Scouting with Stats: Belgium
Written by: Christian Wulff @ahellofabeating
Let’s be clear: you should never recruit a player solely based on their stats. A well-trained eye on a player over several games is always preferably to gazing at a spreadsheet. But here’s the thing: Scouts only have one pair of eyes and can only be in one location at a time. They demand a salary and are costly to fly around the world. Clubs with limited financial resources – such as the majority of Scottish clubs, will only have a few, if any, full-time scouts, instead often having to rely on the same limited networks of ex-teammates, old coaches and agents to bring in players.
The increase in the availability of video footage through scouting companies such as InStat and WyScout do help – scouting analysts can now instantly view detailed and extensive footage of players around the world, handily dividend into different categories like passes, tackles, shots etc. But again, time and resources are limited. Extensive scouting analysis takes time, whether it is in person out in the freezing rain or in front of a laptop in a cosy, heated office.
This is where the statistical input can have maybe its greatest value in a player recruitment process: helping to focus the (often limited) scouting resources a club has to where it might have the greatest chance of success. Essentially, stats should become a filter. For example, where do you start if you need a left-sided attacker who can create chances or a powerful centre forward that can be a physical presence and with good link-up play?
Statistics lets you – within minutes – cast an electronic eye over almost the entire footballing world. Good stats analysts will know which categories to evaluate and how to judge those numbers, ending up with a list of potential players that match the criteria given by the manager or sporting director.
The second part of the process can then start; watching footage of the players, talking to contacts that might know more the willingness of the player to move, their personality and adaptability and if the club would consider a sale. A further short-list produced, clubs can now allocate their precious resources to watch the players live, given them a further perspective not available through stats or video analysis.
Stats can also be used at the other end of the process, as a final sense-check; A recommended player from a scout or agent comes in and the reports are good. Statistical analysts can complete the picture; what does the data over several seasons say, does it flag anything up that is worth further investigation?
Player recruitment (and the wages that follows it) will usually be the greatest expenditure of any club , so the process that leads to such investments should be subjected to scrutiny and due diligence, as it would be in any other business. Statistics, together with traditional scouting and the gathering of non-playing information will help give as much context and re-assurance as possible. It won’t eliminate all mistakes and guarantee successful transfers, but done correctly it will at least increase the chances that this new employee your club is set to spend its precious resources on is worthy of the investment.
With the limited resources available to most of the teams throughout Europe – especially compared to an increasingly rich footballing elite – they will need to look for potential rough diamonds and accept a much larger degree of risk. Clubs with a lot of money can buy the best players in a league and even if they fail, they’ll have more than enough money to throw on another player. Smaller clubs most look for the unappreciated players, and take the chance on raw talent that they can develop and hopefully bring to their full potential.
So what if we applied this approach to help suggest players for the clubs in the Scottish Premiership using statistical analysis.
Our data provider Stratabet covers a multiple of leagues in Europe and on other continents that should have players that can provide very good value for Premiership clubs. Using statistics, we at Modern Fitba will in this scouting series try to identify players that could be worth a much closer look by SPFL clubs.
We start with a footballing nation that is in a middle of a golden period – Belgium.
One of the main areas a stats analyst will look for in trying to identify ‘hidden talents’ when it comes to attacking positions is underperformance in Expected Goals (xG), i.e. a player who has come to a lot of chances (quality of which is measured though xG) but has not produced the actual goals usually expected from such chances.
As I discussed in our guide to Expected Goals this might seem counter-intuitive: why would you consider missing a lot of chances a good thing? Simply put, if a striker is missing a lot of chances over a certain period, they will start converting them at a higher rate very soon. We know that the vast majority of players will over a longer period align their actual goals scored with the xG value of the chances they get to – if they are getting to a lot of chances and not scoring, the likelihood is that they will start scoring as long as they keep getting into these same positions.
For all players who played at least 900 minutes (or 10 full 90min) in the Belgium Jupiler Pro League last season, Silvere Gaanvoula was the greatest underachiever when it came to xG. He’s an interesting prospect:
Name: Silvere Ganvoula
Born: 22th June, 1996
Club 2017/18: Anderlecht, Mechelen (loan)
Nationality/Caps: Republic of Congo / 6 (1 goal)
Height: 1.91m (6’3’)
It’s been a frustrating season for him. Looking at the data and shot map above, 66% of his attempts for the season came from within the Danger Zone; he gets into some great positions in the box, resulting in good quality chances. 52% of his attempts are headers, 2nd highest in the league – Ganvoula is definitely using his height and physique to his advantage.
His xG per every 90 minutes played (p90) is 0.47 – the 9th highest in the league – and he took 2.9 attempts p90 which was the 4th best in league. So in terms of both quality and quantity of chances he got at the end of – for two different teams – he’s been excellent.
