Should Buttler bat up the order in ODIs?

Having a model for ODI Cricket is great when it comes to considering selection, or gambling, but it’s challenging to come up with further practical uses. Fortunately, some recent tweets about batting orders gave me an idea – using the model to suggest the optimum batting order.

England have batsmen with averages and strike rates to get excited about. The current top six is Roy, Bairstow, Root, Morgan, Stokes and Buttler.

Jos Buttler’s career strike rate is 120. He once scored a century in sixty-six balls. If England get a good start, at what point should they push Buttler up the order so he isn’t watching from the pavilion when he could be swishing sixes? He has finished “not out” in 23 of his 116 innings – and could have contributed more in each of those matches if he had been on the field earlier.

Firstly, let’s consider how Buttler has performed by batting position:

Fig 1 – Buttler’s performances batting in ODI Cricket, up to 13/07/2019 by position in the batting order.

The more excitable among us would conclude that six is Buttler’s weakest position, and he has to bat at four or five based on the above averages and strike rates. Personally (and somewhat arbitrarily), I’d like a 20 innings sample size before concluding. All the table above says is there’s no compelling reason why Buttler can’t bat anywhere in the middle order.

So what number should Buttler bat? Using a model of ODI cricket, simulating England batting against their own bowlers at Chester-le-Street*, we can predict performance for England’s usual batting order and compare that to Buttler jumping up two places to number four.

The Duckworth-Lewis method tells us that the way batsmen play at each stage in the innings is a function of how many wickets have fallen. The hypothesis is that the earlier the second wicket falls, the more conservatively England will bat, and thus the less useful it is to promote Buttler. It would actually be counterproductive, because if he’s out he’s not around to score quickly at the end of the innings.

Scanario time: we join the action at the fall of the second wicket, Roy and Bairstow the men out.

Fig 2 – modelled impact of moving Buttler to bat at four rather than six. x-axis represents the over in which the second wicket falls.

The chart shows that promoting the swashbuckling Buttler too early has a slightly adverse impact on expected runs (he’s not the person you want at the crease as you rebuild – hold him back). If the second wicket falls any time after 20 overs, it is beneficial to move Buttler up to number four. The later in the innings the second wicket falls, the more important it is to promote Buttler.** That said, the benefit is less than one run for overs 20-30, so if the batsmen are concerned a fluid batting order could cause them to underperform, coaches should take heed.

Note that the benefit starts to shrink very late in the innings – as the number four will only face a handful of balls anyway.

To put the previous chart into context, here’s a comparison between the two scenarios:

Fig 3 – Modelled median runs scored after the fall of the second wicket. x-axis shows different stages in the innings when the second wicket falls.

The two curves are very similar. If your eyes (and/or phone resolution) are up to it, you’ll see that the blue line (Buttler in at six) underperforms the orange line, especially in the latter stages of the innings.

England have an analyst with a ball-by-ball ODI model. Has he already done this analysis and are England already applying it? Consider the evidence of the World Cup Group Stages:

  • #6 vs Australia. Comfortably chasing 224 when the second wicket fell with the score on 147 in the 20th over. Morgan bats at four, England don’t lose another wicket.
  • #4 vs New Zealand. Second wicket falls in the 31st over. Buttler promoted to four.
  • #6 vs India. Second wicket falls in the 31st over. Third wicket falls in the 32nd over. Buttler bats at six? Takes revenge on the ball, scoring 20 from eight balls.
  • #6 vs Australia. England 53-4 in the 14th over when Buttler comes to the crease. Couldn’t realistically hold him back any longer.
  • #6 vs Sri Lanka. England three down inside 20 overs, Buttler held back to number six.
  • #5 vs Afghanistan. England 169-2 after 29.5. Morgan goes in, hits 148 (71). Buttler doesn’t get a turn until the third wicket falls in the 47th over.
  • #4 vs Bangladesh. 205-2 (31.3). Buttler promoted to four.
  • #6 vs Pakistan. England three down with just 86 on the board. Buttler comes in when the fourth wicket falls.
  • #6 vs South Africa. 111-3 (19.1). England play it safe and Buttler bats at six. Fair enough.

England’s strategy broadly follows the recommendations in this post (and therefore what an ODI simulator would recommend). Two exceptions: against Afghanistan and India. It would be fascinating to know why Buttler batted at four against New Zealand, but not (in similar circumstances) against India.

We can conclude that with their current batting order, England should move Jos Buttler up the batting order if the second wicket falls after the 30th over. A word of caution – the 90million balls I modelled were for this specific scenario, and not the general case. If you would like me to consider another scenario, please do get in touch via the “Contact” page or @edmundbayliss on twitter.

*If I had my time again I wouldn’t have had England playing against themselves and at a ground with high ODI batting averages. Regrettably, I neglected to update those inputs after a World Cup game there. If the modelled runs in this piece feel high to you, that’s why.

**It’s human nature to pick one reason for an outcome. “If the coach had just tweaked the order, we would have put enough on the board”. It’s seldom that clear cut. These batting order changes are worth up to three runs, very much in the “extra one percent” territory.