Limited-overs run chases are a record-keeper's delight. When a team bats first in a limited-overs game, the task before its batsmen is "make as many runs as you can in the remaining overs". When chasing, the equation is as specific as it can ever be in cricket. At every ball the batsman knows exactly how many balls are remaining and how many runs are required. This allows us to ask some interesting questions about players and about the nature of run chases.
I've come to think of Virat Kohli and AB de Villiers as the machine and the polymath of the current era. In limited-overs cricket, and especially in limited-overs run chases, Kohli's mastery often appears frictionless, while de Villiers' most famous stands have been heroic if ultimately unsuccessful. Yet de Villiers has a formidable record in run chases. When it comes to T20, Kohli's record is mixed and his approach appears to be eccentrically old-fashioned compared to the cutting-edge masters of contemporary T20.
This article offers a comparison of these two modern masters in limited-overs chases, and looks at the two players' approaches and capabilities. In the process, it offers a way to think about choices batsmen have to make during these chases.
Kohli's and de Villiers' career records in chases are below. These include all games in which they batted for at least one ball, and for which ball-by-ball data is available.
The table above shows the player's scoring rate and batting average (these represent speed and consistency of run-scoring respectively). It also shows the runs scored at the other end while the player is at the wicket. (In the table above, only runs scored off the bat of the other batsman are considered.) Third, it shows the runs scored in the innings while the batsman was not at the wicket. The record shows that while Kohli's run output in ODI chases is outstanding (average of 67, strike rate of 94), he does enjoy strong support at the other end (44, 87). de Villiers makes 60 on average, at a strike rate of 95, but his average partner manages only 37 and 81. Kohli scores 8% faster than his average batting partner, while de Villiers scores 17% faster than his average batting partner. When they're not at the wicket, their team-mates manage about the same run output. In T20 chases, Kohli seems to prefer consistency (42, 132), while de Villiers seems to prefer power (36, 147).
The ball-by-ball data enables the player's run output from a given delivery to be compared to the required scoring rate as it was at the point when that ball was going to be bowled. For example, suppose 125 are needed in 100 balls and the player takes a single, leaving 124 required from 99. This would increase the asking rate by 0.003 runs per ball. If the batsman scored a boundary instead, it would reduce the asking rate by 0.004 runs per ball. In other words, the net improvement in the asking rate, from the batting team's point of view would be -0.003 runs per ball in the first case. In the second, it would be +0.004 runs per ball. This net change is the scoring rate relative to the required rate for that ball; or, net scoring rate (NSR).
The net scoring rate can be calculated for entire innings for each player. This is the aggregate scoring rate relative to the required rate for each ball of the player's innings. It can also be summarised for each player for each over in a team's run chase, and for the nth ball of the player's innings. These are the two types of summaries that are considered in the remainder of this article. (In each summary, all deliveries available in the record in which up to six runs per ball were required before the delivery was bowled are included.)
The graphs below provide three kinds of information. First, the grey bars provide the distribution of deliveries each player has faced in a given over in the innings during run chases. Second, they provide the player's NSR in that over in the innings. Third, it gives the successively cumulative NSR. This is NSR as aggregated in each successive over. This cumulative figure provides a picture of a player's approach in a given part of the run chase. Remember that when a player scores slower than the required rate, the NSR is negative. When a player scores faster than the required rate, the NSR is positive. And when a player scores at exactly the required rate, the NSR is zero.
The evidence suggests that Kohli adopts a classical approach to ODI chases. Typically, he does not try to keep up with the asking rate in the first 25 overs of the chase. In the second half of the chase, he begins to catch up with the rate, and by the 37th over, his individual innings is typically proceeding at the required rate. In the final 10-12 overs of the chase, if Kohli is still at the wicket, he tends to be a law unto himself and the asking rate ceases to be a problem. He has to be dismissed or he ends up on the winning side.
