The Data Scout – Sandro Ramirez

Everton’s pursuit of 21 year-old forward Sandro Ramirez is complete.

Apparently, the lad’s a goalscorer.

Sandro himself: “My key strength is goalscoring. It’s always good to bring goals to a team and that’s what I’m hoping to do.”

Koeman: “He will bring a lot of good qualities and he’s a really good goalscorer, too. There was a lot of interest in Sandro because, of course, he’s a goalscorer”

14 goals for mid-table La Liga outfit, Malaga, last season in his first ‘proper’ year of first team football does indeed look pretty good. And, blimey, there were some good uns:

Controversial claim alert:

Long term, Sandro may not be that natural goalscorer.


Feel free to come back and haunt me at a later date, but hear me out…

Here’s all Sandro’s shots mapped out nicely on a viz. I say nicely, but this fella was the Elmer Fudd of La Liga last season discharging his double barreled shot gun whenever the goal came into view (and sometimes when it didn’t):

The Potential Problem

Sandro didn’t have the shot profile of a centre forward last year, just the goal tally of one. There aren’t enough shots from good areas in that set of shots. To illustrate this, I’ve filtered the viz for higher value chances according to my expected goal model (xG) and run it for the players with similar goal tallies to Sandro in La Liga last year:

Note the dearth of dots compared to the others. The difference is even more stark if you add the big boys like Messi, Neymar and Suarez in. I’ll put it another way: simulate that set of shots over and you can see that the average shooter in the average season would most likely score 7 goals, not 14:

The chances of a shooter getting 14 goals here is miniscule. The shooter is more likely to score 2 goals than he is to score 14.

Alright, to really break it down, Ronaldo was par with xG last season, Messi was 36% over his, Luis Suarez was 40% over his and Sandro was 88% over his. Unless Everton start laying it on a plate for him in front of goal, Sandro’s finishing will almost certainly regress back to less superhuman levels.

It’s not as if Malaga didn’t supply the ball into the box in those good areas in front of goal. They did. It just wasn’t Sandro getting on the end of them. Veteran striker Charles Dias got on the end of more of them than the youngster despite playing half the minutes:

What are Everton really getting then?

Good question. The lad can clearly play. The lad can clearly strike a football properly. But what’s his style and how did he fit in? Malaga were reliant on slinging the ball into the box from out wide when it came to creating chances in the mixer:

Sandro might as well have been playing for a different team. Here’s how his chances came to him:

Malaga don’t have too much in the way of great passing in the opposition half according to my xP model. But for a centre forward, the model likes Sandro’s passing. He has an eye for a difficult pass too, perhaps frustratingly being more wayward with the easier ones.

Despite this, his direct creativity was low overall. He only had two assists to his name last season which was in keeping with his expected assist (xA) numbers. This again perhaps highlights the the style clash with the rest of his team.

As stated above, the team in general crossed the ball a lot but he seemed quite good at sliding the ball into the box from more central areas to create opportunities for team mates:

Everton fans looking for a no.9 who holds it up better than Lukaku might be a bit disappointed. My model think’s he’s ok at it, but not much better than that (and my model think’s Lukaku’s better at it than you probably do).

We only have one full season of data for Sandro, but it seems likely that Sandro is a more of an all-round player than both he and his new manager give him credit for (at least in short public utterances).

Whether he’s the out and out goalscorer they think he is will depend on him and his team mates getting him free into the box a lot more than he has to date in his short career. Otherwise we’re likely to see a forward who gets deployed not just centrally but all across the top line as he finds his feet in the Premier League.

Follow me on twitter @footballfactman

Posted in Sports | Tagged , , | 6 Comments

The Data Scout – Gylfi Sigurdsson v Ross Barkley (Extended Version)

All the data below is taken from the last two seasons of Premier League football.

Do they even play the same position?

My favourite way to test this using data is by seeing where players receive the ball on the pitch.

I’ve split the pitch into twelve positional zones and used a machine learning method called k-means clustering to group players together who receive the ball in similar areas.

This puts Sigurdsson in a group with the likes of Ozil, Coutinho, Silva, Firmino and Rooney.

Barkley fits in with a group of players (Ramsey, Steven Davis, Wijnaldum and Pogba) who receive the ball in more orthodox central midfield areas rather than further upfield ‘in the hole’.

Nominally, Sigurdsson is more of a 10 than Barkley. Actually, however, the difference between the two is pretty small with both lying on the edge of their clusters about to overlap. Do they play in the same position? Most of the time, yes.

