Kohli's T20 Problem

Kohli is significantly more conservative in T20 than he is in ODIs

Just as it shows that he is a world class Test and ODI player - arguably the finest of his generation - the record shows that Virat Kohli is an average T20 batsman. This aggravates a lot of people, but it happens to be true, so its worth showing that it is.

When a batsman faces a delivery in a limited overs match, two facts are known. First, the number of wickets in hand for the batting team. Second, the number of legal deliveries remaining in the innings. These are the only two finite resources available to the batting team in a limited overs game. In the first innings, the challenge is to spend these resources to accumulate the biggest possible score. In the second, the challenge is to spend these resources to reach a given target.

The two resources listed above can be reduced to one - balls per wicket in hand (BPWIH). For example, at the start of the 9th over in a 20 over match, if a team has lost 2 wickets, then it has 8 wickets left, and 72 deliveries left. The BPWIH is 72/8 - or 9. If they have 6 wickets in hand, it would be 72/6 - or 12. At the start of the innings, it is 120/10 - or 12. In a 50 over ODI, it would be 300/10 - or 30, at the start of the innings.

The two charts which follow give the distribution of deliveries faced by a batsman in the 1st innings of an ODI and T20, as organised by the BPWIH. The deliveries a grouped by rounding the BPWIH to the nearest integer. So a BPWIH of 10 represents all deliveries for which its value lies between 9.50 and 10.49, and so on.

As you can see, the vast majority of deliveries lie in the 30 or under BPWIH range for ODIs, and 12 or under BPWIH range for T20. 3.6% (or 1 in 27, or 4.5 in 20 overs) of deliveries faced by a batsman in a T20 game involve a BPWIH for 13 or more. 11.3% (or 1 in 9, or 33 in 50 overs) of deliveries in ODIs involve a BPWIH of 31 or more.

The charts also give the average scoring rate achieved by the average batsman (the black line) and by Kohli (the red line), and the distribution of deliveries faced by Kohli (yellow bars) and by the average batsman (gray bars) by BPWIH.

The table below presents the data in the charts above via a rolling average. In the ODI case, a BPWIH range of 10 values (30-21, 29-20, 28-19,…) is used. An identical range is used for T20 - 4 values (12-9, 11-8, 10-7,…). The share of the total balls faced by the average player and Kohli which fall within each BPWIH range is given. The record is considered for BPWIH values of 30 and under (the higher the number, the fewer the resources) for ODIs, and 12 and under for T20.

Finally, the column on the right gives Kohli’s scoring rate compared to the average player’s scoring rate for deliveries in each BPWIH range. If the % is positive, it means that Kohli scores that many percent quicker than the average player, if it is negative, it means that Kohli scores that many percent slower than the average player.

As you can see, Kohli scores at least 11% quicker than average consistently in ODIs. Note that BBB records for ODIs are available only for 21st century ODIs, and overwhelming for ODIs after 2002, so the figures for the average player are contemporary.

In T20, Kohli is scores about 8% slower than the average player for about half the deliveries he faces (deliveries with BPWIH from 12-8 are about 51% of the total).

The striking thing about this is that Kohli bats significantly more cautiously in T20 than he does in ODIs. Part of this has to do with the fact that contemporary T20 batting overall is significantly more conservative than contemporary ODI batting. About 11% of teams are bowled out in the 1st innings of a T20 game, while about 32% of teams are bowled out in the 1st innings of an ODI. But even accounting for this general cautiousness in T20 batting (watching a T20 game where 160 is scored in 20 overs is like watching an ODI in which 210 is scored in 50 overs, but without the exciting jeopardy of the bowlers actually threatening to dismiss batsmen), Kohli is slower than the average player.

In most of his innings, Kohli digs a hole for his team by falling behind the average expected scoring rate, and in a small minority of these innings, he manages to dig his team out of the hole he puts them in. In most cases, the hole he digs keeps getting filled in by the player at the other end. But this is an exercise in futility until Kohli stops digging. This is not unusual though. ODI cricket in the 1980s is replete with examples of this approach. There’s more justification for it in ODIs, since it is basically much harder to survive 30 balls per dismissal than it is to survive 12.

The thing is though, the record presented here shows that there’s no basis to the proposition that Kohli plays in T20 as he does in ODIs. The way he plays in T20 is significantly worse. He’s an all-time great ODI player. And that’s because he scores quicker than the average ODI player.

Scoring rates in T20 are higher, and perhaps Kohli’s game is not suited to keeping up with these rates, and this is why he scores slower than the average T20 player. There could well something be something to this theory. But this doesn’t change the fact that on the available evidence, Kohli is an average T20 player.

There are a lot of views which hold that this type of review does not take into account something called “context”. This review considers all contexts in the proportion in which they actually occur. The role of the anchor is often presented as being distinct. But as the figures show, Kohli does not anchor the ODI innings (which is longer, in which the average bowler is better than the average bowler in a T20 game, and in which the bowling side is more likely to be trying to dismiss the batsman than the bowling side in T20). As the figures also show, the range of situations in ODIs are wider than the range of situations in T20 (1 in 9 deliveries in an ODI is bowled when the batting side has fewer resources than it started the innings with. 1 in 27 deliveries in a T20 is bowled in such circumstances). Situations of greater batting jeopardy are more likely in ODIs than they are in T20.

These justifications which essentially say that there’s something unique to the contexts which Kohli is confronted with which are not presented to any other player are not serious. They seem to be attempt to defend Kohli rather than describe the game as it actually is. The purpose of pointing out that Kohli’s an average T20 player is that the contrary view is widely held and represents a misunderstanding of the nature of the competition.

It remains one of the ironies of T20 that a sizable majority of its fans yearn to find cricket in it and do not recognize the consequences of the fact that compared to 50 over cricket, a wicket is not a scarce resource in T20. The main consequence is that all conventional conceptions of constructing an innings do not apply. But this is not new. The exact same thing happened to ODI cricket in the 1980s. Then, starting with the pinch-hitting opener, new ideas about exploiting resourced in ODIs were inevitably explored. ODI cricket in the 1980s utilized resources inefficiently (much as teams do in T20 today), and someone found a way to be more efficient, and other teams were forced to follow their example. ODI cricket is 4260 matches and 50 years old, and today’s ODI cricket bears little resemblance to the ODI game in the first few hundred matches. It is very likely that the same thing will happen to T20.

T20 is still looking for its Dean Jones, let alone its Viv Richards. Kohli is T20’s Sunil Gavaskar or Geoff Marsh. If you want a picture of a great contemporary T20 player, someone who approaches a Dean Jones of T20, see AB de Villiers.