Real Betis, Matt Milnes, Global T20 Canada and struggling ODI Teams

24th June, 2019.


A tweet from the @TalkingLaLiga Twitter account caught my eye last week, stating that Real Betis were one of a number of football clubs using analytics to assess players.  I've posted screen shots of several of their tweets in their thread below:-

For those unaware of Lo Celso, the Argentine attacker was purchased for €25m from PSG following a successful loan spell (with option to buy) at Betis, and based on media reports, will be sold for considerably more this summer.  I'd imagine that the profit on Lo Celso's sale alone will be enough to keep the analytics department at Betis running for decades.  

The second tweet I posted above shows that Betis recruited Lo Celso using data with a specific goal in mind - to replace a departing player.  They looked to find players who had close similarities to Fabian, who signed for Napoli for his reported €30m contract buyout clause value, and Lo Celso had a high statistical likeness to Fabian.  I'll revisit the 'likeness' discussion later in this piece, with context in cricket.

As regular readers are likely to be aware, I completely endorse such methods and analytics in general with regards to scouting players, and I'd be pretty confident to state that recruiting using solely via data would be more successful long-term than solely via a coach/scout's eye test.  For a start, a coach or scout is unable to watch every single match in every single league on TV or live at the stadium, but data can capture every single match in detail, so it has a major benefit in volume of coverage and time efficiency.  Having said this, recruitment via data is not necessarily a completely 100% solution for recruitment or retention, but married with player knowledge from top-level coaches, it is a very powerful combination, and that's probably the ideal solution.

Betis' VP's comments that they will be able to turn to their own databases to see what a player looks like, as opposed to reactive scouting, are very interesting.  Naturally, player databases are extremely useful for any club in any sport and this is a service that I have been providing to cricket teams.  Some English counties, for example, are finding this useful with regards to being in a position in April to make 28-day approaches for players who are going out of contract at the end of the season (I also provide details on contract expiry dates).

One player who was signed following the conclusion of his contract at the end of last season was Matt Milnes at Kent.  Previously of Nottinghamshire, he was unable to get regular game time at a county with a considerable stock of pace bowling options.  However, his data in the Second XI was excellent, and being at a good age open to further improvement, his signing was a smart bit of recruitment by Kent, who were promoted to Division One at the end of last season.

Here's what I wrote about Milnes during my early off-season report on new county signings (which can be viewed at

My County Championship algorithm, which is able to take into account performances in Second XI cricket, as well as future age-curve projections to establish the future ability of a player across different formats when they reach peak age, established that Milnes' Division One bowling expected average for the 2019 season was 29.82, and he's been able to improve even on those high expectations, taking 34 wickets so far in the County Championship at an average of just over 23.

Another player not hugely dissimilar to Milnes in terms of potential ability based on Second XI data and being 'blocked' at a county with deep pace bowling stocks is Toby Lester at Lancashire.  I was pretty surprised that he signed a new contract at the end of last year, because having just turned 26 at the start of this season, he could be a regular starter for a number of other counties.  It was interesting to see that Warwickshire have recruited Lester on a month-long loan several days ago, and they were rewarded with an excellent start from the left-armer, taking 4-41 on debut against champions Surrey.

My point is this - there are numerous players in the Second XI who either have extremely high potential or strong current ability, yet are not nearly first-team regulars - particularly at counties with Test match grounds.  While cricket doesn't have football's transfer system, and therefore teams cannot benefit from gigantic profits from player trading, as Betis are likely to do with Lo Celso, using analytics can enable teams - either in the county game here in England, or at overseas franchise level - to exploit market inefficiencies.

In the internet age, market inefficiencies across the majority of industries are now few and far between, given the prevalence of historical data with regards to problem-solving.   However, this has still to translate itself to the draft and auction markets for T20 franchise tournaments, with there being numerous questionable decisions made by franchise with regards to their recruitment.

Recently, I wrote an article for The Cricketer magazine which proved popular among readers, and you can check it out at  In this piece, I discuss the potential reasons as to why teams recruit badly, so I won't go into any further detail in that area.  However, with the recent CPL and Global T20 drafts, it is utterly evident that teams are still making the same mistakes that I listed in that article in The Cricketer.

The draft system for player recruitment is well established in the US across several sports, and while smart teams will look to try and gain expected value in later draft rounds, most observers will probably have a pretty good idea of which players will be selected in the early rounds - essentially the most popular players.

However, in cricket, player recruitment in the early rounds is far from predictable, with a number of players receiving early round picks despite not being particularly close to being statistically top-level players, or being anticipated to be an early round pick.  In the CPL, for example, the St Lucia Stars rather surprisingly picked Isuru Udana as a first round pick and then Fabian Allen as a second round pick, despite neither player having particularly strong T20 data.   At the age of 24, all-rounder Allen does have some potential - particularly as a boundary hitter - but being picked as a second round pick is a real gamble from the franchise, who have struggled in recent years. 

Franchises at the Global T20 draft in Canada last week were rather seduced by veteran big names, and reputation.  As it is difficult to source a pre-draft player list, it is impossible to know who went unsold, but I would be confident that some players who went unsold would be higher up a smart and statistically-driven shopping list than some who were lucky enough to get contracts.  

Based on my data, the major puzzling picks in the first two rounds were as follows:-

Round 1 - Brendon McCullum, George Bailey, Daren Sammy.  

McCullum's data is on the wane, which has seen his franchise opportunities reduced.  However, this didn't stop Toronto Nationals snapping him up as a round one selection.  

George Bailey was a strange first round selection by the Montreal Tigers - I'd be worried given that he's almost 37 years of age, and had a bad 2017/18 Big Bash.  My concern would be that his better 2018/19 campaign would potentially be an outlier and I'm sure he could have been recruited in a cheaper round in any case (unless he had a very high reserve price, in which case better options probably existed).   

