PLAYER INFLUENCE RATINGS
The player influence ratings are a measure of the influence of the player on the match. They do not consider the result of the game. They are calculated for each match for batting and bowling, and not for each innings. They are listed as "MR" in the player tables.
The influence rating for Test and ODI batting considers four factors:
1. The runs scored (r)
2. The runs scored at the other end while the player was at the wicket. This is calculated as the ratio of the player's runs per dismissal to the runs per dismissal at the other end. (rpw_ratio)
3. The player's run share. This is calculated as the player's share of the runs scored by the team. (rsh)
4. The quality of the opposition bowling (this is based on the formbook model) (opp_bowl)
Note: The run share (rsh) for the ODI batting is the ratio of scoring rates for player and rest of the team. Since opp_bowl for ODIs is given in terms of runs per 300 balls, the factor of 10 is applied.
test_batting_influence = r*rpw_ratio*rsh/opp_bowl odi_batting_influence = 10.0*r*rpw_ratio*rsh/opp_bowl
The influence rating for Test and ODI bowling considers five factors:
1. The wickets taken (w)
2. The opposition batting strength (based on the formbook model) (opp_bat)
3. The player's share of the teams bowling (bowl_share)
4. The player's share of the team's wickets (w_share)
5. The player's share of runs conceded by the team (rc_share)
bowling_influence = w*opp_bat*bowl_share*w_share/rc_share
The T20 batting and bowling rating involves 5 factors:
1. The average score for a batsman (for the batting rating) who plays the number of balls in the innings in question (exp)
2. The projected number of runs the batsman would score in 120 balls for less than or equal to 10 dismissals (proj_r)
3. The opposition bowling (for the batting rating) or batting strength (for bowling rating) given by the formbook model.
4. The number of runs the bowler (for the bowling rating) would concede in 24 balls (proj_r)
5. The balls faced by the batsman (bf) or the balls bowled by the bowler (bb).
t20_batting_influence = bf*(proj_r/opp_bowl)*(proj_r/exp)
t20_bowling_inflence = bb*(opp_bat/proj_r)*(exp/proj_r)
T20 games are better measured using overwise data. Overwise data is available for about 4800 out of 7400 T20 games as of October 2018. A Runs Above Average Replacement (RAAR) measure is developed for T20 games using this over level data. The details for this are available here.