Statistically Assessing England's Fringe Test Players - May 2020

30th May, 2020.

Email: Sportsanalyticsadvantage@gmail.com


In the interviews for the chief selector post of the England cricket team in 2018, interviewees were asked whether selection was an art or a science.  As regular readers of my work will probably have guessed, my perspective is that it is weighted strongly towards science, or in other words, it should be largely data-driven.  In fact, if I had been fortunate enough to have been asked to interview for the role, I'd have even gone as far as to suggest selection should be largely a science, which has the potential to be misled and over-complicated by art.  Quite simply, if you spend enough time looking at selection metrics and drivers for success in formats, there are enough metrics which an experienced user can generate to answer pretty much any selection or recruitment-driven question, and this has the benefit of disregarding hunches, speculation and bias, which can be a big risk if you are considerably art, and therefore 'eye-test', driven.  

For example, if you want to work out the best batting/bowling composition for a Test team at a particular venue, it's possible to use data to work out expected value of say, a 6 batsmen/1 keeper/4 pace team as opposed to a 6 batsman/1 keeper/3 pace/1 spin team.  If you want to know the effect of having a left-arm pace bowler in your bowling attack as opposed to all right-armers, you can use historical data.  If you want to know how a county player is likely to perform at Test level, build an current expected data algorithm which takes into account tens of thousands of balls of historical data.  If you want to work out the likely peak level for players, build an age-curve algorithm which takes into account the average rate of improvement or decline as players get older.  I've devised these particular algorithms, and one English county is already benefiting from using my age-curve data to assist their recruitment, which allows them to understand not only the current expected level of a player, but also their expected peak level.

Of course, there are many areas which can be assessed and quantified, and these are merely several examples.  However, given that my detailed data disagrees with at least one selection (and often considerably more) for most international teams worldwide, I'm not at all convinced that international selectors are particularly data-driven with the teams they pick.   To clarify, though, I'm not remotely suggesting that there is no need to visually scout players, or having a network of people to discuss player personalities with, but this should be like a 'double-tick' method - essentially, a player can be highlighted by the data and then have further due diligence performed on them - a potentially powerful combination.

Yesterday, the 55-player group of players who will represent England in their summer red and white ball matches was announced, and as expected, immediately there was frenzied discussion on social media with regards to the composition of the squad and the likely split between red and white ball players in the group.  My main issue with the group stems from a few players who data cannot make a particularly strong case for across any format, while a number of players who have impressive data from a strong and relatively recent sample size were seemingly ignored.  Given England have a reputation of being rather data-driven, it makes me wonder whether sometimes they use statistics when it suits them, and if it doesn't, they'll just pick players via hunches - perhaps this is a blend of art and science, but in my view, if a team is going to be data-driven then it needs to be consistent in their approach.

Given the commercial relationships that I currently have, I don't want to look at T20 data for English players, so what I want to do in this piece is look at fringe players from County Cricket who should be competing for a spot in the England Test team.  I also want to do this because I don't write a lot about red-ball cricket, but being totally truthful, this is simply because it doesn't have a great deal of commercial application unless I was to work as part of a national team selection panel.  However, it's interesting to take a different direction on occasion, and as several people in the industry have described me as a 'T20 analyst' recently, I want to ensure I'm not pigeon-holed as such!

First up, I want to start with something extremely basic, and demonstrate how something very basic can then turn into something much more detailed with the addition of several quite creative filters using methods and metrics which I don't see a lot of people inside cricket using and this will enable us to work through a long list of county players to generate a shortlist of fringe players who should comprise England's red-ball thinking over the next year or two.   The two charts below highlight County Championship data for both batsmen and bowlers, with the first chart illustrating batsmen scoring 1180+ runs across the last two seasons, comparing their batting average to their balls per dismissal.


Of the top four players closest to the ideal top-right hand corner, only Hasan Azad (not picked in the group of 55 players) has yet to make his debut, and the other three are there or thereabouts currently with the England Test team.  Moving down the group of players, former captain Alastair Cook rated well, and to a slightly lesser extent, former England players Gary Ballance and Keaton Jennings.  Sam Hain (picked in this group of 55 players) and Chris Dent also may surprise a few readers with their numbers. 

