Prince EWS
Global Moderator
We have a couple of threads around for standardised averages and I know I've leaked some of this work into other threads at times, but I've made it into a full-blown project over the past couple of months so I thought I'd start a thread for it.
Introduction
The average - be it batting or bowling - has long been used as the standard measure for one's performance in Test cricket. As well all know, however, it has several limitations on it and research always needs to be made to check just how reflective an average actually is. Variables such as opposition strength, era run-scoring standards, pitch conditions, player longevity and match situation can skew an average one or another.
Standardisation of averages is a process that tends to remove or nullify some or all of those variables, to give a truer reflection of one's worth. Of course, no measure is absolutely perfect and you always have to take everything into context, but the goal of standardisation is simply to provide something more accurate than traditional scorebook averages.
My Method
What I've attempted to do is standardise averages across time and opposition strength, nullifying those variables to fullest extent possible and providing player averages that largely account for the changing run standards in cricket across decades and the differing standards of opposition teams. Of course there are still several other variables within these averages and I don't proclaim this to be a perfect measure, but I do maintain that it's a better measure than traditional averages.
What the method essentially does is compare players to the rest of the world in each decade against each opposition, using a base average of 30.02 (the global Test average, however the base average only changes the numbers rather than the rank of the players for comparison purposes). For example, Australia's bowlers averaged 27.75 in the the 1990s, so averaging 35 against against them in the 1990s would be worth be 20 a standardised average of (27.25 / 30.02 * 35 = 31.77). This has been done for every single bowler in Test history, having broken down their records by decade and opposition. I'm working on an identical measure for batsmen, however it's not quite finished. (EDIT: yes it is)
Possible Improvements Down The Line
Both of these will, IMO anyway, significantly increase the accuracy of the opposition standards I'm measuring. As it stands, for example, a match in 1980 counts for more when determining how good a batting lineup was in 1989 than a match in 1990 does, which is an admitted flaw. Instead of grouping everything into calender decades, the measure of a batting lineup in 1989, for example, will be its average from 1984 to 1994. That will obviously require a lot more data collection.
Introduction
The average - be it batting or bowling - has long been used as the standard measure for one's performance in Test cricket. As well all know, however, it has several limitations on it and research always needs to be made to check just how reflective an average actually is. Variables such as opposition strength, era run-scoring standards, pitch conditions, player longevity and match situation can skew an average one or another.
Standardisation of averages is a process that tends to remove or nullify some or all of those variables, to give a truer reflection of one's worth. Of course, no measure is absolutely perfect and you always have to take everything into context, but the goal of standardisation is simply to provide something more accurate than traditional scorebook averages.
My Method
What I've attempted to do is standardise averages across time and opposition strength, nullifying those variables to fullest extent possible and providing player averages that largely account for the changing run standards in cricket across decades and the differing standards of opposition teams. Of course there are still several other variables within these averages and I don't proclaim this to be a perfect measure, but I do maintain that it's a better measure than traditional averages.
What the method essentially does is compare players to the rest of the world in each decade against each opposition, using a base average of 30.02 (the global Test average, however the base average only changes the numbers rather than the rank of the players for comparison purposes). For example, Australia's bowlers averaged 27.75 in the the 1990s, so averaging 35 against against them in the 1990s would be worth be 20 a standardised average of (27.25 / 30.02 * 35 = 31.77). This has been done for every single bowler in Test history, having broken down their records by decade and opposition. I'm working on an identical measure for batsmen, however it's not quite finished. (EDIT: yes it is)
Possible Improvements Down The Line
- Separation of statistics into home and away samples (EDIT: done now)
- More accurate measures of opposition standards by having figures of each year +/- 5 rather than calender decades. (EDIT: done now)
Both of these will, IMO anyway, significantly increase the accuracy of the opposition standards I'm measuring. As it stands, for example, a match in 1980 counts for more when determining how good a batting lineup was in 1989 than a match in 1990 does, which is an admitted flaw. Instead of grouping everything into calender decades, the measure of a batting lineup in 1989, for example, will be its average from 1984 to 1994. That will obviously require a lot more data collection.
Last edited:
) and scored a gut-full against a good South African lineup. Can't see it topping 35 though.
