Cricket Rating Systems

Cricket is a bat and ball sport that probably originated in England more than 300 years ago. It is a game that lends itself to statistical analysis and cricket fans have used these statistics to argue the merits of individual players and teams for many years.

Since no single statistic reliably predicts which players will perform best in any given game nor which teams will perform best in any given competition, various cricket fans have developed more complicated rating systems that attempt to do that (with mixed success). With the increasing popularity of the Internet, many alternative rating systems for both cricket teams and individuals are published and updated regularly by devotees.

Individual statistics and ratings

There are three main areas in which cricket statistics are recorded: batting, bowling and fielding. Raw data such as runs scored, wickets taken and catches held are used a crude measure of skill in those three areas. More refined are the batting and bowling averages.

* The batting average equals the total number of runs an individual scored divided by the total number of times they were dismissed. A batting average over 50 is considered very good.
* The bowling average equals the total number of runs an individual conceded divided by the total number of wickets they took. A bowling average under 25 is considered very good.

One potential flaw in these averages is they don't take into account the strength of the opposition. Another concerns the period over which you calculate the average.

To try to compensate, various people have come up with alternative rating systems. The best known of these are the ICC cricket ratings, developed by accountants Price Waterhouse Coopers and endorsed by the International Cricket Council.

* ICC (separate Test/ODI) - players are rated on a 0-1000 scale. Factors included are runs scored or wickets taken, ratings of opposition bowlers/batsmen, total runs scored in the game and the result of the game. (see [http://www.lgiccrankings.com/icc_php/about-rankings.php website] )

* ESPN STAR (separate Test/ODI) - Like the ICC ratings, a number of additional factors are included in this ratings as well as runs scored or wickets taken. They include the contribution to the overall team effort, the importance of the game, and whether the game was played at home or away. Only results form the past 18 months are included with a higher weighting given to more recent performances. (see [http://www.espnstar.com/cricket/scr/scrdetails.jsp website] )

* Amul (combined Test/ODI) - This is a very unsposhisticated system that simply combines the runs scored (or wickets taken) by a player in both tests and ODIs over the previous 12 months, then ranks them according to most combined runs or most combined wickets. (see [http://www.cricket.amul.com/about.html website] )

* Chula (separate Test/ODI) - These ratings are designed to rate a player's career rather than their current standing in the game. As such they can't be compared directly to the three previous systems. Each performance is given a rating including runs scored or wickets taken; strike rates or ecomomy rates (ODIs only); contribution to the overall team effort; ratings of the opposition; the pitch; milestones reached; and the match situation amongst others. On this basis, for example, the best test innings of all time is West Indian Brian Lara's 153no against England in 1998-99, a long way short of his world record 400no. (see [http://www.people.cornell.edu/pages/cww22/About.htm website] )

Team statistics and ratings

While individual statistics such as batting and bowling averages have been used for over 100 years, attempts to rank teams are a relatively recent phenomenon.

Former Wisden editor Matthew Engel proposed a system in 1995 that ranked all the test teams on the basis of the most recent series performance against various other test teams ( [http://content-aus.cricinfo.com/wisdenalmanack/content/story/153279.html Engel 1997] ).

Since Wisden is widely known as the "cricket bible", the rankings were the subject of much discussion in the cricket world and eventually prompted the ICC to institute a test championship in May 2001, based on the Wisden system ( [http://www.icc-cricket.com/icc/handbook/2005-06/Pages-73-80.pdf ICC Handbook 2005-06] ).

In January 2003, South Africa ascended to the top position after a home series victory against Pakistan. This led to criticism of the existing system because South Africa had lost its most recent home and away test series against Australia and reached number one largely because Australia hadn't played lesser test nations Zimbabwe and Bangladesh.

The ICC reviewed the rankings and decided to institute a new system developed by David Kendix, an actuary and cricket scorer. This system had already been ranking One Day International teams since October 2002 ( [http://www.icc-cricket.com/icc/handbook/2005-06/Pages-73-80.pdf ICC Handbook 2005-06] ).

The stated concept behind the ICC-Kendix ratings is that of a batting average. A team that wins will improve its average while one that loses will see its average decrease. See ICC Test Championship or ICC ODI Championship for more information on how these calculations are made.

The discussions about the strengths and weaknesses of the Wisden and ICC systems no doubt prompted other cricket fans to try their hands at different rating systems. The different systems tend to fall into four categories:

Points-based systems

The simplest rating system is a points-based one, like a football league table. Teams get a certain number of points for wins, draws and losses and they are ranked according to who has the most points. Because cricket series aren't played as part of an organised league, teams invariably play a different number of matches to other teams so rather than use total points for rankings, all the points-based systems use an average of total points divided by games played. These systems include:

* Amul (Combined Test/ODI) - Like Amul's individual ratings, their team ratings simply combine wins, draws and losses from both tests and ODIs in the previous 12 months.

* CEAT (Combined Test/ODI) - bonus points are awarded for away wins and series wins; additional ODI bonus points for winning tournaments; only results from the past 12 months are included. (see [http://www.ceatcricketrating.com/aboutus/pointssytem.htm website explanation] )

* Howstat (Test only) - bonus points are awarded for away wins and series wins; points from series played more than a year ago are reduced by 20% per year to a maximum of five years. (see [http://www.howstat.com/cricket/Statistics/TestRankingsExplanation.asp website explanation] )

Ratings-based systems

Ratings-based systems originated in the chess world and are premised on the idea that a team or an individual has an inherent rating and one with a higher rating should beat one with a lower rating more often than not. If a lower rated team upsets the higher rated one then its rating will increase at the expense of the other team.

