Limited-overs cricket divides the game into two halves: batting first and chasing are markedly different from each other. With a target in sight, batting sides approach their innings considering what they need to achieve. Compared to the uncertainty involved in setting a good target, this largely makes chasing a better strategy, which is why teams winning the toss have chosen to field first in 77.6% of completed matches in the IPL since 2015. When they opt to chase, teams win 57.5% of matches. On the other hand, even when they lose the toss and are asked to chase, they win 52.5% of the time. Teams have identified that chasing presents a clear advantage. But what does the average chasing side do right? How is the standard chasing win constructed in the IPL?

In this article, we will analyse chases of 140 or more in the last five IPL seasons, to restrict our data to reasonably challenging targets in the recent past.

Looking at the chances of winning a chase, a total of between 140 and 160, which would be considered slightly under par on an average T20 wicket, presents a 61.1% chance of being chased down; 177 is the average target for teams in this dataset, and that falls at the end of the 160-180 bracket, which is a traditional "par" score and corresponds to a roughly 57% and 44% success rate.

To break up how winning sides construct their chases, let us first look at the powerplay. The median winning team achieves 31.07% of its target by the end of the powerplay, for the loss of one wicket. The median losing side, in contrast, gets to just over 26% of the target at that stage, while losing two wickets. Remarkably, the average percentage of target scored in the powerplay is mostly consistent for winning sides, across target ranges.

The average winning sides score a tad more than 30% of the target in the powerplay, which makes up 30% of the allotted overs. In addition, the median number of wickets lost is one, across all target ranges. Typical winning teams follow the consistent strategy of staying just abreast of the required rate while conserving wickets.

Losing sides, on the other hand, make less and less of the target as it goes higher, and lose a median of two wickets in the powerplay. They fall behind the curve even in the first segment of the innings.

The number of wickets that fall in the powerplay is well correlated with chances of winning a chase. As the following table shows, each additional wicket down at six overs brings the win probability down by huge margins. No doubt, this is correlated with low run rates that come about due to the loss of wickets.

However, if a side makes more than 30% of the target in the powerplay, staying close to the required run rate, there's no clear trend saying more wickets lost leads to a lower win probability.

Bowling philosophies in the powerplay are varied. Batsmen are most conservative at the start of the innings, and some teams try to sneak in a few "quiet" overs to retain more attacking options for later. The above data suggests that bowling teams should look to attack more and take wickets early on, deflating the innings before the batsmen start to cut loose to utilise the field restrictions towards the end of the powerplay overs.

The seventh over, when batting teams are waking up to the second epoch of their innings, should be the designated slot for getting through a part-timer's over. The Melbourne Renegades have employed Tom Cooper to this end.

Tom Moody, on a recent episode of the Pitch Side Experts Podcast, agreed about attacking bowling early in the piece: "In the first six overs… the value of wickets outweighs the fact that you may go for a few boundaries, so I'd much rather focus on setting up the first six overs with an attacking approach, knowing that if I've got two or three in the bag after six overs, we're in a very strong position to control the innings."

On the other hand, the chasing team could employ dispensable pinch-hitters at the top of the order for short, fast knocks that make the most of the first three overs, when both teams are playing circumspectly. This ensures an early lead over the required rate, without the loss of a wicket meaning much. In the UAE, where pitches are more sluggish than in India, making hay while the ball is hardest might prove to be a key strategy.

***

Minimising the loss of wickets while going at the asking rate till the powerplay ends seems to be the way an average team goes about winning a chase. How does the rest of the innings pan out?

The mean percentage of the target scored in wins is more than that in losses at all stages of an innings. The average unsuccessful chasing side is always behind the average successful one in every phase of the innings, and the gap increases as the innings progresses. The average lost chase reaches 90% of the target if the innings lasts 20 overs.

The curve for the typical won chase follows the line of equality closely: the percentage of the target scored is almost always hugging the percentage of deliveries taken. This suggests a bare minimum optimal strategy for chases, which has also been suggested by various operations-research studies on cricket: try to go at the required rate, always.

The average number of wickets lost tells the same tale: a normal chase-winning batting order always has more than one wicket extra in hand compared to a losing one. This difference dwindles as the chase approaches the end and wickets in hand lose their relative value, but from overs six to 16, the gap is always more than one wicket.

That the gap closes towards the end of the chase reflects the dispersion of aggression in a typical victorious chasing innings. Batting sides up the ante as the innings draws to a close.

The distribution of aggressive intent from a batting side can be seen in the runs it scores in context of the current required rate. Till how late in the innings are batsmen willing to not strike faster than what is needed? The next graph takes the runs scored off a given ball minus the required runs per ball and averages it for each over. In conjunction with the probability of losing wickets in each over, this explains the distribution of batting resources in a normal chasing win.

An average successful chase can be broken into four phases. In the first two overs, batsmen are settling in, gauging the conditions and conserving their wickets, happy to score below the starting required rate. This increases the asking rate, but they then capitalise on the fielding restrictions in the latter half of the powerplay, going at 0.1 to 0.2 runs per ball faster than what is asked for. From the seventh to the 12th over, they again switch to going below the needed rate, cruising while not drifting too far below it. Noticeably, the seventh over is the most conservative - with the lowest chance of losing wickets, and a run rate well beneath what is required.

The intermediate phase increases the required rate, but since fewer risks are taken, it sets a launchpad for the final stretch. The pacing in the middle overs and the gradual rise of the scoring rate means that the required run rate stays manageable till the 13th over begins, after which the batting steps on the pedal.

In contrast, the average differential of runs scored and runs needed per ball is negative for all overs in lost matches. This is in addition to the wicket probabilities being higher, again, for all overs.

The average unaccomplished chase keeps drifting farther away from the right course, the sluggish scoring and the loss of wickets feeding each other and deflating the innings cumulatively. Wickets slow the batting down, boosting the required rate, leading to more risks and more wickets.

At the end of the 12th over, the median winning team breaks even with their run-scoring rate compared to the asking rate. At this stage, the median winning team leaves 70 runs to get, with eight wickets hand, while in losses, the median equation is 90 runs needed with seven wickets in hand. Good chasing teams work towards the target throughout the innings, seldom leaving too much work for the end.

Nevertheless, can we pinpoint a "par" target to leave for the last eight overs, which gives you a 50-50 chance of knocking the target down?

To accomplish this, we will fit our data to a mathematical model that predicts the chances of winning, given how many runs need to be got and how many wickets remain at the end of 12 overs. Our method of choice will be logistic regression, which uses available data to smoothly predict the chances of a binary result (in our case, win or loss). This will tell us how much an average team can leave for the last phase, depending on the number of wickets they have intact, to give themselves an even chance of victory.

This figure shows the predictions of the model after accounting for data from chases of greater than 140. The three different lines correspond to situations with zero, two, and four wickets down at the end of the 12th over. If a team has lost no wickets, anything under 90 runs needed will give them a greater than 50% chance of winning. Having 60 runs to win will give them an 80% chance of a win.

The equivalent value for a 50-50 chance if a team is two wickets down at the same stage is about 80 runs, and 70 runs if four wickets have been lost. Notice that this par target decreases by about ten runs for every two wickets lost at this stage of the innings. This quantifies the payoff between conserving wickets and scoring runs in the middle overs: for every wicket you lose, you should be about five runs closer to the target to maintain even odds of success.