Why T20 hitting is not just a more risky version of batting

Batting is the art of accumulating runs safely. Its measures of merit say that a longer innings is better than a shorter innings, a bigger score better than a smaller one, a higher average better than a lower one. In terms of ESPNcricinfo's control measurement, a higher rate of control is better than a lower rate of control for the batter.

Hitting is different. It is the art of plundering each delivery for as many runs as possible. Its measures of merit must be different from those for batting. The hitter's technique is designed to make it possible for them to consistently hit the ball to the parts of the boundary that are not defended by a fielder. Among batters, if two score at the same speed, the one who averages more is the better batter. Among hitters, if two score at the same speed, the one who averages less is the better hitter, because this player achieves the said speed earlier in their innings.

Ruturaj Gaikwad described the distinction between hitting and batting recently: "[In T20] you have to be ready for each and every ball and have three particular options in your mind for each ball. Then, all of a sudden, to come [into] red-ball cricket, where you don't really have to look for runs, you have to focus on staying on the wicket."

This is why hitters play innovative and unusual shots. Given the point is to score as quickly as possible, the hitter's orthodox shot is one that evades the fielders. Batters try to defend the top of off stump as well as possible against three slips; the orthodox shot for a batter is one that avoids dismissal.

This distinction between hitting and batting lies at the heart of what has been described in several excellent articles in these pages as the Indian T20 squad's attempts to be "more aggressive" under their new leader, Rohit Sharma. Batters need to survive and are highly selective about the balls they choose to attack. Hitters attempt boundaries as often as possible and consequently are more likely to score a boundary off the average delivery, but they are also more likely to get out on the average delivery.

Does this mean that hitting is simply a riskier version of batting? The answer is yes in one trivial sense - hitting results in dismissal far more frequently than batting does. In the last 20 years a Test wicket has fallen, on average, once every 62 balls, while a T20 wicket has fallen once every 18 balls. The average Test innings lasts about 100 overs; a T20 innings lasts only 20 overs.

In this article I suggest that something more basic is at play in the distinction between batting and hitting. To consider it in terms of risk is to privilege the batter's orthodoxy, and to describe what hitters do through the lens of batting.

If all batting is about selecting the right delivery to score from so that runs are made safely, then there should be evidence in the record for players trading run-scoring speed for survival at the wicket. This relationship - the more a player prefers to play shots, the more frequently the player is dismissed - should be viable in all forms of cricket, including T20.

One way to measure this is to count how often a boundary is attempted, for that is where the risk of dismissal is maximal. Hitting the ball to the boundary typically involves swinging the bat harder, meeting the ball earlier, committing to the shot earlier, playing further away from the body.

A measure available for this is the one that describes whether or not an attacking shot has been attempted. But this is distinct from whether a boundary has been attempted. For example, if the bowler has both long-on and long-off positioned relatively straight, and fine leg and third man up in the ring, then hitting the ball hard straight down the ground is an attacking shot but it cannot really be called a boundary attempt. Given this field, an attempt to send the ball behind square is a boundary attempt. To say whether or not a boundary has been attempted requires considering the placement of the field.

In the absence of direct measurements, let's consider two indirect measures. First, look at how often a player scores a boundary by calculating the balls faced per boundary. Second, consider how often a player is dismissed per boundary scored by calculating the boundaries scored per dismissal. Note that a boundary is a four or a six.

The chart below shows the relationship between these two measures for all players. (To qualify for a format in this chart, a player must have at least 4000 runs in that format between 2003 and 2022; T20 figures include those from T20Is too.)

In Test cricket (blue dots), a greater range of boundary-hitting prowess is viable. Even if one thinks that Kraigg Brathwaite (24.6 balls faced per boundary) and Virender Sehwag (7.9 balls faced per boundary) are not equally good players, consider that Shivnarine Chanderpaul averaged 20.5 balls per boundary, Azhar Ali 24.0 and Cheteshwar Pujara 18.6. At the other end of the spectrum are Chris Gayle, with 9.9 balls per boundary, David Warner (11.1) and Brian Lara (8.2).

The less frequent the boundary-hitting, the fewer boundaries per dismissal, and in Tests and ODIs, the longer the player's average innings in terms of balls. In Tests, Sehwag was dismissed once every 60 balls and Chanderpaul once every 124 balls. In ODIs, Jos Buttler, Adam Gilchrist, Sanath Jayasuriya, Sehwag and Gayle scored boundaries once every six to eight balls, and were dismissed once every 33-42 balls, while at the other end, Joe Root, Ross Taylor, Kane Williamson, Babar Azam and Virat Kohli score boundaries every 10-13 balls and are dismissed once every 56-63 balls. Being more selective in Tests and ODIs produces a longer stay at the crease, because scoring runs in smaller denominations is useful in Test and ODIs (scoring rates are typically slower than a run a ball).

