Steven Smith bowled offspin because he had a short boundary and two left-handers on strike. This is what ODI cricket is now.
The Australia-New Zealand clash resembled a simulated cricket game. Both teams seemed so determined to follow modern cricket thinking that their decisions could have been chosen by an algorithm.
This World Cup has been the most analytically driven in history, as players, coaches and analysts from T20, or even inspired by it, are with the national teams.
Take Australia. Their head coach is Justin Langer, who used advanced analytics in his time with Perth Scorchers. His assistant is Ricky Ponting, who has talked about match-ups many times as a commentator in the Big Bash League, and also as a coach in the IPL. Aaron Finch is their captain, who led Melbourne Renegades' data-driven game to a title win in the BBL this year.
This game between two of the probable World Cup semi-finalists showed just how much influence these T20 methods and Moneyball-inspired ways now have, and also some of their flaws.
One of the most popular trends in T20 is the left-right combination.
In this tournament Australia have been pushing right-hand-left-hand partnerships at first drop, although they didn't do it in the easy win over Afghanistan, nor against Bangladesh. When early wickets fall, Usman Khawaja comes in, as part of his role is to be back-up opener. But in the middle overs when a wicket falls, Smith comes in when Finch goes, and Khawaja comes in for David Warner. That has happened two times each.
There are two reasons batting teams like this kind of partnership. The first is about the spinners in the middle overs. A left-right combo means that at least one batsman has the ball spinning in to him, which is seen as favourable.
The problem here is, the effect of right-left is nowhere near as strong as teams believe it is.
Let us start with accuracy. We're constantly told that a left-right pair plays with bowlers' radars. Compared with two right-handers at the crease, it does. But bowlers are at their least accurate with two left-handers facing, and it's not even close.
One left-hander means you get a wide every 6.4 balls more often. Add another and wides happen 5.2 balls more often again.
This is rather incredible, because left-hand batsmen are not rare. They face 34% of all balls in ODIs. Yet they are still the great disrupters.
The real advantage, theoretically, in splitting up a same-handed partnership is when spin is on. But even there, other than a slight boost of strike rate (about three points), there isn't much difference at all. When two right-handers are at the wicket, they bat at a slightly better average than when it is left-right. The only time a partnership deviates from the norm is when two right-handers face seam; the average dips to under 30. For spin, it doesn't have that kind of effect.
With all that in mind, is it worth upsetting your batting order, unless the other team possesses two spinners who turn it the same way, and all your batsmen are better against the ball spinning in than away?
The interesting thing in this particular game is that because Australia lost so many wickets, they ended up with a left-left partnership of Khawaja and Alex Carey. New Zealand had two specialist spinners, who turn the ball in to left-handers, and that caused them match-up concerns.
When Kane Williamson bowled his last, and seventh, over, Mitchell Santner and Ish Sodhi had bowled the same number between them. Sodhi came back on to bowl some unlucky death overs, but Santner bowled just the three overs in the match, which for the front-line spinner is bad. And this was on a pitch that helped spin and for which New Zealand brought in the extra spinner.
Santner's three overs went for 23 runs, which seems poor. But five runs came from a wide down the leg side, and only 17 runs came off the 16 legal balls he delivered to left-handers. Santner would not have a career in professional cricket if he couldn't bowl to left-handers. And in fact, perhaps he is better against left-handers than right-handers. But let's look at the other two spinners first.
Sodhi's first five overs went for 26, and he was hibernated while Carey and Khawaja batted together. This although his run rate in the game was identical when he was bowling to right- or left-handers. And Khawaja did not pick his wrong'un.
You could argue that Williamson bowled the best of the three spinners. At the press conference he was clear on why he bowled himself: "The match-ups kind of didn't really fall our way, with both our spinners turning the ball in to two left-hand batters. Hence, why I bowled a few more overs again."
So let's look at all three bowlers against left- and right-handers.
The worst bowler against left-hand batsmen here is Williamson. And not even by a little - though he is nearly half a run an over more economical than Sodhi, he averages about 15 runs more. Santner averages eight fewer against left-handers than right-handers, so even allowing for the fact that he is about half a run an over more expensive, he is far better against lefties than righties. Sodhi is the only one who is better against the kind of batsmen you would expect him to be.
This is the problem with very basic match-up information. Everyone who has ever played the game knows that the ball spinning away is generally harder to play than the ball spinning in. But that doesn't hold true for every batsman, nor for every bowler.
Carey does struggle when the ball spins away. But Khawaja doesn't; he is pretty much as good when it spins in as when it spins away.
The other interesting wrinkle is that in this tournament Khawaja has struggled against pace bowling. New Zealand dropped him twice against seam. And Carey shows a marked preference for spin over seam.
