 | « Back to article | Print this article |
Angry and frustrated after the Duckworth-Lewis system cost them victory over the West Indies in the Twenty20 World Cup in Guyana on Monday, England captain Paul Collingwood has demanded that rain rules be revised for the shortest format of the game.
Certainly, he is justified, as England had set a mammoth 192-run target for Chris Gayle's men, but a rain disruption that washed out over an hour of the match led to the target being revised to 60 in six overs, which the hosts chased down easily and won by eight wickets.
The defeat raises some questions about the rain rule. Srinivas Bhogle examines them, and comes up with some suggestions:
What is the probability that you'll win a T20 match after scoring 191 batting first?
I'd agree with Paul Collingwood. It must be something like 95%, especially with long boundaries and a sluggish outfield.
So why did England lose?
Because it rained, the Duckworth-Lewis (D/L) method had to be used, and D/L said West Indies (WI) only needed to reach 60/2 in 6 overs to win.
Isn't that kind of low?
It is. And I heard commentators saying that the West Indies would win at 43/0 in 5 overs or even at 46/1 after 5 overs.
This seems blatantly unfair
It is. No wonder Chris Gayle gleefully opted to field after winning the toss.
The side batting second therefore has a distinct advantage
It does. In fact we're back to the good old (pre-1990) days when ODI targets were decided only based on run rates, without considering wickets lost, and every captain winning the toss chose to field first on a cloudy day.
But isn't that the raison d'être of D/L ... that D/L considers both 'overs remaining' and 'wickets lost'?
Yes, indeed. In fact D/L also claims that their method is toss-invariant and fair to both teams at all times in the match. And to be fair it is ... certainly in ODI games.
So what's going wrong?
They're trying to force-fit a 50-over model for a 20-over game and it isn't working!
Why is no one complaining?
I thought Paul Collingwood spoke tough after Monday's game, but D/L has now attained a hallowed status ... so everyone is shy to say that this T20 emperor has no clothes.
Do you really believe that this emperor is prancing around naked?
Frankly, yes. I know that everyone's saying that D/L needs more T20 data to fine-tune their model. But we now have enough data! We've played 175 IPL matches now and that's a reasonable amount of data.
So why aren't we seeing a new D/L format adapted for T20?
I can only guess that it's not working out! T20 may be a different sort of animal. Perhaps Duckworth and Lewis just can't find those ten 'well-behaved' curves that need to drop rapidly from a resource percentage of 100 (or less) to 0 in just 120 steps, instead of 300.
I'm sure I didn't understand that! But are you saying that the D/L rationale itself may not work in T20?
Yes. So far the D/L horse is proving to be rather unruly with T20. Why flog it into submission and put up with tantrums that deny England an almost certain win? Why not try some other way?
You mean dump D/L? And replace with what?!
With the method that D/L itself replaced around 1992: the 'most productive overs' (MPO) method, apparently devised by Richie Benaud.
We first arrange the runs scored per over in descending order. In the Sri Lanka 20-over innings against Zimbabwe, e.g., the order was: 16,14,13,13,11,10,10... 6,6,6,5,4,4.
Now if Zimbabwe can bat only 5 overs, the rule will require them to score 16+14+13+13+11 = 67, plus 1, i.e. 68 to win. If Zimbabwe have 6 overs then it will be 16+14+13+13+11+10 = 77, plus 1, i.e. 78 to win This seems much more reasonable than the 43 or something that Zimbabwe had to score in 5 overs using D/L (which they foolishly didn't attempt).
What would have been the WI target in the other match?
85 in 5 overs, which is admittedly very high. But remember that England had a few massive overs, including an over with 27 runs and two overs at 16 runs apiece. That's pretty unusual. But at the end of the day if England could score 84 in some five overs, WI too should have to replicate or better that score to win.
And this method doesn't consider wickets lost, etc.?
No it openly acknowledges that only runs matter in T20, not wickets.
Are you comfortable with that?
Look, the biggest D/L anomaly is that it sets ridiculously low targets for 5 or 6 or 7 over innings. The big challenge is to correct that.
Could we be replacing one monster by another equally deadly monster?
Possibly. But my gut feeling is that we'll get to a happier place in T20 by tweaking the MPO method instead of the D/L method.