In T20, as the chart shows, this relationship between the frequency of boundary-hitting and the frequency of dismissal is weakened. There are only six players in the T20 record who average five or more boundaries per dismissal. Their frequency of boundary-hitting ranges from 4.6 balls per boundary for Gayle to 6.8 balls per boundary for Mohammad Rizwan. At the other end of the spectrum are nine players who achieve fewer than three boundaries per dismissal. Their frequency of boundary-hitting ranges from five balls per boundary (Shahid Afridi, Thisara Pereira) to between seven and eight (Shakib Al Hasan, Mahmudullah, Ravi Bopara). At both ends of the T20 spectrum there are, essentially, players who are better (Gayle, Warner, at the top end) or worse (Rizwan) at scoring boundaries. In T20 the room to attempt alternative approaches to run-scoring simply does not exist. With the result that there are essentially more or less selective boundary hitters, and among these, the better ones manage more boundaries per dismissal and the less able ones manage fewer boundaries per dismissal. The more selective players score slower than the less selective ones. In a fast-scoring format like T20, this is not viable in the same way that it is in Tests and to a lesser extent in ODIs.

Readers should note that in T20, the players considered in this article score 60% of their runs in boundaries. The corresponding figure for ODIs is 47%, while for Tests it is 50%. Further, in Tests these players have hit one in every 15.5 balls to the boundary. The corresponding figure for ODIs is one every 10.7 balls, while for T20s and T20Is it is one every 5.8 balls. Brathwaite and Azhar have scored 40% of their Test runs in boundaries, while Sehwag scored 63% of his Test runs in boundaries. Among T20 players this range goes from 48% (Bopara, Steve Smith, Mahmudullah) to 75% (Chris Gayle and Andre Russell).

A second way to consider this idea is to look at the modal over for players for each format - that is, the over in the innings in which the player is at the crease most often. In any innings, a batter is at the crease in a specific range of overs. Considering all innings by a batter, we can calculate the over (or overs) in an innings in which the batter is most frequently at the crease. This is the modal over. In Hardik Pandya's brilliant 71 off 30 balls in Mohali recently, his modal over was the 15th, in which he was on strike for seven balls.

For example, in ODIs the modal over for the No. 3 batter is the tenth over. For No. 4 it is the 19th, for No. 5 the 27th, for No. 6 the 37th, for No. 7 the 43rd, for No. 8 the 45th. In T20s, the modal overs are: No. 3, 5th over; No. 4, 10th; No. 5, 14th; No. 6, 16th, No. 7, 18th; and No. 8, 19th.

The graph below displays this data for all formats in the ball-by-ball record for players batting in positions three through eight. Note that the T20 and T20I curves are nearly congruent. Openers are ignored because they always start their innings in over 1, and so for them that is the modal over, the one in which they are guaranteed to be at the crease.

The horizontal axis plots the over of the innings as a percentile rank. For Test cricket, the first 100 overs are considered, since this is approximately the length of the average Test innings (a Test wicket falls once every 62 balls). When that scale is applied to T20s and T20Is, the 90th percentile over is the 18th over. In ODIs, the 90th percentile over is the 45th over, and in Tests it is the 90th over. The vertical axis shows the frequency with which each over is the modal over.

In Tests (and to a large extent in ODIs), the frequencies flatten out. After the first 15 or so overs of the innings, any over is roughly as likely as any other to be the modal over for Nos. 3 to 8. In T20s this is not so. The later in the innings we are, the modal overs are bunched closer together (from the No. 5 position onwards). This is because in T20 batters run out of deliveries faster than they run out of wickets.

The essential distinction between Test cricket and T20 is that in Test cricket, wickets are the scarce resource. There is a relative abundance of deliveries, accommodating a range of batting approaches, while in T20, deliveries are the scarce resource, since there is a relative abundance of wickets (ten) over only 120 deliveries. This is not a particularly counterintuitive insight but its implications have been resisted in T20 international cricket and in the way T20Is are discussed.

Two recent high-profile games involving Pakistan that featured two of the six players above who managed at least five boundaries per dismissal by being significantly more selective than average about the balls they choose to hit - Babar Azam and Mohammad Rizwan - show the problems with this approach. In the T20 World Cup semi-final of 2021, Pakistan batted first and made 176 for 4. They got to 47 for 0 in the powerplay but in the next seven overs managed only 45 for 1, despite having ten wickets in hand. In the 2022 Asia Cup final Pakistan, chasing 171 to win, got to 37 for 2 in the powerplay. In the next seven overs they managed 54 for no wicket - less than eight runs an over.

From a batter's point of view, 45 for 1 and 54 for none are both excellent returns in a seven-over period because wickets have been preserved. But under the hitter's orthodoxy, these scorelines reflect bad play. The hitter's orthodoxy says that team that had wickets in hand should have scored more runs if they didn't lose wickets, or lost more wickets if they didn't score more runs. Forty-five for one (one four, one six) and 54 for none (five fours, zero sixes) in the middle third of a T20 match in which about 170 runs were made in each innings reflect excessively selective attempts at scoring boundaries. The contrast between Pakistan's approach and that of their opponents in these games is evident in the table below. What marks Pakistan's play is an unwillingness to take chances; if they had taken chances, it would have manifested itself either in quicker runs or more wickets.