Implementation of the rating systems to rank sports teams.

• Elo
  • logistic (logistic decay of score difference)
  • power (power decay of score difference)
  • fte (FiveThirtyEight calculation)
• Glicko

If we are interested in just running the ratings for a specific matchup, we can call the rating objects:

### initialize rating with parameters
elo = Elo(h=70)
glicko = Glicko(c=42.43, sigma_min=30)

Additionally, we can also run the ratings for historical data, stored as pandas DataFrame:

## using Elo:
elo = Rate(model=Elo, season_regress=0.33)
elo.rate_fixtures(train=fixtures)

## using Glicko:
glicko = Rate(model=Glicko, c=10)
glicko.rate_Fixtures(train=fixtures)

You can plot the ROC curve for one or multiple models:

plt.figure(figsize=(8,8))
elo.plot_roc_curve()
glicko.plot_roc_curve()
plt.show()