AB de Villiers is less conservative in the first 25 overs of the ODI run chase. He tends to stay abreast of the required rate for the most part, and about midway through the chasing innings, he tends to score at least as well as the required rate, if not better. However, as the chasing summary at the beginning of this article suggests, his approach is shaped, at least in part, by the relative lack of batting support at the other end compared to a player like Kohli in a strong batting team like India.
In ODI chases neither batsman appears to be daunted by the required rate. Each seems to be able to respond comfortably and make up ground whenever it is lost. In T20, it is a different story. Kohli tends to score about one run per over slower than the requirement during the first ten overs of a T20 chase. He faces anywhere between 2.5 to 3.4 balls per over in each of these overs, so on average, his NSR per ball is -0.321 during these 11 overs. For the next four overs, he scores at the required rate, but cumulatively he is still about ten runs behind the requirement at this stage. After the 15th over, he explodes, but by this time he's making up lost ground, and on average, he does not make up enough to break even by the end of the chase.
By contrast, while de Villiers also tends to score slower than the requirement in the early overs of a chase, he stays within about four runs of the requirement for the most part. Like Kohli, he also tends to score quickly after the 15th over, but he does not have as much ground to make up as Kohli does.
It is worth noting here that while de Villiers' average innings in a T20 run chase lasts 24 balls per dismissal, Kohli's average innings in a T20 chase lasts 32 balls per dismissal. So de Villiers is more likely than Kohli to be dismissed within the first 15 overs of a chase. Having said that, when this happens, his average contribution to the chase tends to have left his team-mates with less ground to make up than in Kohli's case.
Seen in the context of a traditional cricketing conversation (in which dismissal is always a bad thing for the batsman and the batting team), Kohli is more conservative than de Villiers. However, seen from the logic of T20, Kohli could be said to take bigger risks than de Villiers because he tends to allow himself (and by extension, his team) to fall further behind the requirements of the chase for larger parts of the chase compared to de Villiers.
So far we have seen how these two players perform in different parts of the chasing innings. In the next part, we will see how their innings take shape. Note that in the case of the ODI record, the charts are limited to the first 100 balls of each player's innings. The T20 charts show the full length of each player's innings. Among the innings for which the record is available, de Villiers' longest innings in a T20 chase lasted 50 balls, while Kohli's longest innings in a T20 chase lasted 61 balls.
In his average ODI chasing innings, Kohli spends the first 12 balls without any concern for the required rate and allows himself to fall behind the requirement - on average, he falls about three runs behind the required rate. He keeps up with the required rate without making up ground for the next 28-30 balls. Typically, after his 40th delivery he begins to make up lost ground, and by his 60th, he has caught up. After his 60th delivery he's unstoppable. On average, Kohli survives 71 balls per innings in chases. His average contribution in a run chase has an NSR of +1.7. Even considering that he plays in an Indian ODI batting line-up, and those tend to be strong, this is an astonishing record, especially given the length of his average innings in an ODI run chase.
As is the case with Kohli, the required rate in an ODI chase is almost never an issue for de Villiers. He is marginally more aggressive than Kohli and his average innings in a chase lasts 63 balls compared to Kohli's 71. By his 63rd delivery, de Villiers' NSR is +5.0. And yet, de Villiers played in 64 wins and 43 defeats in ODI chases, while Kohli (at the time of writing) has played in 88 wins and 45 defeats in ODI chases.
This adds to evidence that chases are won by teams and not individual players. One is left with the tantalising possibility that perhaps de Villiers might have profited from adopting a more conservative approach in chases. If he managed 63 balls per innings while achieving an NSR of +5.0, how much more consistent would he have been had he shaped his innings in ODI chases in the way Kohli did? Against this, the fact that de Villiers had less support at the other end (average 37, strike rate 81) compared to Kohli (44, 87) has to be considered. This put greater scoreboard pressure on de Villiers.
In T20 chases Kohli's approach appears similar to that in ODI chases. He starts slowly (relative to the required rate), falling about four runs behind the required rate over his first ten balls. Over his next 20 balls (11-30), he falls a further two runs behind the required rate. From the 31st delivery in his average innings, Kohli begins to make up lost ground. He reaches a positive NSR only after his 50th delivery on average. In the average ODI chase, recall that it takes Kohli 60 balls to reach a positive NSR on average. In a 300-ball chase, this makes Kohli a phenomenal chaser. In a 120-ball chase, it makes him a highly risky chaser. Kohli's average innings in a T20 chase lasts 32 balls, by which time, on average he has a -5.8 NSR.
By contrast, de Villiers is largely untroubled by the daunting required rates in T20. His average innings in a T20 chase lasts 24 balls per dismissal, and on average he stays within 2.3 runs of the required rate at all times (his NSR never drops below -2.2). By his 16th ball, de Villiers begins to score quicker than the requirement, and by his 30th he is above the requirement. This approach is less risky for de Villiers' team's prospects than Kohli's approach because even when de Villiers fails (which occurs more often than with Kohli), it happens sooner rather than later, and it leaves his team with less ground to make up compared to Kohli.
The comparison suggests that Kohli tends to treat T20 games as abbreviated ODI chases, while de Villiers is not only naturally more aggressive than Kohli but also far more capable of tailoring his approach to the requirements of the chase than Kohli is. Kohli does not get out as often as de Villiers. This makes Kohli very consistent from a conventional cricketing standpoint. It also makes him extremely reliable in moderately steep T20 chases (typically, those involving targets below 165).
One commonly heard refrain about Kohli is that he averages 50 in all three formats. This is neither here nor there. The average is irrelevant in the T20 game. As Tim Wigmore and Freddie Wilde show in their excellent book Cricket 2.0, a characteristic feature of the first decade of T20 cricket was the inability of batsmen brought up on longer forms of the game to learn to value their wickets less in T20 than they were trained to do in the longer forms of the game. Sachin Tendulkar never fully made this shift. It is not surprising that the best T20 batsmen tend to be players who have not had distinguished careers in international cricket (either Test or ODI). de Villiers remains the solitary exception. Kohli's record shows that perhaps he values his wicket excessively in T20.
Why does Kohli approach his T20 innings in this way? If the above record can be seen from publicly available data, it is unlikely that the sophisticated analytics departments in India's national team or in Kohli's IPL franchise are unaware that a conventional conservative batting approach is a highly risky approach in T20.
Does capability shape approach? Or are Kohli's choices tactical? There is some evidence to suggest that Kohli can, at times, hit boundaries at will. After the 15th over of T20 innings (this is not restricted to chases), Kohli hits 27% of his deliveries to the boundary, compared to 26% for de Villiers. de Villiers hits a higher proportion of sixes compared to Kohli. Kohli's average boundary is worth 4.7 runs, while de Villiers' is worth 5.2 runs in this period. During the first 15 overs of a T20 innings, de Villiers hits 17% of his deliveries to the boundary, while Kohli hits 14% to the boundary. This gives de Villiers a scoring rate of 138 runs per 100 balls faced during his first 15 overs compared to 124 for Kohli.
In the first ten balls of his innings, de Villiers hits 16% of his deliveries to the boundary, compared to 13% for Kohli. From the 11th to the 20th ball, de Villiers manages 19%, compared to 16% for Kohli. After 20 balls, de Villiers hits 24% to the boundary to Kohli's 18%. While it is clear that there are parts of the innings when Kohli can find the boundary at will, these periods tend to occur not only late in T20 innings but also late in Kohli's innings. Whether this is a matter of approach or capability is an open question. The evidence does suggest that Kohli might try to play T20 differently in the new decade compared to how he did in the 2010s.
In conclusion, consider where Kohli and de Villiers sit among the 20 most prolific batsmen in T20 chases. Note that in the table below the NSR and balls faced are given per innings and not per dismissal. REQ SR gives the average required scoring rate facing the player in a T20 chase. SR gives the player's scoring rate.