Duckworth/Lewis - Fair?

[Precambrian]

Average score in the first innings in matches involving the top 9 ODI nations since the start of this year is 238.44

Without Bangladesh involved 237.53

Past 2 years including Bangladesh 241.85

Without Bangladesh involved 245.04

Country-wide there could be fluctuations. Like India and Pakistan where 300+ scores have become the norm than exception.

 

[Athlai]

Last two years all ODI teams in 1st innings
Asia 257.09
Africa 232.10
Americas 238.13
Europe 196.47
Oceania 226.70

 

[Redbacks]

Perhaps the 1992 system would have been fairer to chop off the median overs.

The majority of mathematical models for pricing etc. are aimed at determining the 'indifference' value at the present time. This seams to be the crux of the D/L system. Both teams should be indifferent to the revised target as they have been given a 'best fit' present value that is equivalent to the initial situation.

 

[zaremba]

Perhaps the 1992 system would have been fairer to chop off the median overs.

The majority of mathematical models for pricing etc. are aimed at determining the 'indifference' value at the present time. This seams to be the crux of the D/L system. Both teams should be indifferent to the revised target as they have been given a 'best fit' present value that is equivalent to the initial situation.

This is the beauty of D/L. Time after time I've seen a rain delay, seen the revised target and thought, "fair enough"

 

[AlanJLegend]

Duckworth Lewis Discussion thread

Given the recent D/L situations in the Australia/New Zealand series, I think it would be interesting to have a bit of discussion regarding this 'score prediction' series.

In the 2nd ODI of the series, the NZ's total was reduced by 5 overs and 8 runs. In the 4th ODI, Australia's chase was reduced by 16 overs and 34 runs.

I have no idea how the system works. I have looked it up on Wikipedia and tried to make sense of it, but it is certainly not simple. Looking at the above reductions, they appear pretty unfair to the chasing side. What do you guys reckon?

I can see that it is somewhat reasonable in that chasing a score at a high run rate is easier over 20 overs than a score at a moderate run rate over 50 (if that makes sense), however I can't see how 5 overs for 8 runs really makes sense.


I would be interested to hear what the CW community thinks of the D/L system, if it is a bad system, what would be better?

 

[Sir Alex]

Hi Alan as I was guessing a quick search revealed the following link

http://www.cricketweb.net/forum/cricket-chat/40603-so-how-d-l-method-fair.html

there is a pretty good discussion in that.

cheers.

 

[vic_orthdox]

Thanks for digging that up, Sir Alex. I'll merge it in a second.

WRT AlanJLegend, I think it reflects the fact that teams are getting better at chasing, and that with a known target one can better structure their innings. They can also afford to take more risks, as the loss of wickets may not be as important with less overs available to score.

 

[Sir Alex]

Np mate.

Yes I think it is the fairest system available in the given cirucmstances. The fact that it is a bit boring to comprehend doesn't mean it is a wack job. It gives value to wickets held in hand at the time of cut off which I believe is a fantastic concept as compared to the silly maiden overs chop that existed during the 1992 World Cup.

 

[Prince EWS]

I can see that it is somewhat reasonable in that chasing a score at a high run rate is easier over 20 overs than a score at a moderate run rate over 50 (if that makes sense), however I can't see how 5 overs for 8 runs really makes sense.

The 5 overs for 8 runs thing occurred because New Zealand had lost early wickets. Obviously losing early wickets is less relevant in a shorter chase as you have less time in which to use your resources, but they lost they wickets when they were still expecting to have 50 overs to play with, so the value of them had to be adjusted to fully reflect what had gone on. On the face of it, it does seem odd, but it does make sense.

It's hard to get your head around at first but once you fully understand how it works and why, you'll realise it's by far the best method and is generally very fair to both sides.

 

[Neil Pickup]

I think the 40-over and 50-over leagues that have played out in the UK over the last few years indicate the difference between innings of that length is negligible - par scores in the Pro40 have been virtually identical to the FP Trophy matches. Innings structure with wickets in hand, here at least, results in very similar end totals.

If we look at ODI v T20, a "good" first innings score (i.e. one you expect to win more often than not is, I'd imagine, 250). In a T20, you're probably feeling about the same for 160. That's a decrease of 30 overs (60%) vs a decrease of 90 runs (36% less). Someone else who doesn't need to go referee in three minutes can dig through Statsguru and find the exact 50-50 win-loss first innings scores, I'm sure, but I'd imagine they'll be in that range.