We can take a similar approach with bowlers too, with the chart below showing bowlers taking 60+ wickets in the County Championship across the last two seasons:-

This chart highlights the bowlers with the best data of the group, with all those highlighted having a balls per dismissal figure of below 50 (with the exception of Tony Palladino, who was very marginally above this figure), and bowling average below 30.  Given the numbers of Ollie Robinson (Sussex), who has not made his international debut, but was one of the names in the 55-player group, it's not surprising that he's been touted for international inclusion recently.  Lewis Gregory, also included in the group, has to be in consideration as well, given that he is a capable batsman in addition to strong bowling abilities, and is also an excellent white-ball player and brutal boundary-hitter in T20 as well.  

However, these charts are extremely basic so they need further detail and filtering to create a shortlist of players who should be England's thoughts in red-ball cricket.  Of course, given that we have 25 batsmen and 28 bowlers highlighted here, there are far too many players to comprise a realistic shortlist for England to consider.  In my article from January, 'Can England and English Counties Understand Age-Curve Profiling Better?' I discussed the typical age for an England player to make their debut, and created the following chart demonstrating the age which the 39 players who have represented England from the start of 2019 onwards made their international debuts:-



Here we can see that the majority of these 39 players (87%) had made their international debuts when aged 25 or below, and only three players made their debuts at 27 or older.  No current England player made their international debut at the age of 30 or older.  It varies a little across player type and format, but interestingly, my research shows that the general peak age of a player is around 29, so England look like they are doing this aspect of selection pretty well - getting players international experience before peak age and allowing them to develop.  Given this, if we are going to profile players suitable to contend for the current England squad, it seems reasonable to focus on those players aged below 30, and probably even lower.  Filtering for players aged no older than 27 years of age, or players aged no older than 30 who have already made their full international debut gives us this more condensed long list:-

Batsmen (in alphabetical order):-

Adam Rossington
Dominic Sibley
Gary Ballance
Hasan Azad
James Bracey
Keaton Jennings
Ollie Pope
Rory Burns
Ryan Higgins
Sam Hain
Tom Abell
Tom Kohler-Cadmore
Will Rhodes
Zak Crawley

Bowlers (in alphabetical order):-

Amar Virdi
Ben Coad
Brett Hutton
Craig Miles
Craig Overton
Ed Barnard
Jack Leach
Jamie Overton
Jamie Porter
Lewis Gregory
Matt Milnes
Matt Salisbury
Matt Taylor
Ollie Robinson
Ryan Higgins

Just by adding a realistic age filter, we've managed to trim the list of 25 batsmen down to 14, and the 28 bowlers to 15, and this enables us to do some further detailed analysis on this condensed long list of players.

One obvious problem we haven't mentioned so far is that the data in the charts fail to take into account which Division a player participates in.  To look this from a football context, if a striker scored 25 goals in the English Premier League, he'd be virtually certain of national team inclusion, but if a striker scored 25 goals in the English Championship (Division 2), then it might be enough to get them a move to a Premier League team, but would be unlikely to generate a national team call-up.  This is because there is a considerable gap in ability between the Premier League and the Championship, and it's reasonable to anticipate Championship performance levels to worsen in the higher standard of the Premier League.  The same can be said in cricket too, with there being a gap between Division One and Division Two, which must be taken into account - it still is extremely surprising that stats websites showing career data fail to note the divisional split in county cricket.

We can take this into account by using the algorithm that I have developed which I mentioned at the start of the article.  By using historical data, this takes into account the difference in standards between Test cricket, Division One cricket and Division Two cricket, and indeed, where applicable, the Second XI as well.  I don't want to discuss the details behind it, but it also takes into account recency, with more recent performances being much more relevant than older ones.

