England's Test Batting & Selection

10th September, 2019.

Email: Sportsanalyticsadvantage@gmail.com

Two days ago, Australia defeated England at Old Trafford to go 2-1 up in the Ashes with one Test to play - enough for them to retain the historic urn.  Sadly, the same batting issues which have manifested themselves throughout both the current series and quite a few previous ones have come to the forefront again, and it took a herculean effort (plus a little luck) from Ben Stokes for England to register their solitary victory in the series to date, and perhaps paper over some cracks.

Just as a reminder, here are England's batting scores during the series so far:-

1st Test: 374 & 146
2nd Test: 258 & 258/5 declared
3rd Test: 67 & 362/9
4th Test: 301 & 197

Add this to the 85 & 303 against Ireland this summer, and England have been bowled out for a double-figure total as often as they passed 350.  As is often the case, this is another series where so far they have failed to pass 400 with the bat - they conceded 480+ twice against Australia though, at Edgbaston and Old Trafford.  Across the five Tests against Australia and Ireland, they're averaging 23.11 runs per wicket.

These issues really are nothing new.  I've written a number of pieces previously discussing England's batting issues, so I won't dwell on those problems too much.  For those who would like to recap on what I've written previously, here are some links:-

My general thoughts, in line with the pieces above, are that England currently have a lack of high quality batsmen suitable for Test cricket, whether they are currently in the team or not.  In some of these pieces, I'd advocated Rory Burns for a while prior to the Surrey man being picked, and Burns has had a solid Ashes - coming out with more credit than most.  

Statistically, Ollie Pope is the glaring omission from the current team, as I wrote recently here, and it's difficult to find an argument against him being picked for the fifth Test at his home ground, The Oval.  Pope coming into the lower middle-order for the final Test of the series with an eye towards the future would have made a lot of sense, but he's been omitted.   There are a few other players who you could also argue are there or thereabouts, but I'll keep those names to myself at this stage.  

Apart from this, as I discussed in the final piece linked above, the only short-term strategic solution I can find with regards to sorting out the batting would be to pick more bowlers who can bat, as opposed to bowlers who aren't particularly capable with the bat.  I'd also question whether a regular specialist non all-rounder spinner is needed in English conditions, although I think Jack Leach has had a reasonable enough series with the ball, while there's a few other selection/strategy contentions as well, which I'll also save for a rainy day at this current time - plus an issue of whether some counties aren't providing a decent enough bridge between the 2nd XI/Academy and the first XI (more on this later).

With England having underwhelmed with the bat in the Ashes, there has naturally been plenty of speculation in both the media, and on social media, as to how to solve their issues, mainly focusing on selection.  I'm truly amazed that many of those who have a considerable voice in the media are continually advocating players who haven't got particularly strong county data.  As the piece linked above 'Why England's Batting Will Fail in the Ashes 2017' indicates, it is incredibly rare indeed for English batsmen to improve their Division One average in the Test arena, so both logic and numbers are against those arguing for certain county players to be given opportunities.  If a batsman can't perform at a high enough level against county bowlers, it's extremely unlikely that he'll be able to perform better against the likes of Cummins, Hazlewood, Starc and Pattinson, for example.

England's selectors should be focusing their attention on those batsmen who have a current expected average in excess of 45 for Division One of the County Championship, which roughly translates to a ballpark average of 40 at Test level.  I'll clarify this further - this is my algorithm-driven current expected Division One average (based on data from the last 2-3 years), not career averages, which are pretty much worthless when it comes to establishing how a player will perform in the future.  It's also worth noting that this is not Division Two expected averages, although it's not particularly difficult to create an algorithm to evaluate expected performances for Division Two players at either Division One or Test level, and I have done so.  There is generally a reasonable gap between Division One and Division Two of the County Championship (there are still some very good players in Division Two, however).  Obviously I'm not going to go into much detail about this particular algorithm, although I will mention that, among other areas, it is weighted for recency and quality of opposition.

Certainly, data should be used to ascertain a player's viability at Test level.  Across the 18 first class teams, almost 350 different players have been used this season in four-day cricket, at numerous different venues - it is both logical and efficient to profile players using data given the huge workload it would be to view each player's performances, either at the grounds, or via video.  The only issue for using data in this respect is that there needs to be acceptance of its value from all concerned - perhaps the head of selectors, head coach and captain all need to buy into it - and I'd envisage that's far from guaranteed in many set-ups.

Despite reports to the contrary elsewhere, a Test average of 40 is still around the benchmark for an above-average Test batsman.  Top six batsmen for the top three teams in the ICC World Rankings (India, New Zealand and South Africa) from the 1st January 2018 onwards 37.34.  England's top six average 28.83 during this time period, and no batsman batting regularly in the top six for England averages in excess of this 37.34 mean figure for top six batsmen across the top three ranked Test teams.

