First of all you cannot linearly reduce power play overs; for the simple reason it is a fielding restriction and has nothing to with runs scored. You have assumed that powerplay favours batting side, and it may be true,but nothing in the law says that. So while writing rules you can't simply say reduce the powerplay first.

Secondly your arguement is based on one example, which may not hold for all cases. Can you please tell me if engalnd had scored 200 runs , how much WI will be chasing. assuming it to be 101 in 9 overs, I am not so sure whether that is simpler compared to 200 in 20 overs, further reduce it to 5 overs and assuming you get need to chase 60 runs , I am not sure that is so easy.

I'd take a different approach. In your example, my target score would still be 80, and the number of overs would still be 9 (with 3 PP overs). But I'd reduce the number of wickets from 10 to 5 (approximating 0.5 from 9/20). This would still not be 100% balanced, but you can never get that with any form of math based score revision. But it evens the target out better than what is done now (even considering that the 5 wickets that are removed would be the lower order players).

You make some fair points but your analogy still seems to be wrong. WI were not 81/0 they were 0/0. This means nobody had their eye in and no partnership had yet developed. Going beserk straight away is quite hard unless the pitch is absolutely flat (This batting skill in fact separates the T20 batsmen like Gayle & Sehwag from ODI ones like Ponting).

Personally I think the main problem is that the data Duckworth and Lewis used for their mathematical formulas and permutations was mainly if not all from ODI's. The way innings was built in T20's are different so this needs to be taken into account.

If anything I think Powerplays should just go from T20's. Their purpose was to make 50 over cricket more exciting, but even this is controversial.

This will make the game easier to play and understand for new fans bearing in mind that T20's role is to expand "cricket" into new countries. Smart and aggressive captains will still put fielders close in to increase the chance of a wicket.

It wouldn't be very hard to come up with a different set of targets. Simply come up with a whole variety of scenarios and ask a whole load of cricketers which side they think is advantaged. Tweak the scenarios until the responses are evenly balanced. Then adjust your model so it generates those results. OK, this is a belts and braces approach rather than starting from a mathematical perspective but it might have the advantage of acceptance by those who have to live or die by it.

I believe there is a parameter in the D/L calculation that can be sdjusted to take scoring rate into account. They changed it for ODIs a few years ago, because scoring rates had increased.

There was an article similar to this following the England match. Btw, you didn't mention the number of wickets England lost, which is important.

I think the powerplays are fine as it is.

I mostly agree expect that 80 runs to win in nine is not quite as easy for the batting side as when the score is 81 for 0 after eleven. At 81 for 0 you have 2 batsmen well set, in the later case the batsmen are starting cold. However having said that I remember the game and immediately thought that the target was too easy. Something needs to be adjusted. Regarding the Powerplays I suggest that they are scrapped all together.

Firstly, let me state that I am a huge fan of D/L. As someone with a stats background, I can understand the logic behind it and think that it works extremely well in ODIs.

However, it is clearly less robust for T20 but I believe that the main reason for this is the lack of data available for T20 upon which to build a suitable database. The thing that makes D/L so powerful in ODIs is the sheer number of games from which to use as a base for calculating targets.

We don't have this in T20 - yet. Over time, D/L will necesarily become more robust as a method for setting T20 targets and until then we will probably have to settle for the occasional "blip" like the Eng vs WI game.

Finally, in response to Sarath's comment - the number of wickets England lost in that match is entirely irrelevent.

A team has a number of resources (10 wkts & 120 balls in T20) to score as many runs as they can. It doesn't matter how (ie 150-5 in 20 overs is the same as 150 a/o in 10 overs.)

Continued from previous post...

... so, in this case, team A who scored 150-5 in 20 overs used 50% of their "wickets" resource and 100% of their "balls" resource.

Team B scored 150 all out in 10 overs and used 100% of their "wickets" resource and 50% of their "balls" resource.

They used their resources in completely different ways but ended up with the same result - 150 runs.

Therefore, both methods were equally successful (or unsuccessful) - team A "wasted" half of their "wickets" resource and team B "wasted" half of their "balls" resource.

If team A had taken a few more risks, they may have lost more wickets but scored more runs.

If team B had batted more carefully they may have survived more balls and scored more runs.

The skill in limited overs matches (ODI & T20) is finding the most efficient way to utilise resources - that's the basis behind D/L.

I mostly agree with Gizza, but there is another thing to take into account aside from two batsman not having their eye in. At 0/0 the ball is brand new. At the ypothetical 81/0 the ball has been smacked around for 11 overs, surely that has to have had some impact.

