The Duckworth Lewis Stern (DLS) method, used to calculate par scores and revised targets in rain-affected cricket matches, has been under the scanner ever since the ICC Men's T20 World Cup 2022.
With rain being a constant feature in Australia, usage of the DLS method was widespread throughout the competition, and a section of cricket fans questioned whether the system is perfectly flawless, or needs to be mended.
Essentially, the discussion was regarding the plausibility of having a new system specifically for T20 matches, given that the DLS method uses a combined (ODI+T20I) database for its calculations.
The difference between ODI and T20I cricket has never been as glaring as it currently is, but while the DLS method was devised to be used in 50-over cricket, there is still no specialised method used in T20I matches.
To have all the questions regarding DLS answered, The Quint spoke exclusively with Professor Steven Stern – the current custodian of the system.
The DLS method came into existence long before the inception of T20 cricket. Do you think there is a need to have format-specific tables now, that is, different tables for ODIs and T20Is?
Steven Stern: It is true that the method came into existence well before T20 cricket, but it also did so before ODI cricket reached the scoring levels it commonly does now. The average ODI score in the mid-1990s was around 230, now it has reached around 265.
Given this, every year on July 1, I re-analyse the preceding 4-years' worth of data to assess if, or how, scoring patterns are changing and whether DLS needs to be updated. So far, on the basis of these annual re-analyses there has been no statistical indication a separate program is needed for ODIs and T20Is.
DLS considers T20I matches as the last 20 overs of an ODI match – but is this assessment not flawed?
Steven Stern: It intuitively may seem wrong to some, because it is very rare in a ODI match for a team to reach their final 20 overs with no wickets down. However, when one adjusts for wickets, the scoring patterns for T20s and the final 20 overs of an ODI are identical.
As an example, the following graph shows the average runs scored with any given number of overs remaining and 2 wickets in hand.
In the graph, the dark black dots represent the average score for the overs remaining based on at least 50 ODI matches and the dark blue dots represent the average score for the overs remaining based on at least 50 T20 matches.
As you can see, within minor statistical error, once wickets down are accounted for, the final overs of an ODI are identical in structure to the overs in a T20.
Given the difference in scoring patterns between men’s and women’s cricket, is there a need to have a method designed specifically for women’s cricket?
Steven Stern: I run checks on whether the use of the method should be changed for implementation in the women's game, and to date, there is no need. The primary reason for this is that DLS is based on capturing run-scoring acceleration throughout an innings via percentages.
So, while women's total scores are, on average, about 30 runs lower than men's at present, the percentage of their total runs scored during any portion of a match is the same as that in a men's innings.
The general scoring pattern in T20I cricket is not linear – teams tend to score more in powerplay and slog overs, while the scoring rate decreases in the middle overs. Is this scoring pattern adhered to in DLS calculations?
Steven Stern: As is commonly noted by commentators the world over, when the power plays are on, more runs are indeed scored, but more wickets are lost. Careful analysis has shown us that these two phenomena balance each other, so that while runs/over are different in power plays, runs/resource are not. As DLS works on a runs/resource basis, there is no need to adjust for powerplay overs.
(Resource refers to the number of wickets and amount of overs the batting team has).
But given how the game is ever-changing, will we have different programs for different formats in the future?
Steven Stern: At present, if I were to construct separate analyses for ODI and T20 matches, they would give essentially identical results, which is why there is no need for different programs. If it ever happens that my annual analysis shows this confluence to have been broken, there will indeed have to be separate resource tables for ODIs and T20s.
Is it plausible to inculcate qualitative aspects in the mathematical formulation? Example – Knowing that rain is imminent, the bowling team bowls out all four overs of its best bowler inside the first eight overs – which they would not have done in a regular match. The rain interruption occurs after the best bowler’s spell concludes. Could individual attributes ever come into consideration?
Steven Stern: A team may feel the rain is imminent, and thus change their strategy, but they do so at their peril since only the Cricket Gods know when the rain will actually come and only the Umpire Gods know exactly when it will be heavy enough to warrant suspension of play!
So, if a team opts for your strategy of bowling only its best bowlers early, assuming the match will be shortened, this will be to their advantage if rain indeed terminates play, but if it doesn't, they will be left with their weaker bowlers for the "death overs" and their strategy will have seriously backfired.
As another example, you may say that a team batting first while rain is drizzling but not yet heavy enough to stop play is at a disadvantage when the team batting second, after the rain has finally become heavy enough to stop play, gets to come out and bat in the sunshine.
It is true that this would be unfortunate for the team batting first, but that's cricket! It's why there is a toss and why captains can choose to bat first or second based on conditions.
(At The Quint, we question everything. Play an active role in shaping our journalism by becoming a member today.)