,255
  • 23
  • 56
PMCA
  • 41
  • 1

1 Answers1

1

What I wanted to know was does it make a difference if you calculate the results one by one or as one batch?

According to Professor Mark E. Glickman -

The Glicko-2 system works best when the number of games in a rating period is moderate to large, say an average of at least 10-15 games per player in a rating period.

Note the above link also gives Glickman's algorithms for Glicko 2

Brian Towers
  • 92,895
  • 10
  • 227
  • 372