, ,

I’m getting a bit annoyed at how the sports management company Michael Vaughan is involved with, ISM, seems to be good at getting their players talked about and then selected for England. For example, in 2013 Vaughan bigged up Joe Root on his various media platforms and he was selected to open ahead of Nick Compton, whose county championship form was much better. And this week, after much promotion by Vaughan again, James Vince was picked to join the England Test squad. His LVCC average last year was 32.70. OK, so the selectors see something in both of these players that makes them think they will be better than their stats suggest (and Root has turned out to be pretty amazing), but it does feel as though the media pressure is playing a part. I suppose it is ISM’s job to promote their clients, and well done to them for being good at it, but it suggests the England selectors are too easily swayed.

But is this just my perception? I decided to look up the LVCC batting averages of the most recent England recruits in the season before they got their first Test caps, and compare those who are represented by ISM to those with other or no agents.

If we include James Vince, 22 players have received their England Test cap since 2010. I have divided them into batsmen, all-rounders, keepers and bowlers as follows:
Batsmen: Morgan, Taylor, Compton, Root, Ballance, Robson, Lyth, Hales, Vince.
All-rounders: Shahzad, Patel, Woakes, Stokes, Borthwick, Ali.
Keepers: Bairstow, Buttler.
Bowlers: Kerrigan, Rankin, Jordan, Wood, Rashid.
Some of these are up for debate (was Woakes picked as an all-rounder and Rashid a bowler?) but I’ve played with the uncertain ones and it doesn’t affect the analysis.

Six of these are represented by ISM: Shahzad, Root, Stokes, Borthwick, Buttler and Vince (http://www.ism.golf/cricket).

Let’s exclude the bowlers – batting averages aren’t a good way of measuring their success, and none of them are represented by ISM anyway. For the others, you would expect their batting average in the previous season to represent roughly how they were playing when they were selected to play for England. The figures are summarised in this table, which is sorted by role and then by batting average in the season before they were picked:

Name Season capped Batting average year
before first cap
Nick Compton 2012/13 99.60 No
Adam Lyth 2015 70.39 No
Gary Ballance 2013/14 64.90 No
James Taylor 2012 55.24 No
Alex Hales 2015/16 51.05 No
Eoin Morgan 2010 50.81 No
Sam Robson 2014 47.20 No
Joe Root 2012/13 41.91 Yes
James Vince 2016 32.70 Yes
Chris Woakes 2013 73.50 No
Moeen Ali 2014 59.16 No
Samit Patel 2011/12 47.56 No
Scott Borthwick 2013/14 41.51 Yes
Ben Stokes 2013/14 31.56 Yes
Ajmal Shahzad 2010 22.73 Yes
Jonny Bairstow 2012 48.52 No
Jos Buttler 2014 36.28 Yes
Adil Rashid 2015/16 34.70 No
Chris Jordan 2014 24.84 No
Mark Wood 2015 14.11 No
Boyd Rankin 2013/14 10.66 No
Simon Kerrigan 2013 8.08 No

As you can see, ISM clients always have a lower batting average in the previous season than non-ISM clients filling the same role. Not only that, but all the ISM clients have a lower batting average than all the non-ISM clients – even when comparing ISM batsmen with non-ISM all-rounders (see fig. 1).ISM graph 2Figure 1: Batting averages in the county championship the year before they were capped for England of Test batsmen, wicketkeepers and all-rounders since 2010. Box-and-whiskers plots mark the median (bold line), inter-quartile range (box) and total range (whiskers), with outliers (beyond 1.5 times the interquartile range) marked as points (Nick Compton, I’m looking at you).

Batting average is normally distributed, so I could model this as a general linear model with previous year’s average as the dependant factor and type of player (batsman, keeper or all-rounder – remember bowlers were excluded from this analysis, although the results were the same when they were included) and ISM or not as explanatory variables. This is a very small sample size, and therefore the power of this test is very low, but I nevertheless got a positive result: ISM clients have a significantly lower batting average in the season before they are picked for England than non-ISM clients (GLM, F=10.7, d.f.=1,13, p=0.006). The model estimate for the size of this difference is that an ISM player will have a batting average 24.7 runs lower than a non-ISM player the year before they are picked for England.

This could be read as suggesting that the bar is lower for ISM clients – that you don’t need to do as well in the county championship to be picked for England if you are represented by ISM. Of course, this isn’t proof that that is the case – it may be that ISM are really good at picking out players with international potential, which is also what the England selectors are looking for, and they can both see it despite poor county cricket averages. But it does make you wonder.

I should emphasise that this isn’t in any way an attack on ISM or their clients. (They represent Paul Collingwood, my greatest hero!) I’m just interested in how the world of player management works, and whether it is affecting the international cricket we all love to watch.