When cricket teams lose, the tendency among supporters is to look for scapegoats. These tend not to have anything to do with the team's competitiveness, but rather focus on "respectability". Thus, when India lose a Test match or Test series, attention is inevitably drawn to the batters, though it is the bowlers who couldn't bowl the opposition out twice. In T20, the blame tends to be directed at the batters who score the fewest runs, though it is the speed of run-scoring that determines competitiveness.
In T20 matches the field is spread, and so singles are on offer pretty much on every ball a batter faces. So producing a high average is not very difficult (compared to doing so in Test cricket or even ODI cricket) if a player is prepared to score slow enough.
Virat Kohli's scoring rate after 4008 runs in T20Is stands at 137.96. Let's say it is 138 runs per 100 balls faced. Compared to other players, that appears to be a healthy scoring rate. That is until you consider how long it takes him to achieve that rate. This is given in the table below. Kohli's average T20I innings lasts 27.1 balls, from which he produces 37.5 runs. The same figures for the next 14 most prolific T20I batters for India are in the table below Kohli's figures.
Immediately below Kohli in the table are India's current opening pair. Let's say that they both score at the same rate as him. Except, that they survive 20 and 24 balls per innings respectively compared to Kohli's 27. This means that they get to that scoring rate quicker. The last column below gives the difference between Kohli's scoring rate and that of other players after the average number of balls of the other player's innings. Kohli scores 5.7% slower than Rohit Sharma, 5.2% slower than KL Rahul, 27% slower than Suryakumar Yadav, and so on.
The ball-by-ball record of T20 internationals gives each player's average score after each ball of their innings. All five other batters in the current India line-up accelerate faster than Kohli does. This means that they attempt boundaries more frequently than Kohli does, and that's why they get out earlier more often than he does.
The temptation, especially if one is a fan of Kohli, is to ask, "Why focus on Kohli, who made more runs than anybody else in the tournament?" The above is the answer. T20 is not a game for accumulators. It is a game for plunderers.
Teams have ten wickets to spend over 120 balls - 12 balls per wicket, compared to 30 balls per wicket in ODIs, and roughly 62 balls per wicket in Tests (the average Test innings lasts just over 100 overs in the modern era). So we can say that for a player's innings to not be considered a failure, the player should not be dismissed in their first 12 balls. But we also don't want the player to score slowly just to survive 12 balls. Which is why we also use the expected runs from that delivery in the comparison.
The expected runs from each ball are estimated as the average runs scored from a given delivery. This is defined in terms of three variables at the time the delivery is bowled: (a) the number of balls remaining in the innings, (b) the number of wickets in hand, and (c) the innings scoring rate at the start of the delivery. For example, after 50 balls, with two wickets lost and a current scoring rate of six runs per over in T20, the 51st ball of the innings is expected to produce 1.061 runs. Given a current scoring rate of nine runs per over, the same delivery is expected to produce 1.304 runs. After 80 balls, with two down, a current scoring rate of nine runs per over produces an expected-runs estimate of 1.518 runs per ball.
Note that these are actual average runs from such deliveries available in the record. As more and more T20 fixtures are played, this expected runs record will become "smoother". An alternative approach would be to train a linear model, which uses the same three inputs and estimates outputs for a given (balls, wickets, economy) input, but here I use the average runs from deliveries in the T20 record.
We can now organise T20 innings into four categories:
1. Failures: The player is dismissed within 12 balls and scores fewer than the expected runs from the balls faced.
2. Cameos: The player is dismissed within 12 balls and scores more than the expected runs from the balls faced.
3. Successes: The player faces at least 12 balls and scores more than the expected runs from the balls faced.
4. Under Par: The player faces at least 12 balls and scores less than the expected runs from the balls faced. The distribution of Rohit Sharma's T20 international innings according to the classification above is in this graph.
The distribution of innings across these categories in all T20 internationals for India's top six batters in the 2022 World Cup is below. Kohli plays Under-Par innings more frequently than any other player. Note the high rate of Failures and Under-Par innings for Hardik Pandya, who bats later in the innings than players who regularly bat in the top four, and so is at the crease when the expected runs from each delivery are higher than they are in the first half of the T20 innings.
When only 120 balls are available to the team in the innings, acceleration in run-scoring is as significant as scoring. Kohli's scoring rate in his first 27 balls (the number of balls he faces in his average innings), is 128.6 runs per 100 balls faced. Rohit Sharma's scoring rate in his first 20 balls is 127.6. KL Rahul's scoring rate in his first 24 balls is 134.1. Note that this comparison provides a picture that is distinct from the one provided in the first table in this article. In that table, scoring rates are compared relative to dismissal rates (X balls), with faster dismissal rates indicating propensity to take greater risks earlier. Rohit's scoring rate in T20Is is 139 runs per 100 balls faced, and he is dismissed once every 19.8 balls. But if you consider only his first 20 balls his scoring rate is 127.6. This provides a picture of different rates of acceleration between these players.
In the table above, readers will also note that while one in four of Pandya's innings in which he lasts less than 12 balls are Cameos (Failure and Cameo percentages add up to 45.8, and Cameos are about 25% of that total). One out of five of Kohli's innings of this type are Cameos (4.7% Cameos, 18.7% Failures). KL Rahul starts even slower than Kohli (4.7% Cameos, 23.5% Failures), but if he lasts 12 balls, the majority of his 12-ball-plus innings are Successes, while only two out of five such innings by Kohli are Successes.
However the record is considered, it shows that Kohli is a slow-scoring T20 player as a rule. It is only in the slog that he opens out. A consequence of this is that out of the 120 deliveries available, a large number go uncontested, and are unavailable to other batters. India's problem here is not as acute as Pakistan's. Pakistan have both Mohammad Rizwan and Babar Azam who have T20 scoring profiles similar to Kohli's. Nevertheless it remains a problem for India, much as Kane Williamson's difficulties remain a problem for New Zealand.
There is a lot of discussion in the media about India needing to set up separate squads with separate coaches for each format. As these questions are considered, one issue would be whether players with the scoring profile of Virat Kohli or Kane Williamson are good fits in T20 top orders.
India may not be able to match England's versatility in the short run (England could field six allrounders in their XI in the T20 World Cup final), but they could potentially front-load their hitting talent and use someone like Kohli at No. 6, as insurance, instead of using him to anchor the innings from one end at the top of the order. This will ensure the necessary acceleration, and provide the assurance of there being a backstop in case of early wickets (which is inevitable from time to time). This will reduce the frequency of Under-Par innings from India's top order and raise the ceiling for the scores India can produce.
If the idea is, as many observers have noted, that India need a reboot, then part of this reboot ought to be to take seriously the proposition that T20 is a contest of efficiency. This will require measurements that go beyond basic scoring rates, which can be deceptive, especially for top-order T20 bats.