T20 batting: running out of steam

When should you consolidate? I’ve devised a general rule to calculate when the batting team should slow down and conserve wickets.

Let’s recap the current state of T20 International batting. Teams usually end their innings with wickets in hand. Since 2016 the average first innings score is 166-6*. Teams rarely get bowled out, having enough batting depth to attack throughout, even if a wicket falls early on.

Looking at this another way, on average it takes more than 120 balls to bowl a team out. The openers can go out and play naturally, expecting the top seven to do the business. The number eight batsman averages only three balls per game. I think of limited overs batting in terms of “Expected Balls”: how long would you expect it to take to bowl this team out if they were batting normally? For example, England bat deep, expecting to last 172 balls before being bowled out – this gives them licence to attack in a game that’s only 120 balls long.


The more you get in the first innings, the higher your chance of victory. But get greedy, take too many risks, and you may fall short of a middling score that might have been enough. Any approach to batting has a range of possible outcomes. The goal is to pick the approach that maximises expected win %. How do you do that?

Here’s one example – consider a binary choice where number four in the first innings can either bat normally or anchor (Strike Rate down 10%, Average 20% higher). According to my model, for this current England T20 team, pre innings or at 0-1 anchoring is not optimal. 0-2 it’s marginal. It’s only worthwhile if you’ve slumped to 0-3. Which, coincidentally, is the point at which England’s expected balls drops below 120 (ie. they run the risk of the median innings not lasting 20 overs). This makes intuitive sense: tailor your batting aggression so you almost (but not quite) get bowled out.

Note this assumes England are playing against an equally talented team – hence win % pre game is 50%

The general rule: bat normally unless Expected Balls < Balls Remaining.

A recent example – England were 34-3 (5.3) – which looks precarious, but the Bairstow-Stokes-Morgan middle order meant it was more-likely-than-not that England would bat all 20 overs, and have a reasonable chance of chasing their 180 target. England won the game in the 20th over. Maybe that “lose three wickets in the powerplay, lose the game” maxim is outdated as T20 averages improve. For England, Expected Balls exceeded Balls Remaining, even having lost three wickets in the powerplay.

But this is too simplistic. Not everyone can strike at 150: you can’t expect fireworks from every tail. Here’s the strike rates of the top 10 T20I teams over the last five years. Numbers 9-11 just aren’t as good. I think of teams “Running out of steam” when all the quick scorers are out.**

“Running Out Of Steam” depends of the composition of one’s batting order. England currently have Jofra Archer at number nine. Deep. West Indies aren’t so lucky – Keemo Paul bats at eight with a domestic career SR of 107 – so they Run Out Of Steam at six down.

My hunch is that cricketers know what their tail is like, and how likely it is that tail will be exposed, and bat accordingly. Take another recent example – WI T20 #1 – at 59-5 (5.1) West Indies were vulnerable. One more wicket and they were done for. So Pollard and Allen consolidated, taking 37 from the next five overs. A rain interruption meant the innings was reduced to 16 overs. With just six overs left – it was time to attack, lifting the score to 180 by the end of the innings. Subsequent discussion focussed on the impressive assault, missing the responsible consolidation period that made it possible.


Here’s the “Balls to Run Out Of Steam”*** for the Top 8 T20I sides, based on their most recent XI

As at December 2020

This tells us that England, Australia and Pakistan have the capacity to score more quickly than each player’s career record (ie. if they bat naturally, they are wasting resources by being too conservative). If wickets fall, that should be reassessed****.

Note Sri Lanka put out a particularly weak XI in their last game. Numbers four and below would expect to strike at below 130 – way off the pace. Hence they run out of hitters unusually quickly.

Teams should tailor their aggression, aiming to not quite run out of steam. To do this, throughout the innings the batting team should compare EBTROOS to balls remaining, adjusting the EBTROOS as wickets fall.

Just imagine those clipboards showing live updates of Expected Balls To Run Out Of Steam, and Optimum Strike Rate. (Screenshot from Sky Sports)


*Top 10 teams against each other. Sorry Luxembourg.

**I wish I was good at writing. Spent ages trying to come up with a better name for it than “running out of steam”. Ideas welcome.

***BTROOS = Balls To Run Out Of Steam. This is clunky stuff.

**** There’s an added complexity which I’ll keep for the footnotes: median innings length is not the same as balls per wicket. The difference is only 4% at the start of the innings, but gets bigger as fewer wickets are left. Here’s the same table, but with Median Balls To Run Out Of Steam*****. MBTROOS = Median Balls To Run Out Of Steam. Perhaps MBTROOS could rhyme with albatross. Anyway, here’s the MBTROOS for the latest England T20 lineup:

MBTROOS for England. Note 3 down after the powerplay would mean MBTROOS 86, which are two balls more than the 84 remaining. So keep attacking!