Friday, 16 January 2009

Sachin Tendulkar, nil points

'Selected on a basis of ranking points by the jury'. 'Our rankings take into account a player's entire career'. 'Not a measure of greatness over a career'. 'Sustaining excellence over a long period'. 'Based on detailed data analyses of the year's performances'.

Ah, statistics, those noted chums of lies and damn lies. Today we have had Cricinfo's Awards, based in part on the first and last statements in the paragraph above, and the ICC's Reliance Mobile Best Ever Test Championship Ratings, calculated according to the others (and never mind the apparent contradictions). 

They have dispatched the following information: Virender Sehwag and Dale Steyn gave the Test performances of the year, Shiv Chanderpaul was the Batsman of the Year, Sachin Tendulkar is the 26th best Test batsman of all-time, some 23 places below Ricky Ponting, and Shane Warne has been comprehensively outbowled by Derek Underwood and Tony Lock.

It seems trite to say that some things are quantifiable and others are not. Cricket is a comparatively measureable sport; that's one of its great pleasures. The accuity of the Cricinfo Awards and the patent absurdity of the Reliance Mobile Best Ever Test Ratings (almost too tempting a name, isn't it?) have made for an entertaining day.

But they open a wider and more enduring point, and that's the role played by aesthetics and beauty. To draw a simple, football-based analogy, would David Beckham have lived the life of David Beckham if he had looked like Paul Scholes? 

Life rewards beauty. And beauty distorts statistics. Who was the better batsman, David Gower or Matthew Hayden? The stats are conclusive, the emotions less so. Would you rather watch Mark Waugh or Steve? Gooch or Boycott, Lara or Border, and so on, forever. Today, stats have proven a blunt tool. 







14 comments:

David Barry said...

Who was the better batsman, David Gower or Matthew Hayden? The stats are conclusive, the emotions less so.
The stats are not conclusive. On my adjusted averages Hayden's 43.6 and Gower 42.7. Those are too close to be conclusive - you really need gaps of about 5 runs (over a long career) before you can say which had the higher "true average" with, say, 95% certainty.

The Old Batsman said...

David - that's interesting, I didn't realise they'd be as close. Some loose phrasing by me - should have said something along the lines of 'the headline stats' or 'the averages would appear conclusive'.

What I was driving at more was how the way someone plays sometimes affect judgement of them - Allan Border might be a better example.

BTW do you have a table of adjusted averages up? Might be interesting alongside the Test ratings etc

David Barry said...

This is a list that is a few months out of date but is otherwise complete.

I have occasionally thought that I should have a regularly-updated list on my blog but I have never got around to making such a thing.

nestaquin said...

Dave's list has Tendulkar 33rd. Which is below the recent ICC list although I do recognise that they are measuring different aspects. Still it's around the same mark which is surprising and of interest.

The Old Batsman said...

An update to run alongside the rankings would be great, I think. Obviously the adjustments favour older players, although I don't know if they take in intangibles like standards of fitness and fielding, travel, having to swap often between limited over and test match cricket etc.

still, Sachin coming so low on both lists should stir up some opinion...

David Barry said...

I haven't really worked out what's going on with the older players doing so well on my adjusted batting averages. The same is not true to the same extent with bowlers - the top of the list goes Barnes, Lohmann, Davidson, Marshall, Ambrose, Turner, McGrath.

I have wondered if it is because of greater depth in batting these days. ie, in the olden days you had both more very good batsmen and more mediocre batsmen, whereas today the spread in talent is smaller, so the best players don't stand out from the others like they used to.

But why that phenomenon should only apply to batsmen I don't know.

Nesta, Tendulkar's low position on my list is the one thing that has really puzzled me. I think that it is caused by two factors: the general bias towards great older batsmen as guess-explained above, and his general decline since ~2000. At the end of 2000 he had the highest adjusted average of any modern player, though still only #17 all-time.

The Old Batsman said...

David, I wonder if, along with greater depth, it's to do with shorter two and three test series, increased travel and swapping between formats.

If you think about the structure of old-style tours, they were effectively playing a full season's cricket while on tour, with lots of first class innings etc. They must have been more settled.

Or maybe modern batsmen are just rubbish. Especially that Tendulkar...

Leg Break said...

I've obviously missed something.

How does the Adjusted Average thing work?

12th Man said...

Do the adjusted averages take the opposition bowling into account? i'm sure the batsmen who scored against the mightly Windies attack would do well then.

Leg Break said...

12th.

If that was the case, then Allan Border would rate very highly..

David Barry said...

Runs are weighted by the quality of the attack, which is defined as the "average average" of the bowlers in the innings. That "average average" is weighted according to how many balls each bowler bowled.

That is the only factor. For something so simple I find that it works remarkably well to adjust across eras and compare how batsmen did against quality opposition.

Ideally you'd use ball-by-ball data but that only goes back to 2000ish.

Leg Break said...

Good work.

One minor suggestion though: I assume the average of the bowler is taken over their entire career rather than at the time of the match?

For example, a century scored against Mitchell J would be worth less a year ago than today.

And I realise I’m asking for the impossible,…

David Barry said...

Yes, whole career averages are used, so each time I re-calculate the averages the recent scores change a bit. I could probably do career-to-date but then you'd get more noisy stuff happening with bowlers early in their careers. eg, when Michael Clarke averaged 2.5 or whatever, if he bowled 10 or 20 overs that would skew the scores upwards far more than would be justified.

The Old Batsman said...

'noisy stuff' - i like that expression!