Tuesday, 14 December 2010

Adelaide: What about the last day?

Thanks to Steve Busfield at The Guardian for drawing attention to our WPA statisic in his Ashes blog.

One criticism of our WPA analysis of the Adelaide Test was that the graph did not reflect how tight the game was on the last day, when rainy weather was imminent and a draw was looking very likely.

I should explain why this is.....

The formula we use assumes that all of the remaining scheduled overs will be played.  Our reason for doing this is so that we can quantify how much the bad weather contributed to the final result.  For example, if there is an unscheduled stop in play due to bad weather, the probability of a win might jump to a lower value.   This change in win probability is attributed to the weather itself, rather than to any of the players.

However, Adelaide was a bit different because everybody knew that the rain was on its way on the last day, and England probably needed to take all the remaining wickets before lunchtime.

So let's do it a bit differently.

Here is the analysis of Australia's second innings at Adelaide if we assume that only 35 overs could have been played on the last day:

And here is what it did to the players' overall numbers (showing how each player has changed the probability that his team wins / loses / draws):

            Aus win     Aus lose     draw
Watson       -2.0%       -9.9%       11.9%
Katich       -6.7%       12.6%       -6.0%
Ponting      -7.5%       27.0%      -19.6%
Clarke       -7.3%      -10.7%       17.9%
Hussey        7.0%       -9.0%        2.0%
North        -4.8%        7.9%       -3.2%
Haddin        6.1%       16.2%      -22.2%
Harris       -5.0%       19.9%      -14.9%
Doherty      -8.7%       13.8%       -5.0%
Siddle       -9.9%       10.8%       -0.8%
Bollinger    -2.5%        2.9%       -0.4%

            Eng win     Eng lose     draw
Strauss      -8.5%        9.6%       -1.1%
Cook         14.0%      -15.4%        1.5%
Trott         8.3%       -8.8%        0.5%
Pietersen    21.6%       -6.5%      -15.0%
Collingwood  -8.8%        1.0%        7.8%
Bell          0.8%       -0.1%       -0.8%
Prior         0.7%        0.0%       -0.6%
Broad       -19.9%        3.7%       16.3%
Swann        -1.2%       -0.4%        1.6%
Anderson     43.8%      -18.7%      -25.2%
Finn         12.9%        5.6%      -18.6%
run outs     16.5%       -9.9%       -6.6%

In the new analysis, James Anderson is even more of a hero than before, now contributing 43.8% of a win to England.  Anderson took the wickets of Brad Haddin and Ryan Harris at a crucial point in the game when only 23 overs could be played before the bad weather was due.  These wickets finally put England in the commanding position, and so gave Anderson a huge boost in "win probability added".

In the new analysis, Ricky Ponting is even more of a villain than before, now contributing 27% of a loss to Australia.  With bad weather coming, could Ponting have avoided losing the game if only he'd stuck around longer than 19 deliveries?

Why does our WPA analysis hate Graeme Swann?  He took five wickets in Australia's second innings, but barely contributed to England's win at all!

The answer is in the number of overs he bowled in the last innings: 41.1, which is almost twice as many as any other English bowler.  By bowling more overs, Swann was more responsible for allowing the Australians to stick around at the crease, thereby making a draw more likely.

On top of that, two of Swann's wickets were of tail-end batsmen at a point in the match when England's win was very nearly safe in the bag.

No comments:

Post a Comment