 

[Uppercut]

I think the 40-over and 50-over leagues that have played out in the UK over the last few years indicate the difference between innings of that length is negligible - par scores in the Pro40 have been virtually identical to the FP Trophy matches. Innings structure with wickets in hand, here at least, results in very similar end totals.

If we look at ODI v T20, a "good" first innings score (i.e. one you expect to win more often than not is, I'd imagine, 250). In a T20, you're probably feeling about the same for 160. That's a decrease of 30 overs (60%) vs a decrease of 90 runs (36% less). Someone else who doesn't need to go referee in three minutes can dig through Statsguru and find the exact 50-50 win-loss first innings scores, I'm sure, but I'd imagine they'll be in that range.

Resisting the temptation to dig into Statsguru, I'd suggest that a 'good' first innings total (not the same as a par score) at a top-level ODI (i.e. between two test-playing nations) is closer to 280. The batting powerplay in particular has seen a massive upsurge in 300+ scores.

Varies heavily based on conditions though, obvz.

 

[Uppercut]

Just to put that into perspective, when New Zealand hit 245 in the third ODI, did anyone think they'd got a good total?

 

[Jarquis]

Probably duffed up this statsguru search but....

Records since ODIs began : Team records | One-Day Internationals | Cricinfo Statsguru | Cricinfo.com
5.15*50 = 257.5

Last decade : Team records | One-Day Internationals | Cricinfo Statsguru | Cricinfo.com
5.51*50 = 275.5

In the last two years : Team records | One-Day Internationals | Cricinfo Statsguru | Cricinfo.com
5.90*50 = 295

 

[slugger]

The system of the DL could actually be scraped to point if the ODI structure was slightly adjusted if the game was broken down into quarters. (25-25-25-25)

example:
if it rained in the 1st quarter lets say for 4 overs.
the 1st quarter and the second quarter are both reduced by two overs 23-23-25-25

if it rains in the second you adjust the second and the third
if it rains in the third you adjust the third and the forth.

if it rains in the forth this is the only time you use the DL system because the team the batter first has already batted out their 50 complete overs.

NB only 10 wickets as per normal ODI.

 

[KiWiNiNjA]

Nah, might as well play Baseball.

 

[Top_Cat]

Reckon the D/L is a brilliant bit of work, a great example of merging of maths with real-life data. Been the most widely accepted system for that reason.

As an aside, they're also a decent band;

The Duckworth Lewis Method on MySpace Music - Free Streaming MP3s, Pictures & Music Downloads

Any band who has a song named Jiggery Pokery is alright by me.

 

[Uppercut]

Reckon the D/L is a brilliant bit of work, a great example of merging of maths with real-life data. Been the most widely accepted system for that reason.

As an aside, they're also a decent band;

The Duckworth Lewis Method on MySpace Music - Free Streaming MP3s, Pictures & Music Downloads

Any band who has a song named Jiggery Pokery is alright by me.

I wrote a review of that album, it's around here somewhere. I've also seen them play live during the innings break of the Ireland/England game at Stormont last year. Twas pretty awesome.

 

[Richard]

I wrote a review of that album, it's around here somewhere.

Here in fact.

 

[Neil Pickup]

Resisting the temptation to dig into Statsguru, I'd suggest that a 'good' first innings total (not the same as a par score) at a top-level ODI (i.e. between two test-playing nations) is closer to 280. The batting powerplay in particular has seen a massive upsurge in 300+ scores.

Varies heavily based on conditions though, obvz.

From 2001 to today, teams batting first and scoring 250-260 have won 64 out of 99 matches.
Between 240-250 it's 48 out of 103.

From 2007 to today, the stats are 11 out of 35 for 240-250, and 21 out of 35 for 250-260.

280+ has won 128 out of 154 games since 2007 - not really the number I'm looking for. I stand by my number!

---

In all T20Is, 160-170 has won 5 out 16, and 170-180 has won 8 out of 9: in fact 170+ has only lost three times as a target. So we're definitely looking in the same range of figures - a far smaller reduction in the par score than in the number of overs.

 

[Uppercut]

I may be somewhat jumping the gun, because it hasn't been very long, but my thinking has changed based on the introduction of the batting powerplay (Oct 2008). Since then, 250-260 wins 4 times out of 7, 260-270 wins 4 times out of 7, 270-280 wins 7 times out of 13. It's all rather marginal indeed. I'd call that par.

We'll have this discussion again in a few years when there are a few more numbers for me to play with.