Editor’s note: The article was originally published January 18, the T20 statistics are valid as of that point. It is being reshared ahead of the 2022 IPL auction.

The Indian Premier League mega auction is around the corner and one of the key requirements teams focus on is acquiring a quality death bowler. That being said, there is no complete death bowler in the market given Jasprit Bumrah has been retained by Mumbai Indians.

A simple glance at death bowling economy is not enough while recruiting because those figures are almost always on the higher side and there is not much to separate bowlers on that basis.

A better way of looking at death bowling is understanding how a bowler constructs his overs and whether there is scope for improvement on that front. If a bowler has the base skills – accurate yorker, a good slower ball or a hard length delivery, improving over construction is more feasible.

The yorker is a dangerous ball to attempt – when it is missed, most death hitters can take it apart, as we saw with even Mitchell Starc against Fakhar Zaman in the World Cup. At the same time, when a bowler bowls a series of yorkers, they are showered with praise. Everyone remembers Kartik Tyagi’s epic over against the Punjab Kings. Or T Natarajan’s yorker masterclass against Rishabh Pant and Shimron Hetmyer.

This duality of reaction raises some pertinent questions. Is either extreme favorable? Was Karim Janat who is much slower than Starc equally culpable in the over to Asif Ali at the T20 World Cup?

In short, one of the big questions in T20 cricket: how to construct a competent death over?

Realistically, there is no single answer to these questions - optimal construction depends on bowler strengths. Let us dive into it by looking at a few case studies – most of these bowlers will be targeted as big buys for IPL teams.

Before we do that, though, let us establish some ground rules:

  • An attempted yorker is any ball that winds up as a full toss, yorker or in the slot.
  • Anything behind the slot is considered a length ball (good, back of length, short). It probably isn’t best to group them but since their economies are similar, we can make things simpler.
  • The above two points apply to stock balls, slow balls are grouped in a separate category. As such, slow yorkers are not counted as attempted yorkers; they’re rare so we can make the compromise.
  • Trade-offs are made between these three ball choices.
  • The average yorker accuracy rate is 33% and the highest is 43% which is Bhuvneshwar Kumar’s. That is probably the ceiling.

Tymal Mills

It is only fair we start here because Mills is a massive outlier – he attempts the least percentage of yorkers at the death and yet, has one of the lowest death overs economy rates in the world. The thinking behind it is simple – while a nailed yorker has an economy rate of just under 6 RPO, a missed yorker goes for around 11 RPO. And since yorkers are missed more often than executed, it is extremely high risk.

In the 2021-’22 BBL, only 10% of the balls Mills bowled were attempted yorkers. A quarter were slow balls (a mix of back of hand and leg cutters) and the rest were hard length balls driven into the pitch. With hard length balls, there isn’t as much two-fold risk. You get constant returns and Mills is very good at it. The slow ball change up means the batters have to adjust to the pace drop and cannot commit to either the front or backfoot.

Since the 10-65-25 (yorker-length-slow split in terms of overall % – to be used later in the piece as well) split works for Mills and he is not very comfortable with the yorker, it is close to optimal for him. It can of course be tweaked based on conditions and opposing batter, but he has more or less found his sweet spot.

Others of his breed: Anrich Nortje

Kagiso Rabada

Things just got interesting.

When Rabada nails a yorker, he goes at 4.5 RPO and when he misses, at over 11 RPO. He nails 33% of yorkers, so his attempted yorker economy is roughly 9.3. Meanwhile, his slow ball economy is 8.6 and hard length economy is 7.5.

Despite this, Rabada’s split is 37-36-27. His most common ball is the attempted yorker when it is also his worst ball. For his attempted yorker to achieve the same returns as his hard length ball in the long run, Rabada’s yorker accuracy must be 58% which is essentially unachievable.

Given his current record, it would be ideal for Rabada to drive up his hard length percentage to over 50 while using the yorker and slow ball as change ups. This would particularly benefit him against predominantly front foot players – his poor showing against Andre Russell, Pollard and Fabian Allen in the Caribbean tour can be attributed to poor over construction.

Rabada has the tools to be a world class death operator but his yorker alone will never get him there, simply because such a high accuracy is impossible to achieve. He will have to move towards the Mills category to stay relevant.

Others of his breed: Chris Jordan, Andre Russell, Lockie Ferguson, Mohammed Siraj, Mitchell Starc

Jofra Archer and Jasprit Bumrah

Now we move to proven world class operators.

Archer’s yorker goes at 6.5 RPO but his missed yorker goes at only 9.6 RPO. So, despite Archer having a yorker accuracy of only 30%, he is still going at only 8.7 RPO off his yorker attempts. That alone is incredible. Jasprit Bumrah is similar, he has a yorker accuracy of just 26% but his yorker attempts go at only 8.5 RPO.

