The ICC all-time player career ratings

[TestMatch]

The ICC Player Rankings are the "official" guide to the relative merits of batsmen, bowlers and all-rounders in Test match and ODI cricket. They take into account bowling quality, opposition rankings and pitch conditions. Currently, for example, Joe Root and Steyn have a rating of about 900 in tests.

Below is a list of the average ICC CAREER RATINGS of some notable players.

D. Bradman: 855.37
J. Hobbs: 799.00
Brian Lara: 783.55
L. Hutton: 781.00
Gary Sobers: 780.85
RN Harvey: 777.70
H. Sutcliffe: 775.48
K.C. Sangakkara: 762.80
Viv Richards: 762.74
E.D Weekes: 760.50
K Barrington: 747.20
W. Hammond: 745.40
Javed Miandad: 743.12
R. Dravid: 741.79
M. Hussey: 738.47
R.T. Ponting: 736.86
S.M Gavaskar: 736.53
A. Border: 736.00
J.H. Kallis: 734.60
S.R. Tendulkar: 731.86
K.P. Pietersen: 730.80
M.J. Clarke: 727.24
M.L. Hayden: 723.27
G. Chappell: 712.87
G. Headley: 711.77
S. Chanderpaul: 709.19
Jayawardene: 705.22
V. Sehwag: 703.94
Inzamam-ul-Haq: 697.55
A. Kallicharran: 697.03
G.C. Smith: 695.97
R. Kanhai: 688.58
A.N. Cook: 685.58
A.B. de Villiers: 682.40
C.G. Greenidge: 681.91
Mohammad Yousuf: 681.21
H.M. Amla: 670.02
C. Lloyd: 667.29
F. Worrell: 665.39
S. Waugh: 664.57
D. Haynes: 655.39
S. Anwar: 649.11
M. Waugh: 646.71
G. Boycott: 640.95
G. Kirsten: 639.85
A. Flower: 637.92
Hanif Mohammad: 635.42
V.V.S. Laxman: 632.47
Saeed Ahmed: 632.46
A. Stewart: 631.77
M. Atherton: 630.94
C. Walcott: 630.23
M. Slater: 629.92
G. Gooch: 629.70
S. Jayasuriya: 576.27

Click here for more info on how the rankings are calculated: ICC Player Rankings

 

[Howe_zat]

Interesting but much like career averages in the first place, the idea seems a bit raw. I don't think a career rating should be punished for being an ordinary player for a bit in addition to being gun for most of a career, as oppose to just gun for most of a career.

I'm pretty sure this is what's happening in your analysis as it has Sangakkara over Tendulkar, which is the comparison that first got me thinking about this. Tendulkar averaged 40 between the ages of 16 and 22, when Sangakkara didn't play Tests. Between the ages of about 23 and 40 their records are close to identical, but Sangakkara has the better career average, which seems a little backward.

 

[TestMatch]

I'd imagine it's because Tendulkar never broke the 900 ICC point barrier. Virtually all greats have done this, except him. He also never had massive series. Throw in points lost for India losing all the time, and less points for subcontinent tons (probably lots of runs vs low ranked SL in 90s), and you have his ICC score mostly hovering around the 700s.

I don't think a career rating should be punished for being an ordinary player for a bit in addition to being gun for most of a career, as oppose to just gun for most of a career.

Tendulkar had a slow first four years (though he did things as a teenager in international cricket we'd never seen before; some amazing tons whilst in diapers), so I decided to check his ICC rankings during that part of his career. I overlayed it with Lara's rankings during the same period. See:



Tendulkar is actually ranked higher by the ICC at each point. It is only afterwards that Lara's ICC ratings consistently hits peaks, whilst Tendulkar's kind of plateaus out. So the ICC ratings seem to show the opposite of what you're saying. He wasn't really penalized for his first four years, he was arguably out performing guys like Lara.

Interestingly, checking the ICC ODI rankings during the same period also shows a kind of inverse of the test rankings:

As of 31-Dec-1993:
1st rank - 863 ICC pts - B.C. Lara

As of 31-Dec-1994:
1st rank - 826 ICC pts - B.C. Lara

As of 31-Dec-1995:
1st rank - 880 ICC pts - B.C. Lara

As of 31-Dec-1996:
1st rank - 891 ICC pts - B.C. Lara

As of 31-Dec-1997:
1st rank - 883 ICC pts B.C. Lara

Then in 1998, Lara drops in rankings (not coincidentally, he also stops opening and drops down the order), and Tendulkar becomes number 1.

