Tuesday, May 15, 2007

Bev or Viv?

Today’s issue of Mint (the financial newspaper of the Hindustan Times group, in association with the Wall Street Journal) carries a story on how two actuaries have come up with a new system to calculate batting averages of batsmen in limited over internationals.

The most significant change that has been introduced is the treatment of unbeaten innings. The two actuaries Sanchit Maini and Sumit Narayanan argue, and rightly so too, that adding to the numerator but not to the denominator is not quite fair. Sanchit and Sumit use insurance principles like risk and exposure to play with the calculations.

“According to the new method, if a not-out batsman has played fewer balls than his own average (number of balls faced on an average in his career) up till that game, it is not counted as a not-out. For instance, if a cricketer has played 10 innings, facing an average of 41 balls in each innings, and remained not-out in four, but faced at least 41 balls in only one of these four innings, then he is counted as having been out in three of his ‘not-out’ innings. His average will fall because his score in this 10 innings will now be divided by nine (the number of innings in which he is considered out), and not six as it would have otherwise been.”

It doesn’t sound as complex as Duckworth & Lewis, for sure, so that’s a start. But if the ICC does take it up and starts adding more variables to it, you never know what may come out.

However, the question really is this: Do we really need different statistics to evaluate batsmen? Michael Bevan’s average is much higher than that of Sir Vivian Richards, but does even a rabid Australian think Bevan was better than Sir Viv? Sure, Sanchit’s and Sumit’s formula puts this pecking order in place, but isn’t that retro-fitting?

Any way, let me not quibble. The fun in cricket analysis lies in looking at the game through different lenses – and the number lens is perhaps one of the more interesting. Good shot, Sanchit and Sumit. Now if only you could find a useful way to measure fielding averages…

No comments: