Before the dust has settled entirely on Pakistan's well-merited win in the 2017 Champions Trophy, we'd like to reflect on the circumstances leading in to the qualification for the semi-finals. ("We" are the inventors of the Duckworth-Lewis method.) We'd like to stress, however, that what we have to say is not specific to targets reset by the DLS method (as it is known now): the same problem would have occurred under any of the previous or other current target-resetting methods. Because of our continuing interest and concern for the welfare of the game and the obtaining of fair results, we have something to say on the not-unusual circumstances of one particular match in that competition.
You will remember that the hosts England played their best cricket in the early stages, winning three matches, and were top of Group A. In second place were Bangladesh, which, even accounting for their growth as an ODI side, was a surprise to most.
You will also recall that the weather in England and Wales at the time was variable. Three matches were decided by the DLS method and two were declared "no result", as both sides hadn't received the minimum of 20 overs. Some teams, such as Australia, suffered from rain interruptions far more than others.
In Group B, for instance, Sri Lanka played all three matches uninterrupted. In Group A, each of Australia's games was affected by rain. They lost to England on the DLS, though England were well ahead of the par score when the match was abandoned. Their game against New Zealand was abandoned, with Australia 53 for 3 after nine overs in their chase (this was a game already reduced to 33 overs). And their game against Bangladesh was also deemed a no-result because 20 overs couldn't be bowled in Australia's innings; they were 82 for 1 after 16 overs, chasing Bangladesh's 182. With two no-results and a defeat, Australia exited the competition.
We believe, however, that there are compelling reasons, within the existing ICC ODI regulations, for Australia to have been awarded victory in that game, thus earning two points and qualifying for the semi-finals instead of Bangladesh.
The ICC's own rules are not being applied appropriately in such cases. Using the ICC's ODI Rule 12.4.2 we hope to explain why. We emphasise that we are not suggesting there should be an additional rule, or even a rule change, but only a different application of the existing rule. The rule states:
12.4.2 Delayed or Interrupted Matches
a) Delay or Interruption to the Innings of the Team Batting First
i) When playing time has been lost, the revised number of overs to be bowled in the match shall be based on a rate of 14.28 overs per hour, which is inclusive of the provision of drinks intervals, in the total remaining time available for play.
ii) The revision of the number of overs should ensure, whenever possible, that both teams have the opportunity of batting for the same number of overs. The team batting second shall not bat for a greater number of overs than the first team unless the latter completed its innings in less than its allocated overs. To constitute a match, a minimum of 20 overs have to be bowled to the side batting second, subject to a result not being achieved earlier.
Bangladesh's total of 182 was well below the ODI average, now around 260. In such a match the team batting second (called Team 2) often wins with many overs to spare. But on this day, rain was forecast. After 15 overs, with their captain, Steven Smith, at the crease, Australia were 82 for 1 and drizzle had started. The par score at this point was 37, so they were 45 runs ahead. Smith realised that to speed up the game and get 20 overs in, blocking was a good tactic, and so at 83 for 1 in 15.1, he blocked the next five balls. It was to no avail, however, as after 16 overs the rain became too heavy and the umpires took the players off the field.
At this point, according to the protocol in place, the match manager for the DLS would have produced the table of possible targets that shows how the target reduces as the number of lost overs increases. The contents of the table relevant for the stoppage in this match are given below.
You can see that as overs were lost (from around 8pm that day) the target reduced gradually. ESPNcricinfo reported that after a break in the rain and an inspection, the umpires declared there would be a restart around 8.30pm.
The operating protocol would have converted the time lost into seven lost overs, so according to this table the revised target was to be 166 for a 43-over innings. But the rain returned, the restart didn't occur and overs continued to be lost. No more than 30 can be lost, after which a no-result is declared. You can see, however, that once 28 overs had been lost, Australia, at 83 for 1, had already achieved the revised target of 79. As this happened before more than 30 overs had been lost, in accordance with the ICC rule, they should at that stage have been declared the winners.
However, ICC protocol for umpires in interrupted games appears to be inhibiting the application of the rule in this manner. It seems that the umpires first decide whether conditions are deemed fit to play, as they did for the 8.30pm planned restart in this case, and only then is the revised target assessed. This is what we call a static view of the protocol. We believe that a more dynamic view of the process would be fairer.
In other words, the revised target should be continuously under review as time is lost, which equates to about one lost over for every four minutes deducted. And so with about eight minutes left before the official cut-off time (which was declared as 9:59 pm that evening) the revised target of 79 for the loss of 28 overs would have been already achieved. But as rain was still falling, the umpires had decided that play could not resume and had no need, according to current protocols, to consult the table of possible targets. Since fewer than 20 overs had been bowled, there was no result and Australia were eliminated.
We shall return to this match after looking at a more clear-cut scenario that could easily have occurred in the Women's World Cup only a few weeks later. We'll first describe the actual match situation and then adjust it to produce what would have been an extremely embarrassing situation, as well as being grossly unfair.
In the game between New Zealand and Pakistan in Taunton, Pakistan were bowled out within their 50 overs for a modest 144. The weather was fine with no rain forecast and New Zealand quickly reached 143 for 1 in 14.4 overs. In trying to hit the two remaining runs to win, Sophie Devine, on 93, was caught and bowled by Nashra Sandhu. In the event, New Zealand completed the victory off the subsequent ball, with 35 overs to spare.
However, here is a plausible variation to the actual event. Suppose New Zealand's haste in wanting to achieve their target was due to a forecast that rain would arrive some time into their innings. Hence they wanted to complete the win as quickly as possible, even within the 20 overs according to the current interpretation of Rule 12.4.2 (ii). Suppose now that rain began to fall significantly around the start of the 15th over. Being conscious of the 20-over rule and that they might be denied victory, imagine Devine trying to hit the winning runs that over but falling off the fifth ball as described. As often happens after a wicket falls, the rain becomes too heavy and so, unfortunately but correctly, the umpires decide play cannot continue. And it soon becomes clear that the prospects of resumption are not good.
At 143 for 2 in 14.5 overs, New Zealand were 105 runs ahead of the DLS par score of 38. The table of possible targets for this match (not supplied) would show that for the loss of just two overs New Zealand would have already achieved the revised target of 142. With the early closure of Pakistan's innings there would have been many hours left before the cut-off time to play the 5.1 more overs needed to make a valid match. Under the current protocols everyone would be sitting around for all this time to see if the rain stopped and the ground could be readied for play to resume. Then, and only then, would the umpires have assessed the time lost and the reduced the number of overs for the New Zealand innings. And this could have been as few as 5.1 overs and so only about 20 minutes before the cut-off time. Only then would it have been announced that the revised target had already been achieved and that New Zealand would have won without any further play required.
This static process is wrong and it could be very embarrassing for the game's administrators. If play couldn't restart earlier, there would be no avoiding having to wait until overs started to be lost, but after that point we can easily see that once just two overs had been lost there would be no more cricket at all. That could mean everyone waiting around for up to two more hours unnecessarily. Surely the game would be better for viewing the revised target dynamically and declaring New Zealand the winners as soon as just two overs had been lost instead of having to wait those hours to see if play could have resumed before the cut-off time? And this could be done whether or not conditions were fit for play.
If play could have resumed, it wouldn't have needed to, because New Zealand would have already won. The remaining 32 or so overs were not going to be needed and Pakistan would not have been given any more overs to bowl, so why should it matter whether or not conditions were fit to play? Such a situation could be extremely embarrassing and any credibility in the current protocol would be completely lost. Worst of all, a no-result would have been unsatisfactory and unfair on New Zealand.
The obviously fair and logical outcome we advocate could only be obtained if the current protocol were interpreted such that the target revision process was a dynamic one based on the table of possible targets. In other words, when overs start to be lost, an imaginary hourglass, calibrated with the figures from the table, is upturned. As the sands run out, the reducing revised target is shown at the level of sand remaining. If it reaches the "already achieved" point before the sands of time have all run out, then the team batting second are the winners. In our example New Zealand would have been declared the winners after two lost overs and no one would have had to hang around waiting to see if the umpires decided if play could resume.
Let's return to the actual Australia-Bangladesh game, which unfolded in a similar way to the above scenario, although not quite so emphatically.
The dynamic approach to the target revision process means that as overs are being lost the target is coming down. We can lose a maximum of 30 overs, but you can see that when only 28 have been lost, with six more overs of play possible, two more than the minimum of four to make up the 20, Australia have enough runs to have won. They don't need any of those six more overs. So why does it matter whether or not play is possible in those six overs? Australia should be declared the winners.