As of 31-Dec-1998:

2nd rank - 886 ICC pts - B.C. Lara

I'm pretty sure this is what's happening in your analysis as it has Sangakkara over Tendulkar, which is the comparison that first got me thinking about this

This is due to 2007 and 2012 I'd imagine. Sanga made massive runs in those years and all those mega tons vs Pakistan probably count as "away game" ICC bonus magic points or something.

I kind of expected Sangakkara to have a higher ICC Career average, actually, given his test average. Still, Tendy's undoubtedly better. I personally never warmed to Sanga. He's in the Dravid/Kallis box for me.

 

[Howe_zat]

I think the ICC points system, even though it's supposed to measure peaks, gives a player some sort of ongoing limiter after a bad start, just like career averages do, though.

After the Trent Bridge test this year Joe Root had a rating of 917. That's because he'd had a brilliant year and a half in which he averaged 80 and hit 6 tons.

But Alistair Cook has never had a rating of over 900, in fact his best is 874. This is despite him having just as good a run in 2010-11 in which he averaged 94 and hit 6 tons and a lot more besides. It seems to be that having the a lot more besides, which should either be unimportant (if you are trying to measure a peak) or helpful (if you are trying to measure a career) actually is counted as a negative in the ICC stats. I can't really get behind that.

 

[viriya]

Averaging current ratings to get a career rating doesn't make sense because that wouldn't give equal credit to each innings of a career due to discounting penalties and skewness that is the result of a single great innings.

For example, say a player has a great innings and his current rating goes up to 900. The next innings he fails and gets 0, but since his current rating was so high it only drops to 850. That 850 in isolation is just riding the coattails of the great innings.

A better way to get a career ratings is to average the rating of each innings of a player - not the current rating. ICC ratings does not give out that information.

 

[TestMatch]

For example, say a player has a great innings and his current rating goes up to 900. The next innings he fails and gets 0, but since his current rating was so high it only drops to 850. That 850 in isolation is just riding the coattails of the great innings.

No, the player averaging consistently with low/medium ICC points (say 600s) will still have a higher ICC career average than a player with a large peak but then successive failures (say 900 pts then ducks).

 

[viriya]

No, the player averaging consistently with low/medium ICC points (say 600s) will still have a higher ICC career average than a player with a large peak but then successive failures (say 900 pts then ducks).

The point is averaging current form doesn't make logical sense. It's even more nonsensical when you realize the ICC rating system is pretty meh with only 4-5 factors.

 

[watson]

Testmatch appears to be right.

A form rating would only look at what a player has done in (say) the last year, whereas our ratings take into account a player's entire career

 

[AndrewB]

There's a flaw in taking the raw average of a player's ratings:

Suppose Player A and Player B (from the same team) have had identical careers to lead to ratings of 500 (and so have the same average rating).

In the next two matches, A scores 50, 50, 0, 0, which (let's say) raises his rating to 520, then lowers it back to 500.
In the same two matches, B scores 0, 0, 50, 50, which lowers his rating to 480, then raises it back to 500.

Then they've obviously still had equally good careers; but A's average rating is now higher than B's.

 

[TestMatch]

Then they've obviously still had equally good careers; but A's average rating is now higher than B's.

But your own example disproves you. They're both are on 500. A is not higher than B.

The point is averaging current form doesn't make logical sense.

I'm not sure what you're saying. Regardless, the ICC algorithm penalizes you for "bad form" and rewards you for "good form". Less points are deducted from your score if you had a run of good scores prior. Similarly, more points are deducted from your score if you have had a run of poor scores. The algorithm does the opposite of what you seem to be arguing.

 

[viriya]

But your own example disproves you. They're both are on 500. A is not higher than B.



I'm not sure what you're saying. Regardless, the ICC algorithm penalizes you for "bad form" and rewards you for "good form". Less points are deducted from your score if you had a run of good scores prior. Similarly, more points are deducted from your score if you have had a run of poor scores. The algorithm does the opposite of what you seem to be arguing.

Is the same innings made after a bad patch and a good patch of form better or worse? I would say the same innings should be rated the same, regardless of whether the batsman was in form or not. This method would overrate the batsman in form because his current rating would be higher by default. Current ratings are fine when comparing *current* form, but breaks down if you just average it out over time.

 

[AndrewB]

But your own example disproves you. They're both are on 500. A is not higher than B.

The average ratings aren't the same, which I thought was what this thread was all about.