Duckworth/Lewis - T20 revamp needed?

[mas cambios]

Duckworth/Lewis - T20 revamp needed?

Following on from here...http://www.bigcricket.com/forum/t76352/

I'm not overly bothered about England losing to the Windies (after all, rules are the same for both sides; seems Gayle just played the game better by choosing to bat second and going for it from the off) but it does seem that the D/L method may need some tweaking for the shorter format.

Take the game on Monday, if West Indies had batted 20 overs then they would have needed 9.56 per over. The rain reduced total had them needing 60 off 36 balls or 10 over.

It seems the problem is not so much the total but the wickets left in which to get them. I'm sure most here will agree that chasing 60 off 6 overs with 10 wickets in hand is a lot easier than chasing 192 off 20. Simply put you can throw more caution to the wind.

So, should the D/L method be tweaked to make it fairer or is it fine as it is? Should wickets be removed as well as overs when calculating reduced totals?

 

[Caesar]

Re: Duckworth/Lewis - T20 revamp needed?

DL is based on predicting a score based on remaining resources, extrapolating from what has occurred thus far in an innings. Because of this I think there are few notable problems with applying this system to T20, which I would pigeonhole in two categories:

1) Deficient underlying data
In order to predict scores, DL uses a table for analysis that is based on historical data from international matches. For ODIs this is fine - the format is well established, and teams play a lot of games. For T20, it's a different story. Last year, for example, there were 150 ODIs played, and 48 T20Is.

2) Difficulty extrapolating data
In an ODI, DL requires a minimum of 20 overs (40% of the innings). T20Is only require 5, or 25%. This causes a problem in and of itself. This difficulty is magnified by the fact that with such a reduced timeframe, scoring rates may fluctuate much more wildly. Strategies for T20Is are much less developed than for ODIs, and teams will often score quite differently. Trading wickets for runs in early overs can make huge differences to the resources equation even if it does not change things in the long run.

----------

I think this last point really is the key one. I do not favour using DL for matches where less than 50% (or even maybe 75%) of the match has been completed, for the above reasons. But if it is going to be used, the major thing that needs to be tweaked is the balance of resources.

The matching of value between runs and wickets has always been the contentious point of the DL method, and it's exacerbated in T20I. In reality, wickets are worth far less in the shortest version of the game, and as can be seen from the England match they are currently not penalised enough to prevent people throwing them away when they have a rain-shortened match. Hopefully this problem will be mitigated as more data is collected and reviewed.

So in summary - if you're going to use it, increase the weighting applied to wickets as resources, and extend the minimum overs to at least 8 (40% of the innings) to make it consistent with ODIs.

 

[Boris]

Re: Duckworth/Lewis - T20 revamp needed?

To be fair, I don't think it's the Duckworth/Lewis system that is the problem, but the form of game and the small time frame that is left to decide a match. I've asked around the mathematics circle and they all say that this system is still the best, there isn't really a way of predicting anything accurately in that time frame. Cricket was made for four unlimited innings unrestrained by time, even changing Tests to limit them to 5 days brought around a large problem... the draw.

Everything is worked around, and one way might be to change the format itself, but the way I'm thinking will get rid of the only reason T20s are played, the excitement factor.

I like the idea of having 6 batsmen a side, but 11 fielders and 11 possible bowlers. 5 wickets means the side is all out.

It does mean that those 6 batsmen are all specialists, but of course it does reduce the amount of batting power, and batsman have to be careful not to throw their wickets away.

Also as the bowlers don't bat, 4 bowlers are picked purely for their bowling, nothing else. Only one all rounder is needed, and every side has their all rounder nowadays that can get a spot for either batting or bowling. Then there are the part timers to fill in bowling spots.

It is out of the pattern though. Test cricket, 40 wickets per match. Less then half the time for ODIs, 20 wickets per match. T20s 20 wickets per match with half again the amount of time? Doesn't fit in, but those extra batsmen are there for the extra hitting and excitement.

 

[hattrick]

Re: Duckworth/Lewis - T20 revamp needed?

I going to sound like a broken record but cricket is suppose to be fair and entertaining to both the players and the fans. Duckworth Lewis has failed my country twice. The 1992 farce at Sydney(below) which I always remember when Duckworth Lewis comes to attention and then 2003 disaster against Sri Lanka in Durban in which they scored one short and ended getting booted out of the tournament which admittedly was there own mistake.