Daren Sammy is now 35 years of age and is another player whose stats are on the wane.  He's frequently a non bowling number seven these days, bowling an average of just 1.87 balls per match from 2018 onwards, and facing 9.72 balls per match as a batsman.  This gives him one of the lowest match involvement percentages (percent of a match spent either batting and bowling) of my entire database.

Round 2- Yuvraj Singh. 

Recently, Yuvraj retired from international cricket and IPL commitments, which appears to open him up for worldwide franchise league consideration.  However, there's one major problem.  He's now 38 years of age and across the last few seasons, he's really struggled with his run-scoring.  

Across the last two editions of the IPL, he's scored 163 runs in 10 completed innings (16.30 average) at a strike rate of 110.14.  His boundary percentage is mediocre, at 14.19%, while his non-boundary strike rate of just under the 50 mark is very poor and aligned with the biggest hitters in the game - those who aren't fussed about singles and deal in boundaries, players such as Chris Gayle, Sunil Narine and Andre Russell.  The problem is, Yuvraj doesn't have nearly the same boundary-hitting abilities as those players, based on that boundary percentage mentioned above - most batsmen with a below 50 non-boundary strike rate are able to hit 20%+ of balls for boundaries, making a huge difference in their overall strike rate.  

In addition, his strike rate across the last two editions of the Syed Mushtaq Ali (effectively Division 2 of the Indian domestic T20 scene) is even worse, running at 100 combined across those seasons.  There really is little in the way of statistical evidence to suggest Yuvraj will add much over the average player in any overseas franchise league.

Of course, where Yuvraj is likely to add value is marketing and brand awareness of the league.  However, very rarely in evolved sports are players signed solely for their commercial value, perhaps an area where cricket is still to catch up.

It could be argued that being a big name player was a virtual pre-requisite for selection in the opening two rounds of the Global T20 Canada draft.  Only two players of those first 12 selected were aged below 30 years of age (Kane Williamson and Chris Lynn) with seven aged 35 or higher.  The average age of player recruited in the first two rounds of the draft was 34.79.  I'm struggling to believe there weren't better, younger, options in the draft.

Finally, back to Lo Celso and profiling players based on likeness.  As a start, we can look at the templates of previously successful teams with a view to working out which players are likely to be an upgrade on current squad options.  In ODI cricket for example, I'm surprised that more teams aren't following the clearly successful England template.  I'm not particularly concerned that they've lost two matches in the current World Cup - tight matches and defeats still have some percentage chance of occurring in any match, let alone the variance-heavy nature of short tournaments - and across recent years, they've evidently been the most successful ODI team with a rather batting-orientated approach, with arguably not the strongest bowling attack.

Despite this, rather strangely, other countries seem rather reluctant to replicate England's strategy.  We've seen a general cautiousness in the first 10 over Powerplay, and a number of curious innings from players seemingly lacking intent, either when posting or chasing.  On some occasions, chasing teams appear keener to safeguard their net run rate, as opposed to actually going for wins.  Certainly very few teams are targeting scores much in excess of 300 batting first.

I wonder why this is, particularly when historical data clearly shows that every single one of the last five winning World Cup teams has an overall strike rate at least 15% greater than the tournament average (Mean deviation 15% or greater), as shown below:-



Tournament Batting SR

Winning Team Batting SR

Winning Team SR Mean Deviation


























Moving on, the graph below illustrates the relationship between batting strike rate and boundary percentages of ODI batsmen from 2017 onwards, in matches between the 10 World Cup nations (minimum 300 balls faced):-

We can see that there is a clear relationship between boundary hitting and overall strike rate, and that every single player with a strike rate in excess of 100 runs per 100 balls had a boundary hitting percentage in excess of 10%.  In fact, only two players with a strike rate of over 95 had a boundary percentage below 10% - Peter Handscomb and David Miller, while only Jason Holder, Shaun Marsh and Faf du Plessis had a strike rate between 90 and 95 and had a boundary percentage below 10%.  So of the 42 batsmen in the sample with a strike rate over 90 - surely a virtual pre-requisite of the modern ODI game - only 5 players (11.90%) weren't able to hit more than 10% of balls faced to the boundary.

Of the teams who have struggled with the bat, Pakistan and South Africa spring to mind.  Pakistan have already announced a 'robust' review, so looking at South Africa in particular, they have a rather ageing squad and will almost certainly go through a rebuilding process after a very disappointing campaign.  Fortunately for South Africa, there are a number of promising young players coming through their domestic system, and we can look at data to profile which players are likely to fit the bill moving forward.  Of their World Cup batsmen, only Quinton de Kock and JP Duminy (and almost Faf du Plessis) can boast a strike rate of greater than 90 and a boundary percentage in excess of 10% in ODIs from the start of 2018 onwards - their issues in posting big totals were fairly predictable.

Looking at the data from the domestic Momentum One Day Cup over the last two seasons, we can enter several filters into a database.  For example, if we add age <=27 (higher potential improvement/upside), boundary % >10%, strike rate >90, minimum 200 balls faced, and having played both seasons, we obtain the following list of players who could be capable of making the step up to international level across the next few years:-

Existing World Cup regular players:-

Andile Phehlukwayo

Other players:-

George Linde
Heinrich Klaasen
Kyle Verreyne
Patrick Kruger
Theunis de Bruyn

Close to filter:-

Gihann Cloete
Janneman Malan
Wiaan Mulder
Reeza Hendricks
Sarel Erwee
Temba Bavuma

We can see that there is plenty of strength in depth for South Africa waiting for opportunities, and it wouldn't surprise me if a number of the above names were given chances in the not too distant future. 

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