Conventional ratings systems such as the Elo inspired official ICC team ratings have two major shortcomings. First, they do not take into account the margin of victory. A team winning a series 3-0 with 3 narrow wins and a team winning a series 3-0 with 3 innings wins are considered to be equally superior to their opponent by the ICC's rating. A corollary to this is the fact that for any drawn match both teams end up with equal points, irrespective of how far each may have ended up from victory or defeat. This is a basic misunderstanding of a 'draw' in Test cricket. In Tests, a 'draw' means that the match was inconclusive. This is different from a draw in say football, in which a draw means both sides scored the same number of goals. Second, they use arbitrary cut offs to determine which match performance will be considered in the rating and to what extent these will be considered. The team performance rating devised here at FormBook is an attempt to overcome both these difficulties. The following example illustrates the basic match calculation:
Consider Australia v South Africa at Perth, 2016.
The match summary can be written as follows:
Australia 605/20 in 1137 balls faced lost to South Africa 782/18 in 1343 balls faced.
The cost of each wicket in this Test was 36.5 runs per wicket.
Australia Raw Points = 605/1137 + 18*36.5/1343 = 1.021<
South Africa Raw Points = 782/1343 + 20*36.5/1137 = 1.216
Win Bonus for the match = 0.5*(1.021 + 1.216) = 1.119
Since South Africa won, the Win Bonus is awarded to South Africa.
Australia Total Points = 1.021
South Africa Total Points = 1.216 + 1.119 = 2.335
Australia's share of the points = 1.021/(1.021 + 2.335) = 0.303
South Africa's share of the points = 1 - 0.303 = 0.697
The match performance is recorded for each match on each match page as follows:
Each player in the South Africa and Australia XIs who participated in this game is assigned this team performance. The rating of any team at the start of any test is the simple average of the rating of all the tests which the XI participating in that Test has played in. Tests in which 5 players of the XI have played in are counted five times.
This approach overcomes the problem of having to arbitrarily set time limits for consideration. The matches considered are the matches in which the players in the current XI have participated. The assumption here is that teams are selected to win. And when they don't win, players get dropped. This approach also overcomes the problem of measuring margin of victory. A team's rating at any given time is a measure of its dominance relative to other teams at that point in time. Here is the history of Australia's Test team rating. as an example.
The approach used to calculate ODI ratings is identical. Only matches against Test playing opposition is considered when measuring a team's rating.