Stats and the 'truth' !!!

For all of those with a keen interest in cricket, and especially of cricket facts and figures, the blog 'It Figures' on Cricinfo (link here) offers more than just food for thought. For those who have not yet checked it out, I suggest you stop reading further and click on the above link. An excellent blog by some of the most famed cricket-lovers (and number-crunchers !!).

More than anything else, it brings out the dichotomy between figures and judgement i.e. statistics hide more than they reveal. Of particular interest is the latest post on the blog, which talks about the most 'consistent' bowlers in test cricket history. Now, putting a quantitative measure of the word 'consistent' is itself daunting. Statistically, consistent would mean a distribution in which maximum no. of points are near the median. Translated into cricketing terms, it would imply a batsman who mostly has scores close to his test average (and who is therefore deemed more consistent than, say, a guy who averages about the same but is more of the '100-5-2-150-5' variety). When it comes to bowlers, things are not that simple. So what the writer of the blog has done is to come up with a measure of bowling consistency. What it implies (read the blog to understand it fully) is that a bowler who most often gets a wicket in an average spell (7 overs for a medium-pacer and 11 overs for a spinner) is more consistent than a bowler who bowls 3 wicket-less spells and then gains a bucketful of scalps in his next spell (even though, both the guys end up with similar bowling averages). It is certainly quite interesting and has invited a host of comments from readers (ranging from appreciative to dismissive and a few downright silly - mostly from readers whose favourites could not find a high-enough position on that list).

All in all, a quite brilliant blog in that it exposes us to how our judgement on who the best is might change when subjected to science.

May the blog live long !!!

Cheers
Amit