Systems can be balanced or unbalanced. In a balanced system, the amount one team's ratings increases is equivalent to how much its opponent's rating decreases so the total number of points remains constant. This isn't necessarily the case in unbalanced systems.

* [http://www.image.co.nz/aqb/cricket_test_ratings.html AQB] (Separate Test/ODI) - A balanced system where the winner always increases its rating and the loser always decreases, except that if the higher rated team wins, the change is small and if the lower rated team wins, the change is large. Home and away is also included in the calculation. (see [http://www.image.co.nz/aqb/about_formulae.html website explanation] )

* [http://www.geocities.com/gwozok Test Cricket Ratings Service] (Test only) - A balanced system based on the Elo rating system used to rank chess players. Calculated on a series by series basis, the system works out the expected result in a test series. If a team performs above expectations, its rating increases at the expense of the other. (see [http://www.geocities.com/gwozok/howworks.html website explanation] )

* [http://home.iprimus.com.au/shane_booth/cricket/ratings.html Shane] (Separate Test/ODI) - An unbalanced system based on the Glicko system, a variation of the Elo system (see above). In essence, it uses a more sophisticated algorithm than Elo, to determine how much each team's rating should increase or decrease after each result. A more established team's rating won't change as much as a team with only a few games under its belt. (see [http://math.bu.edu/people/mg/papers/gdescrip.pdf Glicko description] )

* [http://www.geocities.com/xodicricket XODI] (ODI only) - A balanced system where each team contributes 10% of its rating to a pool for the game. The pool is then divided according to the margin in the game. The effect of this is a lower ranked team performing well against a higher ranked team can increase its rating even if it loses. (see [http://www.geocities.com/xodicricket/about.htm website explanation] )

Modified points systems

Some systems combine elements of points and ratings systems. This includes the official ICC ratings.

* ICC (Separate Test/ODI) - a two-step process where points are awarded for a series of games between two-teams, according to the number of games won, drawn and lost and with a bonus for a series win (like CEAT and Howstat). The series points are then multiplied by a factor determined by the prior ratings of the two teams to determine the number of ratings points each earns. A team's total ratings points are divided by the number of games played to calculate its ratings. (see [http://www.icc-cricket.com/icc/test/rules.html website explanation] )

* Rediff (Separate Test/ODI) - originally called the Rao-Bhogle Index, this system is similar to the original Wisden ratings in that it looks at the last series between each team when constructing its points table. However, it manages to overcome some of the difficulties of that system by using the first batch of results from the grid as an indicator of the strength of each team and feeding that back into the calculations so that victories over strong teams count for more than victories over weak teams. (see [http://www.rediff.com/cricket/2001/nov/01ranks.htm Rediff website] )

* Twenty20rankings.tk (International Twenty20) - A system similar to the official ICC ODI ratings in that points are awarded by multiplying the opposition's rating by a factor dependent on the result. The number of points gained is then divided by matches to give a rating. (see [http://www.freewebs.com/icricket/explanation.htm explanation] ) The site also previously used another ratings-based system, details of which can be found [http://freewebs.com/icricket/tmexpalin.htm here]

Averages-based system

The fourth approach to ratings doesn't necessarily look at the result but looks at the batting and bowling averages of each team. This allows a finer comparison than straight win-loss comparisons but can be prey to distortions if teams have a tendency to win big or lose big. To reuse the football table analogy, it is equivalent to ranking teams on points scored for and against rather than wins and losses.

* Herman (Separate Test/ODI) - This Belgian cricket fan started publishing his Test cricket ratings in 1996, hot on the heels of the original Wisden system, so is one of the longest running systems. Rather than focus on pure averages, he uses an iterative process (like Rediff) to modify the averages according to the strength of the opposition. The modified averages are then combined so a team with a positive rating is better than an average and one with a negative ratings is worse (see [http://users.skynet.be/hermandw/cricket/hermrati.html website explanation] ).

* Date (Separate Test/ODI) - This is a complicated system awarding points for runs scored and wickets taken in each match with bonuses for winning matches and extra bonuses for winning away from home. (see [http://www.public.asu.edu/%7Ekdate/Ratings%20System%20for%20Test%20Cricket.htm] )

Comparison

As of June 2008, here is how the various systems ranked the top ten test teams:

And here is how the top 20 ODI teams are ranked:

External links

* ICC [http://www.icc-cricket.com/test/] , [http://www.icc-cricket.com/odi/]
* AQB [http://www.image.co.nz/aqb/cricket_test_ratings.html] , [http://www.image.co.nz/aqb/cricket_odi_ratings.html]
* Date [http://www.public.asu.edu/%7Ekdate/Ratings%20System%20for%20Test%20Cricket.htm] , [http://www.public.asu.edu/%7Ekdate/ODIRatings.html]
* Herman [http://users.skynet.be/hermandw/cricket/hermrati.html] , [http://users.skynet.be/hermandw/cricket/hermodir.html]
* Howstat [http://www.howstat.com]
* Rediff [http://www.rediff.com/cricket/testrank.html] , [http://www.rediff.com/cricket/odirank.html]
* Shane [http://home.iprimus.com.au/shane_booth/cricket/ratings.html]
* TCRS [http://www.geocities.com/gwozok/]
* XODI [http://www.geocities.com/xodicricket/]
* Amul [http://www.cricket.amul.com/teams.html]
* Ceat [http://www.ceatcricketrating.com/ratings/team.htm]