Having done a bit of searching I see there is another method based in India used in the ICL ... yes a 20 over based competition which makes sense.

I do agree that the current situation is far from perfect and this is not a storm in a teacup but rather something that ICC should look at. I come back to my point that 20 over cricket is about entertainment and this is more a less a soapy that never seems to go away.

Jayadevan system, also known as the VJD method is what I am referring to. If your looking for an explanation read these: A comparison to Duckworth Lewis: rediff.com: cricket channel: Is Jayadevan's proposed method better than the Duckworth/Lewis method? and a limited overs implementation of it: rediff.com: cricket channel: Rain rules, okay!

I am sorry but I dont except the current farce which is happening in the Caribbean and how it is influencing the tournament. Their are sponsors, players and fans involved and you want a fair contest which is not happening at the moment. My opinion is that Duckworth Lewis always benefits one side may it be the fielding side or the batting side on certain occasions. I understand that the ICC cannot control the weather but really this needs to be sorted out ...

 

[yashsr]

Re: Duckworth/Lewis - T20 revamp needed?

D/L gives a low target of 60 in 6 overs because:
1) It doesn't take into account powerplays which is its weakness.
2) This isn't a pure case of target in 6 overs. WI first batted 2.2 Overs without losing a wicket. They were pacing their innings for chasing 192 in 20 overs when suddenly they were reduced to 6 overs. The fact that they were already scoring at a healthy rate by being 30 for no loss is irrelevant, that is what they did better and we cannot take it away from WI. So we cannot scale that much...I agree with this point.

For those interested, my Yank method gives a target of 69 in 6 overs which appears quite fair. Had the interruption been in the break, Yank method would have set a target of 84 in 6 overs which again is fair.

For those interested, here is a glimpse of the Yank method which I've developed with my friend:
http://www.mrdemo.com/yank/

 

[Ljp86]

Re: Duckworth/Lewis - T20 revamp needed?

For those interested, my Yank method gives a target of 69 in 6 overs which appears quite fair. Had the interruption been in the break, Yank method would have set a target of 84 in 6 overs which again is fair.

For those interested, here is a glimpse of the Yank method which I've developed with my friend:
http://www.mrdemo.com/yank/

84 runs off 6 overs is in no way fair, in fact it's utterly ridiculous. That's more than two runs a ball, 14 runs an over in fact and with six overs to bat they'd have one over of powerplay.

That is not fair at all.

 

[Boris]

Re: Duckworth/Lewis - T20 revamp needed?

84 runs off 6 overs is in no way fair, in fact it's utterly ridiculous. That's more than two runs a ball, 14 runs an over in fact and with six overs to bat they'd have one over of powerplay.

That is not fair at all.

I think it is, as the interruption was during the break. The team knows they have only 6 overs to bat, so worse comes to worse they can even lose a wicket per over and still be in the hunt. They could change the order around to put their power hitters at the top and chasing 14 an over isn't that hard when wickets aren't a cause for concern.

If the interruption came during the innings, though, that's different.

 

[yashsr]

Re: Duckworth/Lewis - T20 revamp needed?

84 runs off 6 overs is in no way fair, in fact it's utterly ridiculous. That's more than two runs a ball, 14 runs an over in fact and with six overs to bat they'd have one over of powerplay.

That is not fair at all.

Thanks for your comments although I must say that if the first batting team scored at more than 9.5 runs an over, 14 runs an over is really fair when the overs are reduced to a mere 30% of the original. Boris in my opinion is correct and actually spot on.

And just for your information, they'd have 2 overs of powerplay.(30% of 6 overs is 1.8 which approximately equals 2)

Boris, if the interruption occured after 2.2 overs with no wickets lost, the Yank method gives a target of 69 in 6 overs...this is what happened in that match, so the D/L target isn't that bad as people think, but its a little too low nevertheless.

 

[Ljp86]

Re: Duckworth/Lewis - T20 revamp needed?

I think it is, as the interruption was during the break. The team knows they have only 6 overs to bat, so worse comes to worse they can even lose a wicket per over and still be in the hunt. They could change the order around to put their power hitters at the top and chasing 14 an over isn't that hard when wickets aren't a cause for concern.

If the interruption came during the innings, though, that's different.

It's a completely unrealistic target. The team batting first couldn't crack 10 runs an over but the second team has to score at 14 runs an over. Just remember, 14 runs an over is more than two runs per ball. Even scoring one runs puts them well behind the required rate.

