Ashes 2019 - Player Ratings

With the dust having settled on an entertaining Ashes series and, more importantly, the excellent ball by ball data provided by Cricsheet becoming available for the series, it's time to take a look back at our series through the lens of our player ratings model. The model, described in more detail here, uses a combination of logistic and binomial regression to generate ratings for each player involved in a given set of matches - taking into account the quality of the opposition player. So, for example, the wicket of Steve Smith is worth a lot more than than of, say, David Warner.

Before we get into it, it's worth taking a brief moment to reflect on our pre-Ashes article, which used a modified logistic regression to predict the Ashes, and attempted to quantify the effect the returning Steve Smith and David Warner would have on Australia's chances. We gave a 2-2 scoreline a 13% chance of happening, which was the joint 3rd most likely outcome, and the most likely result in the case that just four tests finished in a result.

It wasn't perfect - we had England just about favourites to regain the Ashes (with 55% likelihood) - which of course they failed to do, but we did quote "the fitness of James Anderson, the weather conditions, and the performance of England's frail looking top order" as potential deciding factors, all of which certainly had an impact.

Now to look at our player ratings. The rankings are only based on performances throughout the series - so players who performed well against the series other top performers will benefit.

In our World Cup player ratings model we used a metric that weighted our player ratings based on whether we wanted to prioritise run-scoring/economy or wicket-preservation/taking. Here we use a fixed value for that metric that places more weight on wickets, to reflect the higher value that wickets have in test cricket, and the relative lack of focus placed on fast run-scoring.

Let's look at the batting ratings first. Absolutely no surprises at the top with Steve Smith rating as our best batsman - having averaged more than double anyone else in the series there isn't much more than can be said about the man.

More interestingly perhaps, is that we have Labuschagne in at second, ahead of England star Ben Stokes. Given that Stokes averaged 55 to Labuschagne's 50, this might seem surprising, so it makes sense to dig a little deeper and look at the differences between what these two batmen were facing.

We found that nearly half of the deliveries bowled to Labuschagne were by Jofra Archer or Stuart Broad - both of whom had excellent series, and he faced plenty more from the likes of Leach and Woakes. Stokes on the other hand, only had to deal with Australia's best pace bowlers - Hazlewood and Cummins - for under a third of his balls. Most strikingly, he faced an incredible 327 balls from Nathan Lyon, who didn't have the most impressive series overall. All of this combined means that our model rates Labuschagne's overall performance throughout the series as more impressive.

Looking down the rest of the list it's roughly as you'd expect, with Burns the only (regular) opener to make the top 10. Seeing Wade so low may be a surprise, given that he scored two centuries and had a higher average than Root, Denly or Buttler, but he enjoyed the luxury of facing more part time bowling than most batsmen in the series. He faced nearly ten overs worth from Joe Root, and only Steve Smith faced more Joe Denly deliveries than Wade. The aforementioned Root, Denly and Buttler, however, all had to deal with Cummins and Hazlewood for the bulk of their deliveries.

On a broader note, these rankings seem to reflect the overall feelings about the respective batting lineups. That is, that whilst both are weak, England's is probably a bit better, but that advantage is then trumped by the presence of Steve Smith.

Moving onto the bowlers, and again, no surprises at the top with top wicket Pat Cummins taking first spot. Next is Hazlewood, who bowled excellently throughout and whose economy bettered any of the English bowlers.

Seeing Jack Leach in ahead of Jofra is definitely surprising, however. Jofra averaged 20 throughout the series, compared to 25 from Leach, and also bettered him for economy. As with Stokes and Labuschagne, we need a closer examination of who they were bowling at to understand what's happening here.

The first thing that sticks out is that the average Jack Leach delivery was to a batsman averaging 50 in the series, compared to just 42 for the average Archer delivery. This was in part because Archer opened the bowling, where he spent time bowling at the very poor Australian openers, and also because he was brought on regularly to finish off the tail. Compare that to Leach, who spent a larger proportion of his time bowling at the likes of Steve Smith - 25% of his balls were to Smith compared to 17% of Archer's. Perhaps most crucially, Jack Leach managed to claim the wicket of Smith in this series, something that Archer failed to do. It's worth bearing in mind, however, that our model currently does not take injuries inflicted into account - maybe something to investigate in the future.

There are few surprises in the rest of the bowlers list, with the rankings falling broadly in line with the respective bowling averages.

As a final note, those of you who saw our World Cup player ratings may recall that Archer was not rated especially high by our model there either. In that tournament he was overshadowed in our model by a more understated team mate in Liam Plunkett, just as he has been here by Jack Leach.

Whilst he has no doubt been exhilarating to watch, this initially indicates that his current averages - 25 in ODIs and 20 is Tests - may be unsustainable given the true quality of his bowling. To date he has perhaps had his figures boosted by spending more time than his colleagues bowling at weaker batsmen, and has not been especially successful (at least in raw statistical terms) at bowling at the stronger batsmen. As we noted earlier, he didn't dismiss Smith all series, and whilst he did dismiss Labuschagne twice, he also bowled at him more than anyone else.

It may be that this pattern is just a statistical quirk, and will even itself out over time, or it may be that his figures decline in the long term as he comes up against more consistently higher quality batsmen. Whilst it's certainly too early to draw any conclusions, it does make for something to watch out for as his career develops.

Thanks for reading. To keep up with all the latest thoughts and analysis follow us on Twitter!