The problem? Converting his chances. 98 players in the Belgium league played more than 900 minutes and had on average 1 or more attempts p90. Of those 98, Ganvoula was 80th in Shot Conversion (% of attempts scored), 89th in Shot Accuracy (% of attempts on target) and 70th in Shot Percentage (% of attempts on target score).
These are not pretty numbers and a big reason why even though his xG number show Ganvoula was ‘expected’ to score almost one goal every two full games he played (based on his chances), his actual scoring rate was a goal in less than every five full games. This makes him the biggest underachiever in the league when it comes to scoring goals vs the chances he’s had.
From the footage available, Ganvoula’s finishing technique – especially with his feet – needs work, and he’s taken some ridiculous long-range shots, probably out of frustration. But it is also clear why he’s getting to all these chances; his movement inside the box is clever and usually well-timed, and there’s examples of him using his physique well to hold up the ball and be a part of the build-up play for his team.
So after checking his stats and watching some footage, what about the opinion of someone that has followed him over a whole season? Alastair Gartenberg is the Jupiler Pro League expert for the football analytics and consultancy company Football Radar. His jobs consist of watching every single game played in the Belgium top-flight.
His observations of Ganvoula match well with what we’ve been able to surmise from the data and footage so far:
‘Ganvoula is a tall, rangy striker who divides opinion to say the least. He is very good in the air and his main assets are attacking headers and flicking the ball on for team-mates. He is also quite quick and physical, but technically he is very poor. His finishing is inconsistent at best and in terms of passing and control he is pretty much useless...’
Looking more closely into Ganvoula’s season there are some mitigating circumstances for the young striker. He only started 10 games, and was subbed off in six of them (on average he came off in the 69th minute), with an additional 15th appearances from the bench (an average of 15 minutes each time). When Mechelen got a new coach in October, his minutes on the pitch dried up and Ganvoula’s father sadly passed away at the same time.
These are all factors that can very much play in when considering’s Ganvoula’s very inconsistent finishing; a young striker going through a difficult time off the field, desperate to prove himself and snatching at the chances he gets during his limited time on the pitch.
As he had already played for Anderlecht before his loan move with Mechelen, the club could not send him out on another loan within the same season. With great predictability, the two non-penalty goals Ganvoula’s did score in the league last season both come in the same match against Mechelen:
In terms of an potential switch to the SPFL, Alastair Gartenberg thinks it could be a possibility: ‘Anderlecht have signed two other strikers this summer so I think he would definitely be open to a move away from the club and perhaps he would suit the SPFL quite well. He may be more likely to leave the club on loan as Anderlecht invested £1m on him in January 2017 and he has a contract until 2021, but they may be willing to sell him if they got a good offer.’
Silvere Ganvoula would be a risk for a Scottish club; there are obvious weaknesses to his game and he is coming off a tumultuous year. But there is potentially a big upside; he does have some extreme qualities, especially in the air, and he will definitely not be physically intimidated by any centre-backs in the SPFL. Under a manager willing to trust him and able to build up his confidence, he could be a very good fit for a team where his qualities can be put to good use and where his deficiencies can be hidden.
Perhaps suitable for a club like Motherwell (who had the highest amount of headed attempts last season, both in total and as % of all attempts) or even Aberdeen, where Derek McInnes often likes to play with one strong and traditional centre-forward such as Ganvoula, and desperately need some reinforcements up front.
UPDATE: Just as this article was about to published, news came through that Ganvoula has been sent on loan to 2. Bundesliga side Bochum for the 2018/19 season. So too late for any SPFL clubs now, but it'll be interesting to follow his progress in Germany
One area where advanced statistics can be put to very good use is identifying underappreciated performances from players that excel in creating chances. Looking at a player’s assists total can be notoriously misleading as so much is dependent on the chance being converted into the goal; the pass itself might of very high quality regardless.
The concept of Expected Assists (xA) and Key Passes eliminates that problem; it looks instead at how many chances a player set up (key passes) and the quality of those chances (as measured through those chances’ xG value – i.e. how many assists those chances would on average yield, or xA). The data from Strata that Modern Fitba use also gives us information about the pass before the pass that set up the chance; what we call a Secondary Key Pass, and by extension Expected Secondary Assists (xSA).
Putting these two types of passes together gives us a combined category of Chances Created (key passes + secondary key passes) and xG from Chances Created (expected assists + expected secondary assists). Within this stat category, players are no longer dependent on a goal being scored for them to be credited for their chance creation skills.
The top six players in the Jupiler Pro League last season in terms of Chance Creation per every 90 minutes played are all likely non-starters when it comes to any interest from SPFL clubs. They are either too highly valued financially for Scottish clubs (maybe bar Celtic) or they are probably too old (33-35 years) to fit in term of age profile. The one possibility, Ryoto Morioka of mid-table Waasland-Beveren, has already been picked up by Anderlecht.