Well, all being spoken, I feel that the English have always found a reason or a rule to blame. It seemed to me a fair game. Imagine the over with one run given in it being bowled in the start. Would West Indies still have reached the target? I doubt it. In such a short 9 over game, there was no chance of error in the start. They would have lost their nerves and the game.

When it comes to rain delays, Its always unfair to someone. Cricket is so. Why don't you suggest that batting and bowling conditions change over time and so does the pitch, even that should be balanced out.

Relax, sit back and enjoy the cricket.

Cheers.

You have made some valid points. I agree with you that a direct application of the D/L rule for Twenty20s may be unfair and the rule does need tweaking. However, I think neither of the suggestions made by you is the right change.

The best way to solve D/L issue is that even if it rains the match must be kept 20 overs itself. Because a T20 lasts just around 3hrs and no problem if it ends very late.
If at all no play is possible, then the winner must be decided by super-over eliminator

In my opinion it would be unfair not to call cricket a batsman favored sport. But since twenty 20 is chosen as the way to promote cricket, it should be experimented with. I think power plays as mentioned earlier were to excite the 50 over game, should be removed twenty 20. this could result in a more calm and excitement filled contest.

Good article, absolutely this needs tweeking and I believe that a balance of both points would be a good way forward, perhaps adjust it to losing one powerplay over for every two overs lost - counters the argument that powerplays are favour the batting side entirely.

I find Martin's comments baffling to be honest. Yes it is just one example of the D/L method, but the the very nature of the D/L system means that you can extend the scenario to all circumstances. He goes on to say that 101 in 9 and 60 in 5 overs are not easier tasks than 200 in 20 - of course they are. Holding a scoring average of 10% and 20% more over 50% and 75% shorter periods of time is absolutely a more simple proposition. It is far easier to score at a strike rate of 200 over 8 balls (16) than it is over 40 (80), the statistics of this are quite clear.

For me this is very straight forward and it is illogical to argue against this - how many sides have scored 100 off 10 overs and failed to score 200 off 20...

The concept of D/L (or any other target calculator) is fundamentally flawed. A Twenty20 game lasts for about 3 hours, and if you can't get 40 overs in that time, the game is a draw. PERIOD. We do accept that in Test Cricket, don't we? Consider these scenarios: 1) At Tea on Day 5, the batting side is 8 down, still needing 200 runs to win. Last session is washed out. Result: DRAW 2) At Stumps on Day 4, batting side needs 75 runs to win with 9 wickets in hand. Last day is washed out. Result: DRAW The "unlucky" team can whine, cry, scream, or do whatever they want. The result is still a DRAW.

I think Jeff has it nailed. The database of ODI's and List-A matches simply didn't have enough matches where a team had a position remotely resembling the start of a 20-20 (they'd have had to bat for 30 overs--20 or 25 maybe in some places' List A--without losing a wicket). As more 20-over matches are played at the top level, that part of the table will, I expect, be modified to take the new evidence into account. But this doesn't require changing the fundamental method of setting the new target, only the inputs to that method.

The only thing which shifted the balance to the chasing side is "wickets in hand". When you know that you have 10 wickets in hand and only 9 overs or so, it placed you in much comfort zone rather to keep wickets in hand for 20 overs

I am given to understand that the D/L system of determining second teams' targets in rain affected matches is based on determining available resources (wickets & overs) compared to the resources used by the first team. As such it uses a data base of ALL previous matches to generate its numbers. This means that in ODIs the data pool is nearly 3000 matches (& so is fairly finely tuned). At the moment there have been less than 200 T20 internationals which is a much smaller data base and so you will get 'unusual/unfair' results. Given time and more matches played the data base will grow and even out much more.

A number of good points have been made by the author and the responders. The most critical issue is the relative paucity of data from T20 games with which to build a reasonable statistical model for runs versus unexpended resources. This was brought up by Jeff. I am not sure if Duckworth and Lewis even atttempted to calibrate their existing model, developed for ODI, to the T20 data (whatever is available). Another important issue is the inherent uncertainty in the model prediction of runs to be scored by the second team. My opinion is that this should be taken into consideration and rather than having a single point estimate of the target, an interval estimate should be set as the target. The second team loses if it falls below the lower limit of this interval; it wins if it crosses the upper limit, and it is a tie otherwise. This is a fair system that acknowledges the uncertainty. As the database of T20 games expands, this interval will shrink.