The current process obscures this by only looking at the target when play is possible. Since the umpires decided that play could not be resumed for those four more overs a no result was declared - and what we see as an injustice to Australia was enacted. In practical terms the umpires may well wish to abandon a match well before the cut-off time. But if the team batting second have achieved a revised target before the cut-off time, as was the case with Australia, then they should be declared the winners.
Another near example occurred in the Big Bash League, in which the minimum overs per side is five but the logic of teams winning before the minimum overs have been bowled to both sides is identical.
The example took place in a match in December 2012 between Perth Scorchers and Melbourne Stars. Scorchers batted first and were all out for 69 in 15.2 overs. Stars started their reply and reached 29 without loss in two overs when rain interrupted. It stopped in time for the umpires to allocate a five-over innings to Stars - that is, three more overs. Then, and only then it seems, was the target calculated using the D-L software in use at the time to reveal… a target of 20. In other words, they had already achieved this revised target and so the game was over before a restart, Stars having won. Such a situation was unfamiliar to officials, and absurdly, the players came out and one unnecessary ball was bowled before everyone trooped off.
It would appear then that the table of possible targets, as provided below, hadn't been reviewed during the stoppage, otherwise the situation would have been more apparent.
This is what we believe should have happened. It is clear that after 13 lost overs, two fewer than the maximum of 15 after which a no-result is announced, Melbourne, on 29 for 0, had already achieved a revised target of 28, and so there was no need to resume. The result could have been declared earlier. But, as importantly, if the rain had not stopped, Melbourne would still have achieved their revised target before time had run out and so would have been the winners, whatever the weather situation, with five, not three, more overs left.
If this dynamic interpretation were recognised and implemented then it would have the additional benefit of avoiding tactics that, while understandable, are liable to bring the game into disrepute. We mentioned Smith's defensive batting to try to speed up the completion of the 20 overs in the Champions Trophy match. If our suggestion were adopted, and understood, teams would realise in this situation that the tactic is to get as far ahead of par as possible (recognising that losing wickets raises the par score) if the match is likely to be terminated before 20 overs have been bowled. Isn't this what spectators would prefer to see rather than the defensive tactics Smith observed were appropriate for the current protocol? In addition, under the current protocol, it would have been to Bangladesh's advantage to deliberately bowl wides so as to not complete the 20 overs and allow the rain to stop the match.
We think rules that support negative tactics need questioning. Our interpretation would make such tactics unnecessary.
The main reason quoted by authorities in the past for not supporting our dynamic view is that only Team 2 can win this way, and so it is unfair to Team 1. They argue that in cases where Team 2 are facing almost certain defeat - for example, at 100 for 9 in 19 overs chasing a target of 300 - then there is no equivalent way to give Team 1 the victory that the weather might well be denying them without bowling at least one more over. We accept this point but don't see why it should disqualify Team 2 from winning within the 20 overs, for which we have shown such compelling logic.
We argue that this is just one further point of difference between batting first and second due to cricket's asymmetrical playing structure, and it is something, along with all the other inequalities between batting first or second, that captains might take into account in making their decision whether to bat or field on winning the toss.
We stress again that we are not proposing another rule, but a de facto dynamic interpretation of the existing rule in that the team batting second should be declared winners although they have received fewer than 20 overs if they have already achieved a revised target before time has passed for the loss of more than 30 overs. Put simply, having already achieved such a revised target it does not matter whether or not play is possible in those unneeded overs. The batting team don't need them and the fielding team are not going to get them. Consequently, whether the umpires declare conditions are, or are not, playable is irrelevant.
In summary the scenario we are trying to allow for occurs in these specific but not unusual circumstances.
* Team 1's total is below/well below average for the number of overs per side
* Team 2 make a rapid start and are well ahead of the par score (by whatever mechanism this might be derived - it's not specific to DLS)
* Play is interrupted before completion of the minimum overs to make a match
Current protocols instruct umpires first to decide whether play can resume and then work out how many overs should be lost whereupon the revised target is computed. We believe the revised target should be continuously under review and if Team 2 have achieved a revised target before time has elapsed for the loss of more than 30 overs in ODIs then Team 2 can be declared the winners no matter that it might still be raining and they haven't received the minimum of 20 overs. If those remaining overs are not needed why it should matter whether play could take place in them? It doesn't.
Frank Duckworth and Tony Lewis are the originators of the Duckworth-Lewis-Stern method for calculating rain-revised targets