My algorithm generated the following outputs for current expected Test batting average for the group of 14 batsmen in the condensed long list:-

Although his sample size of data isn't huge and this might cause his expected test batting average to come down a little in the future, here we can see that Ollie Pope's current expected Test average makes him likely to be an elite-level Test batsman, and he should be the mainstay of the England batting line-up for the next decade at the very least.  Current Test batsmen Dominic Sibley and Rory Burns also have strong expected averages, while there are a group of players - Ballance to Hain - who have very similar solid expected numbers.

This analysis was also performed for the bowlers, and the following outputs were generated for the group of 15 bowlers in the condensed long list:-

Some readers may be surprised to see Ben Coad top this list, but the Yorkshire pace bowler has produced excellent numbers across the last few years and looks rather under-rated.  Lewis Gregory and Ollie Robinson, as mentioned previously, rate extremely well and in Gregory's case in particular given his batting value and ability in white-ball formats as well, deserves England opportunities.  Ryan Higgins should be noted as well, given that he actually features in the batting analysis too, and looks like a further decent all-rounder option.

What I want to do next is to dial down further and look to reduce a few players from this condensed long list of players after that assessment of the expected Test averages for those players, and I cut out those batsmen with an expected Test average of below 30 (Adam Rossington, James Bracey, Zak Crawley) and bowlers with an expected Test average of above 30 (Amar Virdi, Craig Miles, Matt Milnes, Matt Salisbury, Matt Taylor).  Note that many of the eight players cut out here either spent one or two years playing in Division Two, and this is absolutely logical given historical numbers illustrating the difficult step up from Division Two cricket to Test cricket.

This now leaves us with a shorter long list of the following players:-

Batsmen (in alphabetical order):-

Dominic Sibley
Gary Ballance
Hasan Azad
Keaton Jennings
Ollie Pope
Rory Burns
Ryan Higgins
Sam Hain
Tom Abell
Tom Kohler-Cadmore
Will Rhodes

Bowlers (in alphabetical order):-

Ben Coad
Brett Hutton
Craig Overton
Ed Barnard
Jack Leach
Jamie Overton
Jamie Porter
Lewis Gregory
Ollie Robinson
Ryan Higgins


We now have a group of 11 batsmen and 10 bowlers who we can perform some more detailed analysis on, and what I want to do here is compare the performances of these players to the performances of their team-mates in the matches in which they played in.  This is quite useful as it demonstrates the extent which a particular player has outperformed their team mates who will be facing the same bowlers in the similar conditions, although it should be noted that this analysis might undervalue opening batsmen compared to other batsmen a little given they have to face the new ball, but it is also possible to compare opening batsmen alone to ensure a like-for-like comparison.  

Here are the numbers for batsmen first, looking at average runs added over top six batsmen in the same team across the last two seasons of the County Championship:-


As with the expected chart, Ollie Pope is virtually off the charts here, and as I wrote about here last summer (before his England recall) he should already be a player England plan to build their batting around for the next decade or longer.  Simply put, you just don't see players of his age produce numbers like he does - he's an extremely rare talent in red-ball cricket.  

Interestingly, Hasan Azad also performed well via these metrics and although the Leicestershire opener has only played one season, I'm still pretty confident in his longer-term ability to replicate, or come close to replicating these performances.  He has very strong Second XI numbers (even dating back years to when he was at Nottinghamshire) which gives a little more confidence in his small sample size.  The current England openers (when fit) of Rory Burns and Dominic Sibley also rated well and with the exception of Azad, it looks like England are picking well when it comes to integrating new batsmen into their squad, although those on social media clamouring for the recall of Gary Ballance will be heartened by these numbers.  His Yorkshire team-mate, Tom Kohler-Cadmore, rounded off the seven players adding over 10 runs more than their top six team-mates on average.

Another way we can profile potential batsmen who could step up to Test cricket would be to compare their expected Test batting average with this runs added over top six team-mate measure, and this produced the following chart for the same players:-


Using these measures combined, Pope is out on his own as the outstanding batsman of this condensed group of players, while the previously mentioned Dominic Sibley, Rory Burns and Gary Ballance - all of whom have been capped by England, with the former two the likely opening partnership moving forward - also featuring strongly.  The uncapped duo of Hasan Azad and Tom Kohler-Cadmore look like the two best uncapped batting options for England moving forward.  

