cricrate: new cricket ratings website

[Prince EWS]

What makes you change your weightings then?

Developing new ways of calculating things that I hadn't incorporated before and/or actually changing my mind about what's important.

If the results don't make sense, you stay with it because the algorithm is always right? This is a rating system, not some kind of betting algorithm where there is a clear definition of what is correct and what is not.

I suppose I'm not making my point very clearly. All rankings system are essentially an objective application of a subjective framework -- taking what actual followers of cricket can determine better than programming, developing it into a framework and then having automated programming apply that framework consistently in a way our heads won't allow. This takes the best of the tools at our disposal. If you're manipulating that framework just to get the results in a certain order then you're not really taking advantage of the precision or consistency elements of your process anyway, so you might as well just do the results first off the top of your head and then try to come out with a framework that matches them really closely afterwards.

 

[viriya]

Developing new ways of calculating things that I hadn't incorporated before and/or actually changing my mind about what's important.



I suppose I'm not making my point very clearly. All rankings system are essentially an objective application of a subjective framework -- taking what actual followers of cricket can determine better than programming, developing it into a framework and then having automated programming apply that framework consistently in a way our heads won't allow. This takes the best of the tools at our disposal. If you're manipulating that framework just to get the results in a certain order then you're not really taking advantage of the precision or consistency elements of your process anyway, so you might as well just do the results first off the top of your head and then try to come out with a framework that matches them really closely afterwards.

But that's not true though. There are so many innings that I don't agree should be in the top 100 right now that I don't question because the overall framework makes sense to me. But then I am informed by some innings that I consider undoubtedly great (subjective framework) to modify the framework itself.

 

[viriya]

Updated after final Ashes test and Sanga's last match:

- Steven Smith returns to top 5, Joe Root drops out:
cricrate | Current Top Test Batsmen
- Sanga ends at #4 with a similar rating to Tendulkar and Lara, Clarke ends at #54:
cricrate | Best Test Batting Careers
- Ashwin joins the top 10 with his best rating, Siddle returns to top 10:
cricrate | Current Top Test Bowlers
- Rahane debuts in the best fielding career list at #3 (volatile spot since only 10 tests):
cricrate | Best/Worst Test Fielding Careers

 

[weldone]

- Rahane debuts in the best fielding career list at #3 (volatile spot since only 10 tests):
cricrate | Best/Worst Test Fielding Careers

should retire now, otherwise he may miss the '3rd spot in the list of all-time fielders compiled by viriya'

 

[The Sean]

Updated after final Ashes test and Sanga's last match:

- Steven Smith returns to top 5, Joe Root drops out:
cricrate | Current Top Test Batsmen
- Sanga ends at #4 with a similar rating to Tendulkar and Lara, Clarke ends at #54:
cricrate | Best Test Batting Careers
- Ashwin joins the top 10 with his best rating, Siddle returns to top 10:
cricrate | Current Top Test Bowlers
- Rahane debuts in the best fielding career list at #3 (volatile spot since only 10 tests):
cricrate | Best/Worst Test Fielding Careers

Viriya, apologies if this has already been done to death in this thread, but the best batting careers ratings seem heavily biased in favour of the modern players and the sheer volume of cricket they play. While I'm not going to criticise any of the blokes on that list for the outstanding careers they've had, I really struggle with the idea that 10 of the top 15 Test batsmen of all time have played in the past decade or so. Is there something within the methodology that over-weights longevity/no. of Tests criteria?

 

[viriya]

should retire now, otherwise he may miss the '3rd spot in the list of all-time fielders compiled by viriya'

heh tbf I don't have "All-Time" as the heading. But yea it's a little ridiculous because there's only data from 2005 to consider.

 

[viriya]

Viriya, apologies if this has already been done to death in this thread, but the best batting careers ratings seem heavily biased in favour of the modern players and the sheer volume of cricket they play. While I'm not going to criticise any of the blokes on that list for the outstanding careers they've had, I really struggle with the idea that 10 of the top 15 Test batsmen of all time have played in the past decade or so. Is there something within the methodology that over-weights longevity/no. of Tests criteria?

This is an issue that was brought up before and for career longevity it's based more on the number of years you play instead of the number of tests (if you play in the 60s for 15 years and 80 tests without missing any that would be equivalent to 15 years now with 135 tests). It still seems to be underrating pre-2000s players like you said, and there is one "bug" I thought of while reading your comment:

I discount player current ratings till a plays his 40th test, and so for batting ratings the bowler quality factor would generally be lower for pre-2000s matches because of the lower chance that a player gets to play 40 tests.. This applies in the other way around for bowler ratings. I'll look to fix this asap and update when I've made the changes.