So the correct match-up was probably seam from both ends. New Zealand tried that for five overs, and when it didn't work, Williamson brought himself on. Which worked, but over 20 overs after the partnership worked and Australia already had a decent total on the board. And the two front-line spinners just disappeared.
No one in world cricket seems to keep data on how far players hit their sixes. And while there is much that cricket should have metrics for - where are the fielders standing? - how far batsmen hit the ball is not next on the list. When it does come in, it could make an interesting coaching tool.
Players have always attacked short boundaries, and T20 has exaggerated this. Even before grounds began to be measured, this was a big deal. Now players seem to be trained to try the shot based on the boundary, not their strengths. Commentators are wise to this and feed fans information on the dimensions of the ground, which it is impossible to see on TV without a graphic.
There is a lot to gain from this information for players. But there is also a psychological effect, where teams play for that short boundary and change their game.
Ross Taylor is probably one of the best slog-sweepers to ever play. Facing Glenn Maxwell, an offspinner who turns the ball in, you would expect Taylor to play the shot, or even his normal sweep. Maxwell was delivering the seventh over by Australia's multi-headed fifth bowler. Williamson had just been dismissed, Tom Latham was scratching around, and the asking rate was creeping towards 6.8 an over. On this pitch they couldn't let it rise above 7.
Maxwell was around the wicket, trying to bowl fairly straight at off stump, and Taylor had four balls. Not once did he try the shot he hits the most sixes with. Instead he tried to dab the ball, work it, and even played a reverse sweep. In the last five years, of the 2086 balls ESPNcricinfo has logged of Taylor playing spin, he had played three reverse sweeps before this one. But we have him down as playing well over 100 sweeps or slog sweeps. When playing the sweep, he scores at 10 runs an over, averaging 87.
But he didn't play this shot on any of those four balls from Maxwell. There may have been more than one reason. One of them had to be that longer-looking leg-side boundary - 68 metres away. On at least two occasions that over, he looked towards the shorter boundary to the off, although Maxwell was bowling for him to hit to leg.
The next over Taylor was facing Pat Cummins. Now the far shorter 58-metre boundary was on his leg side. Cummins went short and Taylor pulled one - not entirely middling it, but still finding the gap between the two fielders. That highlighted how important the short side is. Then he tried his stand-up slog-sweep across the line, skied the ball as high as any building in North London, and was caught.
From a psychological standpoint, those ten metres of difference are huge. Knowing you just need to mishit a ball to get it over is a delicious prospect. But it's also quite clear that even with an extra ten metres on the boundary, the chances of hitting Maxwell for six with the spin were far higher than those of hitting Cummins across the line on a pitch that by that point had a touch of variable bounce.
We don't measure sixes, so we can't tell you what length an average Ross Taylor leg-side six travels. So maybe he knows his range better than us. And while he may be in career-best form, he's not the hockey-swatting god of a few years ago. But here he is in 2011, hitting sixes well over 70 and 80 metres, and here he is in the IPL in 2015, effortless carrying 72 metres with a sweep .
For whatever reason, Taylor didn't target the handy part-time offspin of Maxwell, but he did the searing pace of Cummins.
Replacing Taylor was Colin de Grandhomme. With him at the crease, Smith came on to bowl his legspin. It would usually have been a bizarre choice, but de Grandhomme's reputation against legspin is known. In ODIs he averages 18 against it, while hitting at less than a run a ball; overall he averages 30 at a strike rate of 110.
It is hard to tell how much of that mattered when Smith delivered a half-volley first ball that de Grandhomme hit straight to long-off. Either Australia's plan had worked, or New Zealand had sent in de Grandhomme to dent the run rate straight away and it backfired.
What followed was more interesting. Smith bowled offspin (he has been trying it in the nets) to finish the over. There would seem to be a few reasons for this. One is that Latham (who was struggling) and Jimmy Neesham are both left-handed. Then there is the short boundary again - it was now on the left handers' leg side.
The first ball was to Latham, who has a slightly better record against offspin than legspin. Neesham does not.
From the Champions Trophy until the start of this World Cup, left-handers have been 0.7 runs an over slower when facing offspin than when facing legspin.
But they get dismissed far more often, averaging 7.25 less against legspin than against offspin. Meaning legspin has been better against left-handers than offspin in that time. And none of this is factoring in the bowler. Smith is far better against left-handers in ODI cricket than he has been against right-handers.
Smith is a very part-time bowler (he delivered three full tosses in his first over against New Zealand) and he is now trying an even more part-time skill, offspin; he looked horrendous trying to get to the crease.
This is modern cricket: a part-timer with a casual skill exploiting a match-up that doesn't quite work, while everyone has one eye on the short boundary. Welcome to the 2019 World Cup, T20 data edition.
Stats inputs from Shiva Jayaraman