Most teams struggle to score more than 60 runs from six overs during a powerplay period in a Twenty20 game. So what makes you think they'll score 84 runs in six overs with only two overs of powerplay?

Thanks for your comments although I must say that if the first batting team scored at more than 9.5 runs an over, 14 runs an over is really fair when the overs are reduced to a mere 30% of the original. Boris in my opinion is correct and actually spot on.

And just for your information, they'd have 2 overs of powerplay.(30% of 6 overs is 1.8 which approximately equals 2)

Actually Yash it is a very unreasonable target score. Expecting a team to score at 14 runs an over when the other team couldn't get above 10 is pretty unfair. I couldn't see any international side chasing down that target at all, it's way too high. Like I said above, most sides struggle to score more than 60 runs in six overs in periods which are all powerplay overs. Today's bowlers are too smart to be smashed for 14 runs an over for six overs so the target of 84 is quite unfair in my opinion.

 

[yashsr]

Re: Duckworth/Lewis - T20 revamp needed?

Most teams struggle to score more than 60 runs from six overs during a powerplay period in a Twenty20 game. So what makes you think they'll score 84 runs in six overs with only two overs of powerplay?

I'm not telling that they'll score 84 runs in 6 overs. What I'm telling is that 84 runs is their target. They've conceded 191 runs, so obviously its unlikely they will chase 192 in 20 overs or 84 in 6 overs.

I couldn't see any international side chasing down that target at all, it's way too high.

Check this match out:
2nd Match, Group D: New Zealand v Scotland at The Oval, Jun 6, 2009 | Cricket Scorecard | Cricinfo.com
Scotland scored 89 in 7 Overs against a decent New Zealand attack and New Zealand scored 90 runs in 6 overs. And this was a 7 over game from the beginning, no interruptions..

I couldn't also see any international side scoring 191 in 20 overs day in and day out. England put up a very good score and Collingwood said that a team that has posted 191 would go on to win 95% of the times. He's almost correct....you'd probably chase 191 on average 1-2 times out of 10 at best.

 

[Ljp86]

Re: Duckworth/Lewis - T20 revamp needed?

I'm not telling that they'll score 84 runs in 6 overs. What I'm telling is that 84 runs is their target. They've conceded 191 runs, so obviously its unlikely they will chase 192 in 20 overs or 84 in 6 overs.

Yes and I am saying that the target score is too high.

Check this match out:
2nd Match, Group D: New Zealand v Scotland at The Oval, Jun 6, 2009 | Cricket Scorecard | Cricinfo.com
Scotland scored 89 in 7 Overs against a decent New Zealand attack and New Zealand scored 90 runs in 6 overs. And this was a 7 over game from the beginning, no interruptions..

I couldn't also see any international side scoring 191 in 20 overs day in and day out. England put up a very good score and Collingwood said that a team that has posted 191 would go on to win 95% of the times. He's almost correct....you'd probably chase 191 on average 1-2 times out of 10 at best.

I am quite familiar with that game and it is most certainly an outlier in terms of Twenty20 cricket. The pitch was also akin to a slab of concrete as it was that easy to bat on.

Australia chased 191 down in the semi-final of the World Twenty20 earlier this year, scoring 7/197. Whilst that too is quite rare, teams have more of a chance making 192 off 20 overs than 84 off six. Perhaps if the six-over target was more reasonable (around 70-ish) then we probably wouldn't be having this debate.

 

[yashsr]

Re: Duckworth/Lewis - T20 revamp needed?

Yes and I am saying that the target score is too high.Whilst that too is quite rare, teams have more of a chance making 192 off 20 overs than 84 off six. Perhaps if the six-over target was more reasonable (around 70-ish) then we probably wouldn't be having this debate.

I'd much rather chase 75 in 6 overs than 192 in 20 Overs.

Just take a hypothetical example, Team 1 score 720 in 20 Overs in a T20 match scoring at 36 runs an over or 6 runs per ball and Team 2 are to be set a target for 10 Overs. What would be the target? Surely more than 360, say 450. But you'd say that 720 in 20 Overs is atleast possible and 450 in 10 Overs is just impossible if you remove extras and other virtual impossibilities. So you'd say the rain rule is not correct. But obviously rain rules don't work like that!

