MintCondition wrote:
The article above is interesting, but statistics are not to be played with lightly!
I thought one of the person's comments in relation to that article were pretty accurate:
Posted by: Tanmay Shukla at October 22, 2010 6:52 PM
I'm sorry to write that, as with most cricket statisticians, there is very little rigour or consideration of confounding effects in your analysis. (1) Instead of leaving out Zimbabwe, B'desh, you should see how they performed in the test compared to their average performance over all tests. This should in fact be extended to all teams. Why throw away information? De-mean it. (2) Half-fixing a problem is often worse than leaving it alone (think Bush). While you account for variation across tests, you ignore variation within. Did the pitch deteriorate over time? Which day of the test was the innings played on? These matter as much as overall Test difficulty, if not more. (3) If you do not de-mean the data, a batsman with great teammates is naturally disadvantaged in your analysis. Think Lara, as an example. Since he's almost always played in a fragile batting order, the total runs scored would always be a little lower than if he'd been on a strong team. This would inflate his stats.
Posted by: Tanmay Shukla at October 22, 2010 6:59 PM
I do not expect you to believe me, but while writing the previous pos I still hadn't seen your final results. Regardless of whether or not you do, point (3) explains very well why West Indian cricketers dominate the rankings and the Australians are lower down.
It also just occurred to me that apart from de-meaning the data, you should also exclude the batsman's scores themselves from your calculations. Otherwise you'll end up attenuating your estimates: run for run, the importance you'll give to scoring, say, 50 runs higher that average will be greater when you've scored 50 runs in total, than say 150 runs in total (because the remaining 100 runs you've scored will also increase the average number or runs scored in the match, thus devaluing your "lead" over other batsmen.
oh right i didnt realise the guy had tweaked it to favour batsmen that score a high percentage of teams runs. thats probably unnecessary. haha and im not the lara fanboy thats mr rogerson

give me laxman or sehwag