Wednesday, 1 June 2011

Man of the Match: Robotic Assessment?

It's about time we talked a bit more generally about WPA (we've promised to present a bit more about how the model works) but here's a quickie on a different topic: could WPA be used to choose the man of the match? Here are some assorted thoughts on this, i'm sure people have their views, and we'd love to hear them (here, on Twitter or by email, cricketwpa@gmail.com).

If you're of the school of thought that avoiding a loss is as important as earning a win, the player with the highest WPA - LPA should be the man of the match. This is probably a good way to elect the player of a series, but what about if a win is paramount (say you're 2-1 down and it's the last match of a series- there's little difference between drawing and losing)? If the team that's behind in the series wins, maybe their player with the highest WPA should be MotM; if the other team wins, maybe it should be their player with the lowest LPA?

Now of course this is all theory and assumes that our program is perfect which it isn't- ideally we'd have to run it with 'predicted overs left in the test match' (as opposed to the maximum overs remaining) as perceptions of time remaining can be very important. For example, if rain is predicted in day 5, wickets are more important on day 4 than our model would say. We also assign all the fielding team's WPA/LPA/DPA to the bowler, which neglects both subpar fielding and excellent fielding. We suppose, however, that it would be fairly straightforward to assess potential MotM candidates' fielding efforts. A stunning catch's WPA is easy to look up, as could be the would-be WPA swing for a dropped catch...

[As a relevant side-note here, we should note that we were remiss in our construction of the WPA tables from the Cardiff Test- not only did we forget to penalise Prior and Jayawardene for byes, we also forgot Trott's extremely flukey run out of Maharoof, worth 6.6% of an England win, and -2.8% of a Sri Lanka win! We do normally note run outs on a separate line...]

More generally, is WPA the right tool for MotM assessment? Possibly, but you'd have to accept that it does leave wide open the possibility that an 'important' wicket or two late in the game could hand you the honour, whilst a double century at an 'unimportant' time could leave you feeling unfairly treated (the point of WPA!). Having said that, how often have players taken 5 wicket hauls consisting mainly of tailenders? And how often has an impressive hundred come when a game is well on the way to being drawn? For example, Trott's double hundred did a huge amount to save England from losing, but a strike rate of under 50 in a rain affected match isn't the way to play to win the match!

In WPA terms, the Test comes down to to this- England went from being behind in the game to winning it. Trott and Cook took away half a Sri Lanka win, Swann then snatched half an England win by himself. He's therefore the man of the match. Of course, you could easily say that without Tremlett, Swann wouldn't have been able to be in a match-swinging situation. You'd be right to point this out, but he was there and he really did do the business economically and quickly, when no less than that could win England the game.

No comments:

Post a Comment