How much risk do Kohli, Pant and KL Rahul take when they attack in the IPL?
Nathan Leamon: 'Morgan is clinical, sharp, bright. He's an analyst's dream captain'
England cricketers at the IPL: Curran vs Curran, Moeen Ali's return, and the Stokes setback
Search for consistency continues as flawed CSK and Rajasthan Royals face off
Kohli vs Rohit: what the numbers say about their IPL impact
In the 54 matches in which he has bowled for England in a T20, Adil Rashid has opened the bowling four times. All four were in the recently concluded five-match series against India. While opening with a spinner in the powerplay is no longer novel in the shortest format, this move was prompted by specific knowledge: googly-wielding legspinners spell trouble for members of India's top order.
Rashid's promotion to open the bowling to counter Kohli and Sharma was the most recent instance of match-ups being used in T20 cricket. In Tests, each strategic play unravels over a long time. In contrast, because time is so limited as a resource in T20, each ball is a substantial determiner of the result. Teams look to optimise every moment to squeeze out the tiniest advantage, making T20 the format where gameplay is most closely "managed".
Using the bowler who takes the ball away from a batter, or sending a left-hander in ahead of schedule to counter a certain bowler can be the ten-run difference that massively tilts a match in your team's favour.
In Tests, "how" you execute is important, while in T20 the "what" and "when" gain equal importance because each play has a major bearing on the course of the game.
With match-ups attaining ubiquity in the T20 landscape, it is important to look at statistics contextualised by various batter-bowler combinations. It is well known that the ball turning in to the batter is advantageous for him. Do the numbers bear that out? If you look at the baseline run rate and dismissal numbers from the last three years of the IPL, they do.
The following table shows you the run rate (runs per ball) and dismissal rate (probability of being dismissed) for left- and right-hand batters against different styles of bowling. (Left-arm wristspin is excluded because of very small sample sizes.)
For left-hand batters, their overall dismissal rate facing offbreak bowlers is 4.3 compared to just 3.6 versus slow left-arm. The run rate is also lower against offspinners, by 0.26 runs per ball. Similarly, for right-handers, the runs-per-ball figure is almost 0.1 runs higher when facing offspinners as compared to against legbreak bowlers and slow left-armers, both of whom take the ball away from right-handers.
Legspinners concede fewer runs to right-handers and are also likelier to get left-handers out. This can be illuminated by further splitting their performance by innings phase. Phase one is the powerplay (overs 1-6), phase two the middle overs, and phase three the death overs (17-20).
The table above shows that right-handers play legspinners more conservatively in the middle overs, possibly "playing out" the dangerous match-up while conserving wickets for the end overs. Left-handers try to utilise the advantageous match-up by going harder in the middle overs - scoring quicker but also getting out more often.
It's a similar story when you look at slow left-arm numbers by phase. In the powerplay, right-handers score much slower compared to left-handers and get out more frequently. In the middle overs, they moderate their approach, scoring slowly while preserving wickets. In comparison, left-handers score faster but get out slightly more often.
The data shows that match-ups work in a broad sense, but what happens when you look at players individually? Jasprit Bumrah and Bhuvneshwar Kumar are both classified as right-arm fast bowlers, but they execute their skills very differently. A right-arm seamer is expected to perform at a certain level versus right- and left-hand batters, but how much does an individual deviate from that baseline?
This can be quantified by dividing their rates of conceding runs and taking wickets by the average runs per ball and wicket probability for each match-up. For example, right-hand seamers overall concede 1.27 runs per ball to right-hand batters in the powerplay while picking up wickets 3.61% of the time. In comparison, Bumrah concedes only 1.1 runs per ball and 4.1% of his deliveries get wickets. We can condense these facts into two simple ratios that tell us how well a bowler (or a batter) performs compared to a particular match-up in a given phase of the innings.
Match-up Run Index (MRI) = (runs per ball by a player for given match-up) / (overall runs per ball for given match-up)
Match-up Dismissal Index (MDI) = (dismissal rate for a player for given match-up) / (overall dismissal rate for given match-up)
An MRI value of 1 means a bowler is as expensive as the average bowler of his kind for a given match-up. A value lower than 1 means he is economical. On the contrary, a higher MDI value than 1 means he is more likely to pick up wickets given that match-up. Continuing from our example, for Bumrah in the powerplay, the MRI is 0.86 (1.1/1.27) and the MDI is 1.14 (4.1/3.61). Here is a breakdown of Bumrah's performance on these metrics:
From a strategy perspective, this shows that Bumrah is exceptionally miserly versus left-handers in the powerplay but not a great wicket-taking option. He is exceptional against both batting styles in the middle overs, and especially effective against left-handers in both run-saving and wicket-taking skills.
Because spinners work with lateral deviation off the pitch, match-up indices are much more relevant for assessing their roles. Here is the same match-up-based performance table for Yuzvendra Chahal, which shows that he is a defensive option compared to other legspinners in the powerplay, but a wicket-taking one in the middle overs, with MDI values of more than 1 against both left- and right-handers, which means he is better at taking wickets than the average legspinner against both batting styles. In terms of economy he is almost as expensive as the average leggie to both kinds of batters (MRI values close to 1), but he is a lot more expensive against right-handers in the death overs.