The chart below illustrates the relationship between batting average and century percentage among top six position Test batsmen with at least 40 innings this decade:-

You will see that there's a pretty decent relationship between batting average and century percentage, as would be completely logical - the higher a player's average, generally, they have a higher century percentage, and vice versa.  This was also largely the case when looking at the relationship between batting average and half-century+ score percentage, represented below:-

If we look at the players in this sample who have a top six batting average between 42.50 and 37.50 runs per wicket (2.50 runs either side of 40), they have scored 121 centuries in 1191 innings, and a total of 330 innings out of these 1191 where they scored 50+.  Converting these into percentages, these slightly above-average top six batsmen scored centuries in 10.16% of their Test innings and scores of 50+ 27.71% of the time, which gives us some other benchmark metrics to assess players.  It also enables us to work out the likelihood that a player will hit a given milestone in their innings, and given the relationship between milestones and averages, we can start to ascertain that milestones are simply a natural consequence of a player's Test average, or indeed, expected Test average - the more you are expected to average, the more likely you are to hit 50s or 100s.

With this in mind, it brings us to another selection mishap, which certainly isn't an England-exclusive issue.  How many times have you seen an under-pressure player score a century early in a Test series, for someone - often in the media - to suggest that particular player has sealed his spot for the rest of the series following that century?  Certainly, a team's selectors dropping a top six batsman immediately after scoring a century would be an extremely rare occurrence in Test cricket.   

However, I'd offer a counter-argument, that being that a player's ability to score a century is simply a certain likelihood based on their expected average, and they almost always aren't a considerably better batsman following scoring this century, just as much as they aren't a considerably worse batsman after scoring a duck or a low score.  For any batsman with a decent sample size of data, a century or duck in an isolated innings would have a very small impact indeed on their expected average for the next match, and as a consequence, this will have a limited effect on their likelihood of scoring a century, or a fifty, in an upcoming match as well.  To use a current selection decision as an example, how convinced should the selectors be that Joe Denly's two recent fifties are evidence of him 'getting to grips' with Test cricket, or are those two innings essentially several relatively low percentage chance outcomes occurring in quick succession, which can easily happen, statistically?

Essentially, selection all comes down to expected performance levels.  A selectors job would be to pick the highest expectation squad/team balance based on the player pool and information available to them - if they can do this successfully, then they've done all that can be asked of them, whatever the level of competition.  Understanding statistical variance and likelihood of outcomes is a critical facet in this - kneejerk reactions are, generally speaking, to be avoided.  If a player scores a century in 10% of their innings, for example, they are virtually certain to endure a barren run of non-centuries, at some stage, and selectors should not over-react to this.  

Moving on, I briefly mentioned earlier the bridge between second XI and academy players and the first team earlier, and I want to finish on this.  There is plenty of young talent at counties, but not all of them are getting the opportunities that their performances at second team level warrant.  Recently, I created a database of players aged 22 or below in county cricket, and applied age curve data to each player to assess expected peak levels of players.  There are certainly players who have potential to be Test-quality batsmen (the aforementioned Ollie Pope is obviously one), but a number of the other batsmen with high potential at the top of the list have found it difficult to break into their counties first team, and have lacked opportunity.  This is particularly the case at some larger Test-match ground counties, some of whom appear extremely reluctant to blood young talent - some very rarely use players in this age bracket.  

It is difficult to know the reasoning for this, but I'd envisage that a decent amount of the time, explanations would potentially focus around either coaches under pressure to deliver short-term results, a lack of trust from specific coaches towards young players, or a failure to understand how second XI performances translate to expected performance levels in the first team.  Whatever the case, this reluctance from some teams to pick young players would be an issue for talent making its journey along the pathway to the first team, and eventually, potentially the national team.  Logically, the opposite should be the case - young players tend to be hungry for success, loyal to the coach/team who gave them opportunities, as well as being financially cheaper than older players.  These players should be treated as an asset, not an easily dispensable resource.  Any struggling Division Two team who is under financial constraints could do a lot worse than press the reset button, sign a bunch of these talented young players (they rarely have long-term contracts), coach them well and wait several years for them to improve into a squad capable of competing in Division One.

Unfortunately, due to commercial sensitivities, I'm not willing to share the names of the young batsmen that my age curve analysis indicates have future Test potential (there are plenty of young bowlers with high potential too!), but what I will say is that every county has at least potentially decent quality young players in their second XI - these players just need the opportunities.  However, anyone who is interested in discussing a financial package for this information can get in touch via sportsanalyticsadvantage@gmail.com.

If this article has given you insight into the data that Sports Analytics Advantage can offer cricket teams around the world in formulating team strategies, selection, draft/auction plans, or any other work, please feel free to enquire at sportsanalyticsadvantage@gmail.com.