" Cricket Etcetera was voted as the Best Cricket Blog by Google during the World Cup 2007 "

Wednesday, March 21, 2007

Duckworth-Lewis for Cricket Dummies like me

Vinod's Cricket Brain is definitely bigger and bettter than mine. So is Abhigyan's. My apologies to Vinod for giving you a wrong bit of information while replying to your query on my post on net run rate calculations. Thankfully, on Abhi's request, I decided to dig further, and my ignorance got revealed. But what's a blogger who can't make a story out of something - so here's my post on the Duckworth-Lewis System, and I dedicate this post to cricket dummies out there, not excluding myself.

Basic premise on which Cricket's most complex system called Duckworth-Lewis works:

Resource Percentage - when any team starts a fresh innings in any full 50 overs one-day international match, they start with 100% of resources left. As the number of overs reduce and/or wickets fall, the Resource Percentage ( simply referred to as RP from here on) left keeps reducing.

note: for any match less than 50 overs, even from the start, the resource percentage doesn't start at 100, but at a lower number. There's a table which tells you all this.

Calculating:
Case 1: When the match gets interrupted with the second innings in progress, and the number of overs have to be reduced for the second innings only.
Say play got interrupted at 110/3, 30 overs and when restarted it was reduced to a 40 overs match.
RP Lost = RP remaining when 20 overs left - RP remaining when 10 overs are left
Net RP which Team 2 is left with = 100% - RP Lost
New Target = Old Target x Net Resource Percentage left. So if Team 2, chasing 800 runs is left with 70% Net RP, their new target will be = 800 x 70% = 560! One hell of a target na !!


****
Phew! That was some effort। I just hope i've not bungled on something, else let me know and please forgive. Vinod & Abhigyan raised the query that it was actually possible for the target to be higher than what Team 1 scored, and it's quite possible under Duckworth Lewis. Maybe it has already happened in the past, i don't know, but explaining that would be a far bigger challenge, which I don't feel like accepting right now. Let me know if this was an useless post, and I'm actually the King of the Cricket Dummies.


5 comments:

Vinod said...

Thanks Amit. Yeah, its not without a reason that a professional South African team under the computer-guru and now 'late' Bob Woolmer bungled up in the previous World cup on D/L. Its an extremely complicated system but I think it works well, especially when one compares it to the lousy 1992 rain rules.

I am still confused about the NRR calculations though. In a game where targets are adjusted, do the adjusted tagets apply for both teams? For example, today the SL-Bdesh match. SL made 318 in 50 overs. Rain interrupted Bdesh innings and their target was revised to 312 in 46 overs. But Bangladesh was bowled out in 37 overs.

So, the questions are
1) What is SL's run rate in this game? Would it be 318/50 = 6.36?
3) If Bangladesh was not bowled out and they played 46 overs, what would their RR be for points table calculations? Their total/46?
3) More importantly, what is Bangladesh's run rate give they were bowled out? Is it 112/46 or 112/50? It could be unfair on SL to count only 46 overs because it helps their NRR to calculate based on 50 overs (which would have been the case for bowling out Bdesh without rain).

I am just curious and waiting for cricinfo to update their points table to see what happens. But its definitely not straight forward.

Vinod said...

Okie looks like the match becomes a 46 over game overall. So Sri Lanka's run rate becomes the revised target of 310/46. And Bangladesh's run rate is just 112/46. The D/L in itself might be a reasonable system, but I am not too sure about its application to calculating net run rate.

Amit Bajaj said...

I'm getting more and more absorbed into this, Vinod and it's interesting - you have asked a really good question, i must say. Tough one but i think i've your answer:

The answer is neither!!

SL will be allotted a score of 311 off 46 overs. ( I think that's pretty neat of the ICC and D/L: 1 run less than the new target off the new number of overs. So Lanka's net run rate is now simply = (311 - Bangla score)/46.

Amit Bajaj said...

Though it won't really matter, it actually works out to the benefit of both Bangla and Lanka - and thus a loss to India. In the end, when you have the final denominator - Lanka's & Bangla's would be lower than it should have been and so their other efforts get divided by a lower number.

Agyan said...

I guess that's why you have stolen a march over your fellow-bloggers, you really get into the thick of things. And I can understand why D/L is so complicated, maybe our quantitative skills will me matched only by Anil Kumble (who is an engineer), yet the bloody rule is difficult to fathom, forget calculate.

Vinod - I don't think Woolmer bungled for SA in 2003, he was their coach only upto 1999. In fact, Woolmer learnt from Pollock's and Ford's(?) mistake. In our last tour to Pakistan, when they chased down 328 (the only ODI victory for them in the series), there was some rain threat around 15-20 overs. And Woolmer was out on the balcony with the exact score to be had after each over to win, not tie!!

I don't think the target has been revised upwards so far in ODIs, although it easily might be. The crazy thing I remember was in our trip to West Indies in 2002, when Lara was chasing some Indian score in one of the ODIs. And he had no clue of the adjustment, becoz noone in the West Indies dressing room had managed to work it out for him. I think there is scope for good statisticians to be hired with International teams.