However, number 7 on the list is a very interesting option indeed
Name: Roman Bezus
Born: 26th September, 1990
Club 2017/18: Sint-Truiden
Nationality/Caps: Ukraine / 19 (4 goals)
Height: 1.85m (6’1’)
Ukrainian Roman Bezus has spent the last two seasons with Sint-Truiden, the club finishing 12th and 10th in that period. Bezus came from Dnipro in the Ukrainian league and has 19 appearances for his country, although the latest came over three years ago. Turning 28 in September he is now coming into the latter part of his peak years and might well feel the need for a move to re-start his career, having languished in the lower regions of the Belgium league.
When considering Bezus’ data, the Belgium league system throws us a little curveball. When the 16 teams in the top-flight haves played each other twice, the top 6 teams go on to a Championship play-off consisting of ten games, while teams placed 7-15 is joined by 3 teams from the second tier for two groups of Europe League play-offs (the 16th placed team after 30 games is relegated). However, while Strata provide data for the Championship play-off, they don’t for the Europa League play-offs. This means that for clubs finishing 7-15th (like Sint-Truiden did), we only have advanced statistics for the first 30 games.
Context is crucial in data use, so in order to compare Bezus’s numbers to the rest of the league we only use the data from the first 30 games (pre-split) for all teams.
As mentioned, up until the split, Bezus was the 7th best creator of chances in the Jupiler Pro League when it came to the xG value of the chances set up, with 0.43 expected goals p90.
Looking purely at the quantity of chances set up he is actually 3rd, setting up on average of almost 4 chances (3.8) per every 90. For a player at a club towards the bottom of the league, these are very good numbers. Bezus is a threat both from open play (2.6 chances p90 from a ‘moving’ ball – 7th in league) and set-pieces (1.1 chances from a ‘dead’ ball – 6th in the league). He is involved both at the end of the attacks (2.6 times p90 he had the last pass before an attempt) and slightly earlier (1.2 time p90 he had the pass before the pass that set up the attempt).
We’re not given the location of those secondary key passes by Stratabet, but by looking at his Key Passes we can see that Bezus’ very much creates from a central position, usually in the space between the oppositions back four and midfield.
But there’s more: of the six players that created a higher xG from chances p90 than Bezus last season, none of them was as big a goal-threat as him; even excluding any direct free kicks, Bezus had 2.1 attempts p 90, with an xG value of 0.29 p90 (an expected goal almost every three games) from those attempts.
It’s usually just a very small group of players in a league that is towards the top in both getting on the end of chances and creating those chances for others. As you can see by the graph below, Bezus just about gets into that select group in the Jupiler Pro League pre-split.
While we don’t have advanced statistical data such as xG and Key Passes from the Europa League play-off games that Bezus participated in, looking at more traditional stats it seems he definitely didn’t slow up at the end of the season; he added 4 goals and 7 assists in those ten games.
The footage available of Bezus again confirms the picture painted by the stats; a playmaker who can also contribute with goals, especially from set-pieces:
So we have here a player who creates from open play and set-pieces and who also manages to get to plenty of chances himself. Where is the catch?
Well, Bezus only played just over nine full 90 minutes pre-split (he had almost nine after the split as well), which seems odd for a player who hasn’t been injured and seemingly gives his team so much offensively. Belgian league expert Alastair Gartenberg casts some light on why:
‘Bezus is a technically gifted but his work-rate has often been called into question and although he is definitely one of Sint-Truiden's better players, he has been dropped on occasion due to a perceived lack of effort. I personally think this is in part due to his languid style and I do rate him as a player. I think Bezus could prosper in the SPFL but am not sure Sint-Truiden would be willing to let him go unless they got a really decent offer, as they have lost a number of key players this summer.’
Based on all the information so far, it’s easy to imagine Bezus fitting well into a team like Hibernian, especially if Scott Allan does not return to the capital. But also clubs such as Hearts and Aberdeen could do with a spark of creativity, especially through in the middle, and together with Hibs they should all have the resources to test Sint-Truiden’s resolve in trying to capture an experienced player of good pedigree.
While not EU nationals, both Bezus and Silvere Ganvoula have be capped several time and have played in higher-rated leagues than the SPFL, so the granting of work permits should hopefully not be an obstacle for clubs wanting to bring them to Scotland. For various reasons, these players would constitute a risk, but their data also indicate a potential big reward for clubs willing to take that chance.
Resources are limited for most Premiership clubs. I would recommend they spend some of it to take a long, hard look at Silvere Ganvoula and Roman Bezus.
This article was written with the aid of StrataData, which is property of Stratagem Technologies. StrataData powers the StrataBet Sports Trading Platform, in addition to StrataBet Premium Recommendations.