I also performed similar analysis on the bowlers, looking at their bowling average compared to that of their team-mates in the matches in which they played, to see if the extent in which they outperformed their team-mates against the same groups of batsmen in similar conditions.  This generated the following results:-


When Ollie Robinson was called up by England, I must admit I was a little surprised he got the nod ahead of several other bowlers, but these numbers should illustrate why he's highly thought of by the national team's selectors.  Whether they look at this type of analysis is, of course, unknown, but he adds huge value to Sussex compared to his team-mates.  Performing this more detailed analysis in conjunction with his expected Test bowling averages would indicate to me that he's likely to be a strong bowler in Test cricket currently, and while he might have a job ousting Jimmy Anderson or Stuart Broad in the short-term, but could well be a long-term successor.

As with the batsmen, I looked at expected Test bowling averages in conjunction with this 'value added' metric and this created the following chart:-


With the top-left corner (low expected Test bowling average, high value added over team-mates) being ideal, again Ollie Robinson is highlighted as being an extremely strong option.  However, a few other bowlers rate really well as well, with Ben Coad and Lewis Gregory showing up very well as well via these metrics combined.  We've already discussed Gregory in some detail and he must be virtually smashing down the door when it comes to England inclusion across several formats, so I'll focus on Coad here.  

One of the most satisfying aspects of working with player recruitment data is that you find relatively low-profile gems on occasion, and several teams have already benefited from my analysis in this area.  Coad looks one of those gems - not only are his numbers superb, but he's strongly outperformed his Yorkshire team-mates with the ball in the matches in which he's played, and as a relatively unexposed player (just 61 career matches, 36 in the County Championship) who has just turned 26 years of age, I'd anticipate continued improvement as he progresses along the age-improvement curve as he advances towards peak age.  He should be another bowler who England have on their shortlist for inclusion.

A nod to several other players should be given as well.  Jack Leach, already around the England team, is the only spinner in this group and deserves credit simply for this alone given that he bowls in a generally very pace-orientated country, and my numbers indicate he should be England's number one Test spinner currently, if we are rating bowlers simply on their bowling alone.  In addition to this, Ryan Higgins - as mentioned before - has performed well with both bat and ball, and another all-rounder, Ed Barnard, also showed up very well in this bowling analysis.

Following all this analysis, we can create a shortlist of players who have produced at least a reasonable sample size of county cricket data across the last two seasons, and I would suggest the following players should be worthy of consideration for the England selectors in red-ball cricket, either now or in the not too distant future:-

Already in many recent squads:-

Ollie Pope
Dominic Sibley
Rory Burns
Jack Leach

Further Batsmen:-

Hasan Azad
Gary Ballance
Tom Kohler-Cadmore

Further All-Rounders:-

Lewis Gregory
Ryan Higgins
Ed Barnard

Further Pace Bowlers:-

Ollie Robinson
Ben Coad


Certainly when it comes to batsmen selection in Test cricket in recent times, England have got a lot right.  Pope, Sibley and Burns - the main successes in the last year or so - absolutely deserve their places based on my numbers as well, while England seem to have jettisoned most of the batsmen which my numbers are less enthused about as well.  Bowling-wise, they perhaps make more contentious decisions, based on my numbers at least.

It would also be interesting to see how these players stack up against players in and around the England squad who perhaps aren't mainstays of the England team who have played a decent amount of county cricket as well in recent years - players such as Joe Denly, Sam Curran, Dominic Bess and Olly Stone in particular, as well as Jofra Archer who is now an integral part of the England set-up but was picked via performances in county cricket first.  

Having said this, the main purpose of this article was to use detailed data to illustrate players in county cricket who have the potential to step up to Test cricket, and in addition to this, I don't want to provide all of the answers to all of the questions.  However, decision-makers at any cricket team can get in touch via sportsanalyticsadvantage@gmail.com where I can discuss complete solutions to recruitment, selection and strategy decisions.

Comments