May 15, 2015 DLS or VJD?

In the 2015 World Cup, the Duckworth-Lewis-Stern method, a modification of the D/L system, came into play four times in 49 games. Among the games in which it was used was a low-scoring one between Pakistan and South Africa, in which Pakistan's score of 222 all out in 46.4 overs changed into a target of 232 for South Africa in 47. Three overs were lopped off due to rain during Pakistan's innings, and even though Pakistan were bowled out within 47, they still benefited by nine runs, as the loss of three overs might have affected the way they would have paced their innings. The earlier D/L method pegged the target at 231, while the V Jayadevan method (VJD), which has generally received favourable reviews for its target-setting as well, pegged it at 226.

In this case, the DLS and VJD targets were both pretty reasonable and similar to each other, but there are other instances when they aren't so close. (By the way, in the cliffhanger semi-final, the DLS target was 298, while the VJD one was 300; New Zealand ended up scoring 299 and winning with a ball to spare.) The DLS system has undergone several modifications, and while there have been improvements - especially since the computerised system replaced the manual one - there continues to be areas of concern, especially in the sluggish manner of setting targets in higher scoring games, especially in the 50-over format. The latest modification - championed by Steve Stern, who is the 'S' in DLS - was meant to improve this specific area of the system, but closer scrutiny of the system shows problem areas continue to exist, especially with internal consistency.

The example below illustrates inconsistencies within the DLS system, when comparing results that it throws up for different match situations:

When a team is bowled out for 150, the 20-over target for the team chasing is, according to DLS, 93; the VJD system calculates it to be 91. For a score of 200, the 20-over targets by the two methods are 124 and 118; for a total of 250, the targets are 154 and 142. Till here, the increase in the DLS targets for every 50-run increase in the total is around 30 runs; the VJD method increases it by 27 and 24 runs. However, for a total of 300, the DLS target is 163, which is nine more than the 250 target, and for a 350 total, the DLS target increases to only 174. That means the difference in 20-over targets between a 50-over score of 250 and 350 is only 20 runs - a team scoring 350 will have reason to feel shortchanged, given that the 20-over target is only 20 more than it would have been if the team had scored 250. The 100-run difference in scores converts to an average of two per over, yet the target goes up by one run per over.

The previous version of this system - the D/L method - actually did better in these cases: the 250-run target then was 151, and the target for a 350 total was 181, a difference of 30 runs. In the VJD method, the targets move up quite evenly, and the 350 total generates a target which is 40 more than the 250 target.