What do they do with the ball?

I said above that the players receive the ball in much the same areas. What they then do with it is where the two players start to really diverge.

Consider the viz below showing where each passes the ball when in possesion on the left hand side in the final two thirds (blue successful, red unsuccessful). Barkley’s range is much more expansive:

My expected passing (xP) model, also starts to highlight some big differences and really spotlight those original clusters the two were put into.

The xP model tells us how difficult passes are. For example, a sideways pass by a centre back to a fellow centre back in front of their own box will have a high probability of being completed (95%+).

With Barkley and Sigurdsson, I filtered the database so only harder passes would show on the viz. Basically, that’s those passes that get completed less than 25% of the time in the top leagues. Here’s how they look:

You can see Barkley there trying more vertical passes from deeper central positions while Sigurdsson is all about putting the ball in the box from out wide.

This is where Sigurdsson’s game falls down. He’s well below average at making the more difficult passes stick. Evertonians will be interested to note that Sigurdsson profiles like a poor man’s Rooney here or a very poor man’s Coutinho/Silva/Ozil/Firmino.

Barkley on the other hand is above average at making more difficult passes stick and profiles as a poor man’s Yaya Toure, or a very poor man’s Fabregas.

On top of this, over the last two years Sigurdsson’s involvement has dropped to the point where Barkley is passing the ball twice as much as the Icelander during a game.

Neither comes out as anything like the man to take Everton onwards and upwards but Barkley is a clear winner in the general passing stakes.

I was surprised by how poorly my model rates Sigurdsson’s passing. It’s one you’d want further opinions on from the video guys and the scouts in the stand.

You may have read that Barkley was one of the most creative players in the league last year.

If you look at number of chances created, then yes, Barkley was in the top 10. But so was Sigurdsson.

However, we need to look under the hood because as is clear, all chances are not the same. That’s why there’s such a thing as an expected assist (xA) model. My one is based around where on the pitch the ball was delivered from and to and how often these types of pass were converted historically.

Starting with open play, 201 players over the last 2 seasons have played 1000 minutes and directly created at least 20 chances. Once sliced and diced for how good their set of created chances are, and for how many minutes they’ve played, Barkley ranks 70th and Sigurdsson is 160th.

What? Surely that can’t be right?

You don’t trust my model!? Here’s the top 5: Alexis Sanchez, De Bruyne, Ozil, Fabregas, Hazard. And you don’t have to play for an elite team to get amongst this lot. Tadic, Deulofeu and Payet are right behind them.

Essentially, Barkley and Sigurdsson just aren’t that good at consistent, high quality chance creation in open play. Neither get into, or pick people out in the box enough. Yellow arrows are assists:

Barkley’s had an elite forward in Lukaku to convert the chances he has delivered into actual assists for goals. Sigurdsson’s had…well, we’ll just leave it there.

Set pieces are a different story. Sigurdsson rules:

Having a dead ball specialist is really nice. But Everton have a team problem with creating chances in open play. The Blues are way (way) behind the elite clubs. If Everton lose Lukaku this summer, and don’t get a proper creator in, then the amazing attacking efficiency they’ve posted over the last two years will nose dive.

Who’s the bigger direct goal threat of the two?

Both hit the ball from fairly bad areas and they do it a lot. Both have been worth around 11 or so goals in open play over the last two seasons according to my expected goals (xG) model. Both have scored 10:

Again the difference between the two players shows in dead ball situations. Again, Sigurdsson is king:

The Icelander is adept at getting his free kicks on target. This is a really precious skill to have.

What about off the ball?

Swansea and Everton win the ball back at very similar rates and in very similar areas. Not so for Barkley and Sigurdsson. Barkley not only wins the ball back a lot more than Sigurdsson but in much deeper areas too:

We keep hearing Koeman talking about pressing the ball during his press conferences. To the eye, Barkley often seems physically incapable of being consistently proactive in a higher press. Is this one of the things he likes about Sigurdsson?

After just a short dive into the data, it’s becoming clear that these two players’ roles haven’t been that similar at all in the last two seasons.

If the idea is to bring in Sigurdsson for what he does now, then already the data is flagging up some big worries. If the idea is to bring him in for something different then he’s got to adapt to that, a team in wider transition and a change in living circumstances.


If you’re interested in similar profiles for your own football club recruitment department, then get in touch:


Posted in Sports | Tagged ,