So the WI vs. Eng is just such a high scoring game but the magnitude of the above anomaly applicable is much less....

Thats the best I can put forth in my argument!

If there is any case if anyone would like me to do a comparison with D/L and Jayadevan with Yank, I'd be happy to oblige, specially if its a real match situation although I wouldn't mind an imaginary example but surely not like the hypothetical example I just gave above!

 

[yashsr]

Re: Duckworth/Lewis - T20 revamp needed?

Heyljp86,

We've actually added 2 new concepts in our 'Advanced Yank model' which we will soon put up on our page.

The answer for the case if it rained in the interval would be 80 runs off 6 overs(so its 4 runs less than the 'Basic Yank model' score) with 2 overs of powerplay and 10 wkts in hand.

 

[The Ox]

Re: Duckworth/Lewis - T20 revamp needed?

I going to sound like a broken record but cricket is suppose to be fair and entertaining to both the players and the fans. Duckworth Lewis has failed my country twice. The 1992 farce at Sydney(below) which I always remember when Duckworth Lewis comes to attention and then 2003 disaster against Sri Lanka in Durban in which they scored one short and ended getting booted out of the tournament which admittedly was there own mistake.

Having done a bit of searching I see there is another method based in India used in the ICL ... yes a 20 over based competition which makes sense.

I do agree that the current situation is far from perfect and this is not a storm in a teacup but rather something that ICC should look at. I come back to my point that 20 over cricket is about entertainment and this is more a less a soapy that never seems to go away.

Jayadevan system, also known as the VJD method is what I am referring to. If your looking for an explanation read these: A comparison to Duckworth Lewis: rediff.com: cricket channel: Is Jayadevan's proposed method better than the Duckworth/Lewis method? and a limited overs implementation of it: rediff.com: cricket channel: Rain rules, okay!

I am sorry but I dont except the current farce which is happening in the Caribbean and how it is influencing the tournament. Their are sponsors, players and fans involved and you want a fair contest which is not happening at the moment. My opinion is that Duckworth Lewis always benefits one side may it be the fielding side or the batting side on certain occasions. I understand that the ICC cannot control the weather but really this needs to be sorted out ...

Duckworth-Lewis didn't even exist in 1992.
Duckworth?Lewis method - Wikipedia, the free encyclopedia

If that game had been using D-L SA would have needed 4 off the last ball, not 22

 

[Ljp86]

Re: Duckworth/Lewis - T20 revamp needed?

Heyljp86,

We've actually added 2 new concepts in our 'Advanced Yank model' which we will soon put up on our page.

The answer for the case if it rained in the interval would be 80 runs off 6 overs(so its 4 runs less than the 'Basic Yank model' score) with 2 overs of powerplay and 10 wkts in hand.

Still a really high target yash, really couldn't see many teams (if any at all) achieving that sort of target.

 

[yashsr]

Re: Duckworth/Lewis - T20 revamp needed?

Still a really high target yash, really couldn't see many teams (if any at all) achieving that sort of target.

Whatever you feel but 80 runs off 6 overs is not unachievable although pretty difficult but so is 191 off 20 overs...And we're not trying to set a realistic target for WI, we're trying to discount 191 off 20 overs to 6 overs. Since 191 is huge, even the target in 6 overs has to be huge.

How many teams have won chasing 191 or more? I think it has been done only once or twice in T20 Internationals.

As per the Yank's Advanced method, the par score for 191 in 20 overs in the shortened 6 overs is about 78.6 by our method which means approx 79 and thus target comes to 80.

We're totally convinced....In fact if you have a look at the CricInfo article by Srinivas Bhogle and Rajeeva Karandikar, they write:
"A target of 69 certainly appears more reasonable, but what if the interruption had occurred between the innings and West Indies knew from the start that they only had six overs to bat? The target then would have been 87, or 14.4 runs per over, with only two overs of field restrictions. This appears steep, but we mustn't forget that 191 too is a lot of runs."

For the article, see:
http://www.cricinfo.com/magazine/content/story/459431.html

So in fact, they were even happy with 87 which is a little too high in this case. And our corresponding targets here are 69 and 80 runs which is totally fair. Ljp86, what do you say? What do others have to say?

 

[Ljp86]

Re: Duckworth/Lewis - T20 revamp needed?

