Test Cricket Career Ratings

Latest Changes - Batting Ratings - Bowling Ratings - All-rounder Ratings - Unrated Players -

Ratings last updated after Bangladesh v Sri Lanka – 2^nd Test, 3 Apr 2024

Aim of the ratings

The aim is to produce a rating system similar to the ICC Player Rankings, but whereas the ICC rankings attempt to rate players at a point in time, including an element of form, these ratings are intended to assess the players’ Test careers as a whole.

Methodology

The method of producing the career ratings is essentially to take the players’ ICC ratings after each of their Tests throughout their careers and take a weighted average of those ratings, so that each innings should contribute equally to the final rating.

The ICC ratings include a system whereby batsmen only get a full rating when they have played 40 innings, and bowlers when they have taken 100 wickets. Up to this point, they get a progressively-higher percentage of a full rating. This is to prevent players from getting to the top of the rankings too quickly. In producing the career ratings, these percentages are first removed to arrive at what the full ratings would have been, so that the players are given full credit for their early matches; after taking the weighted average the ratings, the percentages are then reapplied to the average for those players who have not qualified for full ratings.

I have taken some shortcuts to reduce the amount of data input required. Up to 28^th January 2007, I have assumed an equal number of batting and bowling innings per Test, by dividing the total number of innings by the number of matches played (note that an innings in which the opposition made 150 or more contributes towards a bowler’s rating even if he did not bowl). This will mean that some innings will actually carry greater weights than others. In the bowling ratings, I have also assumed an equal number of wickets taken per Test up to the bowler’s 100^th wicket. The number of wickets is used to look up the percentages mentioned above, so there will be errors in calculating the full ratings, but these errors will be small for players whose average number of wickets per Test remained fairly consistent (up to 100 wickets if they reached that milestone). For matches after 28^th January 2007, the cumulative number of innings and wickets after each Test will be kept up to date accurately.

The all-rounder ratings are calculated by multiplying together the players’ batting and bowling ratings and then taking the square root. I prefer this method because it gives a range of ratings more consistent with the batting and bowling ratings than the ICC’s method of dividing by 1000 instead of using square roots, as a player with equal batting and bowling ratings would have the same all-rounder rating. However, both methods would put the players in the same order.

Missed matches – a problem to be fixed

At present, the calculations used to produce the career ratings do not take into account the effect that being omitted from the Test team has on a player’s ICC ratings. The calculations treat his ratings as if they were produced from consecutive Tests.

Each time a player misses a Test, for whatever reason, his ICC ratings are reduced by 1%. This means that, generally speaking, the more Tests a player misses, the lower his subsequent ICC ratings will be. This will also result in a lower career rating, which is why Kevin Pietersen and Mike Hussey, neither of whom had missed a Test by January 2007, were almost certainly higher in the rankings than they should be (at this point, Hussey had only played 26 Test innings, so only got 89.5% of a full rating – without this reduction, his rating would have been about 939, not far short of Bradman’s).

The second effect of missing Tests is that the weighting of the player’s performances in Tests prior to the absence is reduced more when calculating later ICC ratings, e.g. if a player plays two Tests, but misses several in between, his first Test performance will have less influence on his ICC rating after his second Test than it would have done if he had played two consecutive Tests. As a consequence, there is a bias in the career ratings of players who missed Tests, with their later performances being given greater weight; the more Tests missed, the greater the bias. The effect of this depends on the relative performances of the player before and after the absence. Two extreme examples are Marvan Atapattu and Matthew Elliott. Atapattu’s first three Tests yielded one run in six innings. After each of these Tests, he was dropped, missing 7, 14 and 15 Tests respectively. When he returned to the side, he was able to achieve a respectable ICC rating of 550, putting him 41^st in the ratings, within 12 Tests and little over a year, despite his batting average still being a very moderate 33, helped by the low weighting of those early failures. Elliott, on the other hand, lost his place in the Australia team in 1999, at which point his batting rating was 466 after 20 Tests. After an absence of more than 5 years, he returned to the side for one Test, made 1 and 0, and finished with an ICC rating of 5. This gives him a career rating of 296.9, whereas without that final Test it would have been 551.4.

I am planning to amend the system to take missed matches into account, but as it will require a considerable amount of additional data to be input, it will take some time.

David North