There is only one thing that separates Archer and Bumrah: it is their slow ball economy. Bumrah’s is 7.1 RPO and Archer goes at 9.5. Their length ball is steady around 7.5 RPO. Archer’s split is 48-19-33 which is clearly too many slow balls. Moving towards a Bumrah-like split of 36-38-26 would be closer to optimal, with more hard length + yorker and less slow ball.

Despite hard length being Archer’s best ball, his yorker gives him so much leeway that he can attempt it a lot more than say, a Rabada or Mills. Having a more varied split is always more effective because it keeps a batter guessing.

In addition, when a boundary is hit off the previous ball, there are two options available to switch to, giving the batter no patterns to pick on. Someone like Tom Curran, who has a poor length ball almost always switches to a slow ball when his attempted yorker goes wrong, making him very predictable. This kind of versatility is what makes Bumrah and Archer the world’s best.

Others of their breed: Chris Morris, Avesh Khan

Harshal Patel

A lot of the discourse around Harshal, after his sensational IPL 2021 season for RCB where he finished with 32 wickets, is whether he will get found out. But there is very little dissection of his methods. Harshal has an excellent yorker accuracy (38%) which means his yorker goes at only 8.21 RPO. His slow ball dips on the batter and goes at only 7.5 RPO. Meanwhile, his length balls go at over 9 RPO.

Harshal’s yorker could potentially go to the elite (40%+), but we cannot bet on it. His length ball economy is likely to remain the way it is as well, because of his lack of pace. Harshal’s current split is 38-21-41 which is probably close to optimal.

While there is merit to bowling over 50% slow balls, his effectiveness will begin to drop if it is picked. So currently, he is using the slow ball as a weapon but has a fairly good yorker to keep his longevity by not having to drive up the slow ball rate to an excessive amount.

Others of his breed: Mustafizur Rahman, Arshdeep Singh

Bhuvneshwar Kumar

As stated earlier, Bhuvneshwar has the highest yorker accuracy in the last few years (43%). Yet, his recent death bowling numbers seem to be on a decline. One of the reasons is Bhuvneshwar’s missed yorkers going at 13 RPO, also the highest amongst all bowlers.

Bhuvneshwar’s slow ball has also become very ineffective – it is the 6th most expensive in the IPL. Yet, Bhuvneshwar’s split is 31-32-37. That is, he bowls far too many slow balls despite it being a poor option, unlike a Harshal Patel for whom it is a main weapon.

Bhuvneshwar’s missed yorker economy could be high just due to randomness and it is possible that it comes down. What he could do to improve is bring down his slow ball % and focus on driving up his attempted yorkers with a mix of length balls. Andrew Tye has rediscovered his mojo in the BBL by significantly reducing his slow ball % after his knuckleball was decoded. This could be a template for bowlers who excessively use slow balls.

Others of his breed: Jaydev Unadkat, Keemo Paul, Tom Curran, Tim Southee, Sandeep Sharma

T Natarajan

Finally, we move to the other extreme of Mills. Natarajan, who lit up the 2020 IPL, has an elite yorker accuracy of 41%. His attempted yorkers go at 9 RPO, a decent rate for modern T20 with his missed yorkers going at 11.5 RPO.

However, there is little tradeoff here since his slow ball goes at 10.2 RPO and length ball at 11.4 RPO. Natarajan’s yorker accuracy can drop from 41% all the way down to 24% and his yorker would still be more effective than his slow ball.

Natarajan’s split is 75-14-11, which is in fact more extreme than Mills’ bias towards hard lengths. It makes sense, because Natarajan’s yorker accuracy can be rubbish and it’d still be his best ball. However, since the yorker is his weapon, it is essential that his accuracy does indeed remain elite (more than 40%) for him to survive.

Others of his breed: Dwayne Bravo, Deepak Chahar, Trent Boult, Sam Curran

Graphic by Karthikeya Manchala for Scroll.in (Images from AFP / Sportzpics / Reuters / ICC)


  • Specialising on either end of the spectrum like Mills and Natarajan is fair, but only as long as the specialised skill is elite and consistent.
  • If the yorker accuracy needed for it to be the best ball is achievable, then it is worth improving upon it (Boult, Curran).
  • If the yorker accuracy required is unachievable, then specialising in hard lengths is best for high pace bowlers.
  • If the yorker accuracy required is unachievable, splitting up yorkers and slow balls to become unpredictable is best for low pace bowlers.

All statistics for men’s T20 cricket.

Karthikeya Manchala is a cricket enthusiast with a passion for numbers. He tweets @Static_a357.

Correction: Bumrah’s split figures have been updated to 36-38-26.