Whatever you feel but 80 runs off 6 overs is not unachievable although pretty difficult but so is 191 off 20 overs...And we're not trying to set a realistic target for WI, we're trying to discount 191 off 20 overs to 6 overs. Since 191 is huge, even the target in 6 overs has to be huge.

How many teams have won chasing 191 or more? I think it has been done only once or twice in T20 Internationals.

Not many. But I would bet that more teams have achieved 191 off 20 overs than 80 off six overs.

As per the Yank's Advanced method, the par score for 191 in 20 overs in the shortened 6 overs is about 78.6 by our method which means approx 79 and thus target comes to 80.

We're totally convinced....In fact if you have a look at the CricInfo article by Srinivas Bhogle and Rajeeva Karandikar, they write:
"A target of 69 certainly appears more reasonable, but what if the interruption had occurred between the innings and West Indies knew from the start that they only had six overs to bat? The target then would have been 87, or 14.4 runs per over, with only two overs of field restrictions. This appears steep, but we mustn't forget that 191 too is a lot of runs."

For the article, see:
Why Duckworth-Lewis doesn't work in Twenty20 | Specials | Cricinfo Magazine | Cricinfo.com

So in fact, they were even happy with 87 which is a little too high in this case. And our corresponding targets here are 69 and 80 runs which is totally fair. Ljp86, what do you say? What do others have to say?

69 is fair in my opinion, 80 is not.

You've got to remember that a score of 80 from six overs is more than two runs per ball. A dot ball or scoring only one run per ball would put the batting side well behind the rate required and appears to be quite unrealistic in practice.

I think that your method focuses too much on the mathematical side of things rather than what is realistically achievable. 80 might be the mathematical equivalent but is it an achievable target? I don't think it is.

 

[mas cambios]

Re: Duckworth/Lewis - T20 revamp needed?

Whatever you feel but 80 runs off 6 overs is not unachievable although pretty difficult but so is 191 off 20 overs...And we're not trying to set a realistic target for WI, we're trying to discount 191 off 20 overs to 6 overs. Since 191 is huge, even the target in 6 overs has to be huge.

The difference is that the 191 is 'only' 9.55 runs per over. 80 off 6 requires over 13. Somewhere between the two seems fairer, 69/70 maybe. That still allows for having 10 wickets and only 6 overs but means that the target is within reason.

How many teams have won chasing 191 or more? I think it has been done only once or twice in T20 Internationals.

A score of 191 has been achieved 7 times in 185 T20 games. Of those, 3 resulted in wins, 3 in losses and there was a single tie.

 

[yashsr]

Re: Duckworth/Lewis - T20 revamp needed?

Not many. But I would bet that more teams have achieved 191 off 20 overs than 80 off six overs.



69 is fair in my opinion, 80 is not.

You've got to remember that a score of 80 from six overs is more than two runs per ball. A dot ball or scoring only one run per ball would put the batting side well behind the rate required and appears to be quite unrealistic in practice.

I think that your method focuses too much on the mathematical side of things rather than what is realistically achievable. 80 might be the mathematical equivalent but is it an achievable target? I don't think it is.

The mathematical equivalent is actually a target of 84 runs in 6 overs but our new concepts have been developed to input reality which brings it down to 80 runs in 6 overs. We actually take an average score and...well, I'll talk later once we upload the entire T20 documentation...

But whatever you say, 80 runs in 6 overs is achievable...People didn't think that 350 in 50 overs was achievable but you now it is...people didn't think 200 in 20 overs was possible because they were used to seeing about 225 getting scored in a 50 over game....if you have games of such duration like 6 overs, you'll see teams scoring 80 runs in at least 5-10% of the games. Similar is the probability of 191 in 20 overs.

 

[yashsr]

Re: Duckworth/Lewis - T20 revamp needed?

A score of 191 has been achieved 7 times in 185 T20 games. Of those, 3 resulted in wins, 3 in losses and there was a single tie.

Yeah once was in the semi-final in which Pakistan literally threw the match in Michael Hussey's hands....you never know with Pakistan so I wouldn't count this match...fixing etc.

Once was the very same match we're talking about in this topic...England lost partly because WI were set a very low target of 60 in 6 overs where it should have been about 69

Only once has the a team won scoring more than 191 realistically....South Africa vs. WI when Gayle scored a ton but still lost

So 191 is a HUGE Total....The fact that is has been crossed only 7 times in 185 games prove that....and its 6 over equivalent would also be huge...