Methodology



  • Data


    • ODI scorecards from 2005 to date are parsed to rate every win shares performance.

    • Scorecard information is sourced to generate an accurate assessment of each player's performance.

    • Match results are used to generate team ratings. Peak ratings are identified by checking if the rating before and after the particular match considered is lower.

    • Fielding performances are evaluated by parsing detailed text commentary. More detail here.

    • Win shares are calculated by attributing over-by-over team match odds changes to the appropriate player.

    • Players lose 1% of their live rating if they miss a team match. This is to ensure injured or dropped players do not stay high in the ratings without playing matches.

  • Ratings


    • Current Best/Worst ODI Win Shares:
      This is evaluated using a weighted average of a player's win shares. More weight is given to recent performances with past performances diminishing in value with time. Unestablished players that have played less than 20 matches are penalized to avoid cases where a few highly rated matches can catapult a newer player to the top of the pile.

    • Best/Worst ODI Win Share Careers:
      This evaluates overall careers by calculating the average match win shares of each player, with a bonus given to career longevity (where playing the most number of years while missing the least percentage of matches gets the most credit) and a penalty for careers of less than 100 matches (increasing penalty for lower matches played).

    • Best/Worst ODI Win Share Performances:
      This is a list of the highest and lowest rated win share match performances.









  • Performance Factors


    • Odds
      Over-by-over match odds are calculated by using similar historical match situations and the corresponding win/loss frequencies of those matches. For adjusted win-shares, the starting match odds are adjusted to factor the team strengths (Aus vs Afg will not be 50-50 but more like 75-25 favored to Australia for example).

    • Batting WS
      Batting Win Shares are attributed over-by-over based on the number of runs added by each batsman when win odds for the batting team increases, and the number of balls faced by the batsman when the win odds decrease. The likely win odds decrease when a batsman gets out is also attributed to the corresponding batsman.

    • Bowling WS
      Bowling Win Shares are attributed to the bowler over-by-over based on the win odds change for the bowling team. Fielding events such as dropped catches and missed stumpings are positively credited to the bowler since he created the chance while the odds change from a great catch is shared with the relevant fielder so the bowler does not get full credit.

    • Fielding WS
      Fielding Win Shares are attributed based on fielding events that affected the game. A direct hit or a great catch would be mainly attributed to the fielder instead of the bowler, and dropped catch is also considered as "lost" win odds.