Who contributed the most in terms of wins in the Test series?

[Cricketwpa]

Hello- from a long time reader/first time poster... Just wanted to post briefly that, having analyzed the Ashes, we at cricketwpa (cricketwpa.blogspot.com) have been doing a retrospective on the India v SA series so as to determine who contributed most to their team's chances of winning/losing/drawing the test matches. We've been working hard to make a model that evaluates players in a fair manner (and is NOT intended to be applied to betting), and we hope we're just about there! It would be great to see what you guys think, I promise it's not a scam site (it's on blogspot for goodness sake!) and we don't even have adverts. We just hope cricket fans are interested, and given that India is the world capital of Cricket, and there is a lot of good statistical chat on this site, we thought we might post here! And we don't have a forum on our site, so perhaps we will be able to generate some discussion on the ICF messageboard too... If there's enough interest, we will continue to analyze India games (otherwise our readership is mainly in the UK so we may instead catalogue historic England games instead)... Any thoughts people? All the best, Henry (from cricketwpa)

[rohitsilhi]

how do you calculate these figures?

[Cricketwpa]

Hey Rohit! The model works by assessing, in each game state encountered in the match, how a generalized (we call it generic) team would be expected to perform against another generalized team. It takes into account parameters such as run environment, overs remaining (always assuming that rain remains at bay; overs lost to rain or bad light appear as a discontinuity) and of course the current lead/deficit and the wicket resources left to each team (some wickets being more valuable, and some being less so). Basically it has a bunch of 'probability distribution functions' for all the partnerships that could potentially be formed in the match and combines them in such a way that it is able to assimilate all the ways it believes the test can work out, weighted by probability. Some of these scenarios exceed the match time limit, and so are assigned to draws, the others are a win for team A and a win for team B. It is important to note that the win probabilities shown are those for a generic team, not for the team involved (see the section called 'what do we mean by win probability' for clarification of this and why it's not the same as betting odds). When the 'win probability' changes, eg by Harbhajan dismissing Smith, and Harbhajan adds to his team's win probability, he adds to his WPA, and Smith earns negative WPA. A positive WPA over the course of the match means he's contributed more to the game than a typical player of his type (e.g bower/wicketkeeper/batting allrounder etc) would have... There are quite a few finer points about how the distribution functions are formulated for the 'generic' teams and how the scenarios are assessed etc that I would go into, but am afraid of making it too technical and inducing sleepiness but I do hope people like the general concept, the pretty graphs and, of course, the WPA stats!

[rohitsilhi]

in theory the mathematical model describe above seem to be well designed but in practice i think there are many limitations to its use. e.g. harbhajan got 4-10 in the 3rd innings of third test and if the seamer bowler had finish off the job, then india would have surely won! as a result of his 4-10, he gave india a high probability of winning, but according to your calculation he decrease india chance of win by 8.8%? how can this be so?

[punjabi_khota]

^ I would guess that he earned big points when he took the 4 wickets at top, but then gave away 120 runs which earned even bigger negative points, hence making his total negative...

[Cricketwpa]

Indeed. His match figure was -8.8%, which is made up at follows: 1st innings bowling performance (0-75 off 27 overs) worth -21.4% 1st innings knock of 40 worth +7.2% In his second innings bowling, the wickets of Smith, Harris, Petersen and Amla were worth 14.6%, 5.5%, 14.1% and 17.3% (a total of over 50%) but his innings total ended up being only +5.3% after he gave up a total of 120 runs over 38 overs (albeit taking 3 lower-order wickets). Because of these 120 runs, and his wicketless first innings effort, Harbhajan finished the game in negative WPA territory. This is a shame, though, because I think very highly of him! I suppose if the seamers had polished off the lower order, he wouldn't have ended up giving up so many runs at the end; but that didn't happen, he gave up those runs, and by the end of that innings, South Africa were in the dominant position...