 

[mr_mister]

well i like your site viriya. i admire the effort you put into it

 

[Daemon]

Yeah people have been unfairly dickish to viriya. His rankings do look very off and he has some (imo) odd views which he's stubborn about...but it doesn't really warrant some of the **** he cops.

 

[mr_mister]

yeah ive decide i really hate this lillian thompson bloke, its a pity he seems smart but so unnecessarily cruel

 

[viriya]

Yeah people have been unfairly dickish to viriya. His rankings do look very off and he has some (imo) odd views which he's stubborn about...but it doesn't really warrant some of the **** he cops.

What are my odd/stubborn views if I may ask? Anything you think I'm doing obviously wrong?

 

[weldone]

I discount player current ratings till a plays his 40th test, and so for batting ratings the bowler quality factor would generally be lower for pre-2000s matches because of the lower chance that a player gets to play 40 tests.. This applies in the other way around for bowler ratings. I'll look to fix this asap and update when I've made the changes.

You should lower this to something like 15 tests; and over and above that you should measure this by longevity, which you already do by number of years iirc [As I have always said, sample size and longevity are 2 separate issues and mustn't be mixed]. And while measuring quality of opposition, you should ignore longevity.

 

[weldone]

yeah ive decide i really hate this lillian thompson bloke

Ok then I have also decided that I really like this mr_mister bloke.

 

[viriya]

You should lower this to something like 15 tests; and over and above that you should measure this by longevity, which you already do by number of years iirc [As I have always said, sample size and longevity are 2 separate issues and mustn't be mixed]. And while measuring quality of opposition, you should ignore longevity.

I'm exploring options here, and yea 40 tests is too long for early era cricketers clearly.

I do think longevity should be a factor in measuring quality of opposition though - otherwise a bowler who debuts with a 10-fer would sky rocket to the top of the pile.

 

[NUFAN]

Siddle into the top 10 where he belongs. Where does his bowling performance in the 3rd innings at the oval rank, top 50? seem to remember him making hurricanes top 50 bowling performances.

 

[weldone]

I'm exploring options here, and yea 40 tests is too long for early era cricketers clearly.

I do think longevity should be a factor in measuring quality of opposition though - otherwise a bowler who debuts with a 10-fer would sky rocket to the top of the pile.

Sample size should be a factor in measuring quality of opposition, but not longevity. There maybe some doubt about whether scoring runs against Bob Massie (low sample size) is the toughest job in the world; but there can be no doubt that scoring runs against Shane Bond (low longevity, but sample size is above the cut-off) is very difficult. However, that doesn't mean Shane Bond had as good a bowling career as Richard Hadlee (great longevity) - but scoring runs against both of them were mighty difficult. It was difficult against Bond for a few years, but for well over a decade against Hadlee.

Cricket ranking is all about sound logic.

 

[viriya]

Sample size should be a factor in measuring quality of opposition, but not longevity. There maybe some doubt about whether scoring runs against Bob Massie (low sample size) is the toughest job in the world; but there can be no doubt that scoring runs against Shane Bond (low longevity, but sample size is above the cut-off) is very difficult. However, that doesn't mean Shane Bond had as good a bowling career as Richard Hadlee (great longevity) - but scoring runs against both of them were mighty difficult. It was difficult against Bond for a few years, but for well over a decade against Hadlee.

Cricket ranking is all about sound logic.

I gave an extreme example, but you see some bowlers have one great year on debut, and I think just 10-15 tests should be discounted. 40 tests is a bit much but I would say a player is not established till 2 years in his career.

 

[viriya]

Siddle into the top 10 where he belongs. Where does his bowling performance in the 3rd innings at the oval rank, top 50? seem to remember him making hurricanes top 50 bowling performances.

His 4/35? It's not rated anywhere near top 50 all-time if that's what you mean.. It was only 4 wickets and only Cook was an established batsmen of the wickets. Maybe you're talking about a different performance?

 

[NUFAN]

What what what ONLY 4 wickets, only 4? 4 of the best from Pete, what a performance.

 

[viriya]

You should lower this to something like 15 tests; and over and above that you should measure this by longevity, which you already do by number of years iirc [As I have always said, sample size and longevity are 2 separate issues and mustn't be mixed]. And while measuring quality of opposition, you should ignore longevity.

Actually it was 20 innings, not 40 tests - I was confusing the current rating discount to the career rating discount. That said, a hard 20 innings for all players is unfair for earlier era players who had less of a chance to play tests in general. I'm currently pushing through a change that lowers that threshold for pre-1990s era players with some interesting results.