Sunday, July 20, 2008

Statistics, where does it lie?

Consider that a team has lost 5 wickets, where is the final score likely to lie? The last 5 wickets are likely to add only ~113 runs (110-116) irrespective of the starting scores (e.g. 300/5, 400/5, 500/5, 600/5). Strange but true ...

Charles Davis, on Cricinfo's It Figures, does a fine statistical analysis of the likely final score for an unfinished innings!

Which would be a bigger final score, SL's 952/6d or WI's 790/3d? SL might've reached 1028 vs WI's 996!! WI's chances of exceeding 1028 is 24% ...

Can this technique be applied to 'not out' innings as well? We could possibly determine Sachin's final score for 248* but it would be tricky for 400*, I guess. Would the averages be affected drastically?

Who says that cricket is just aggression? In this case, it is regression ...

3 comments:

Veetrag said...

We can statically predict in cases where questions such as 'What will be India's average total in next 20 matches?' but to predict score for one game would be very risky. Reason being, statistics works best when we want to extrapolate some results in cases where averages. Its because in a set of innings the deviation from average will be canceled out, while not in one match.

rb stumped said...

Statistics can predict the outcome for even a game (with likely deviation) and the confidence in the prediction too! e.g. WI would've overtaken SL's 1000+ runs with a 24% likelihood.

ray said...

Hi ,

I was reading ur blog posts and found some of them to be very good.. u write well.. Why don't you popularize it more.. ur posts on ur blog ‘Stumped!’ took my particular attention as some of them are interesting topics of mine too;

BTW I help out some ex-IIMA guys who with another batch mate run www.rambhai.com where you can post links to your most loved blog-posts. Rambhai was the chaiwala at IIMA and it is a site where users can themselves share links to blog posts etc and other can find and vote on them. The best make it to the homepage!

This way you can reach out to rambhai readers some of whom could become your ardent fans.. who knows.. :)

Cheers,