Splitting open a batter's performance in terms of MRI and MDI is also useful - it shows their relative strengths against particular bowling styles. For instance, here is Kohli's record in the powerplay and middle overs the last three years:
England's decision to bowl Rashid to Kohli is vindicated, albeit with a small sample size. Kohli scores at the par rate for a right-hander facing a legspinner in the powerplay, as evidenced by his MRI of 1, but with an MDI of 1.15, he is likelier to get out than the average right-hander.
But a closer comparison within Kohli's own record split by match-ups reveals that his real kryptonite might be offbreak bowling. In both the powerplay and the middle overs, he scores slower and gets out slightly more frequently than the average right-hand bat versus offspinners. He falters in a match-up that should be advantageous to him.
Last year AB de Villiers, Kohli's partner in the Royal Challengers Bangalore middle order, was shunted down the line-up to avoid facing legspinners, but he has an MRI of 1.09 and an MDI of 0.79 facing that style of bowling in the middle overs in the past three seasons, which signals that he is less likely to lose his wicket to them compared to the average right-hander.
Sharma, Kohli's partner in the Indian top order, has scored nine runs for two dismissals against legspin in the powerplay, but plays it much better when he's settled in the middle overs, with an MRI of 1.06 and an MDI of 0.55.
Different varieties of spin to differently handed batters are match-ups often used by bowling sides. To find out who is the best at run-scoring and wicket preservation for a match-up, we can calculate the MRI and MDI values for each batter in every phase and take a weighted average of these values to find a combined MRI and MDI for a batter.
For instance, the following graphic shows the average MRI and MDI values for all batters who have faced 60 or more balls from legspinners in the last three IPL seasons. The average MRI and MDI account for the match-up and the expected scoring rates in each phase of the innings. Both batting hands can be combined on one plot because the MRI and MDI already account for match-up strength.
An MRI of over 1 and an MDI of under 1 are better for a batter; a value of 1 means the player is average.
The best batters are in the lower-right quadrant. Nicholas Pooran with his middle-overs aggression and Chris Gayle with his disdainful six-hitting are the best against legspin. A bunch of right-hand openers, Mayank Agarwal, Prithvi Shaw and Robin Uthappa, form a high-risk high-reward group in the top-right quadrant with high MRI and MDI values. Surprisingly, Krunal Pandya occupies the dreaded top-left quadrant, which implies slow scoring and a high risk of losing your wicket.
How do batters do against offspin? David Warner outshines his left-hander peers in terms of strike rate and preserving his wicket, while fellow southpaws Gayle and Ishan Kishan are weaker than the average left-hander against offspin when it comes to striking the ball. Hardik Pandya is in a league of his own, with a high MRI and low MDI. MS Dhoni manages to not get out too often, but fails to score against offspin, his numbers heavily influenced by his match-up against Sunil Narine, who himself perches on the far right of this plot, fulfilling his role as an attacker of spin who does not need to value his wicket too highly.
Plotting the MRI and MDI values summed across phases for a bowler can tell us the kind of role he should play in a bowling attack. As an example of how this can be used, the following plot shows the aggregate MRI and MDI values for spinners who have bowled more than ten overs to left-hand batters in the last three seasons of the IPL. A higher MDI and a lower MRI is better for a bowler.
Players in the bottom-left quadrant are holding bowlers who concede fewer runs than one would expect given the match-ups they face, but who are less likely to get wickets. Such a bowler could be brought on as a defensive play to stem the flow of runs and force the batter to "play out" his overs, as teams have tended to do against Rashid Khan.
The tactic of using Washington Sundar as a run-stopper in the powerplay is another great example visible on the plot. Moeen Ali has a small sample size of 108 balls over three seasons, but his high MDI indicates he fares well in comparison to the average offspinner against left-handers.
Here is the same plot for spinners bowling to right-handers:
Surprisingly, R Ashwin is better bowling to right-handers than to left-handers in T20, opposite to his Test bowling strengths. Narine too fares better against right-hand batters. The four best legspinners - Rashid Khan, Chahal, Rahul Chahar and Amit Mishra - are expectedly in the top-left quadrant. Krunal Pandya was slightly high on the wicket-taking MDI against left-handers, but becomes a run-saving bowler facing right-handers.
This method of summing MRI and MDI values over different phases is an attempt to integrate context into raw cricket numbers. The aim is to split the ball-by-ball records of each batter or bowler by the phase of the innings and the match-up, and then scale their run rate and dismissal rate by the par rates for that "context".
This adjusts rudimentary statistics by accounting for what the average player does against the same type of bowler. We can then take averages of these scaled numbers to find combined statistics, and then calculate MRI and MDI values for each combination of phase and match-up. We can then add these numbers up across phases and batting styles to get overall MRI and MDI values for each player. This pair of numbers tells us their run-scoring/saving and wicket-preserving/taking ability while accounting for the handedness of the batters and the style of bowler.
The concluding plots show aggregate MDI and MRI values for both batters and bowlers in the last three seasons.