So Diego Costa was all the rage on my Twitter time line last night following Liverpool’s reported £21m bid for the Brazilian Atletico Madrd striker.
Bar one or two notable exceptions, the news has been welcomed with great fervour, especially by Liverpool fans.

In such a situation it would be rude not to run the Statsbomb eye over what I presume to be Liverpool’s intended replacement for Suarez.


First off, let’s start with a look at his stats for last season.  To aid comparison I have added his numbers to the recent data that I pulled together for my Chelsea strikers piece.

(The table is interactive and it can be sorted by the user, it is a wide table but the columns can be scrolled across.)

[table id=14 /]

OK, so I know that he wasn’t Top Dog at Madrid, that honour went to Falcao, but those per 90 numbers aren’t great.  After slating Torres in yesterday’s article for his subded numbers I’m surprised to see that Costa’s shots and shots on target per 90 are on a par or worse than Torres’ figures.

However, what Costa did much better than Torres is convert the chances he is presented with.  This is why his goals per 90 of 0.44 beats Torres’ 0.28 out of the park.

On the downside, his turnover numbers are really bad.  I have created a metric “Turnover per SoT” and this could be subtitled the “Frustration Measure” as it quantifies the numbers of lost possession in terms of the shots on target a player had.  The thinking here is that you could tolerate a lot of lost possessions if, on balance, the player manages to strike a lot of efforts on goal as the trade off is worth while.

You can see that Costa is even more frustrating that Torres, as his 59 turnovers dwarf the 23 shots he put on target.  Costa’s Frustration Measure is more than double that of Ba and Lukaku.

I had a look at Suarez’s Frustration Measure, and remarkebly his is just 1.14 (79 tournovers and 69 SoT).  That demonstrates just how high Costa’s number of 2.56 is.

ExpG Measures

If you haven’t seen the ExpG and ExpG Eff metrics yet please take a quick look at my previous post (here) where I give an explanation of these new metrics.

Costa’s ExpG Eff of 1.09 marks him out as a pretty neat finisher.  This means that he has scored 9% more goals with the chances he had than an average player would have done.  For information Suarez posted a number of 1.08 last season.
His Average ExpG value per shot was 0.18.  You can see that this was higher than the figures posted by the 3 Chelsea players meaning that the shots he takes are of better quality.  Again for info, Suarez’s Average ExpG per shot was just 0.11.  This clearly demonstrates the poor shooting choices that Suarez made last season (and undoubtedly even before then as well).

So what do we make of his stats?

I’m surprised at how quiet he was in terms of the positive stats.  What he did, he did well, but given the Twitter praise that was being heaped upon him I expected to see higher figures being posted.
His strength seems to be in hitting the target, albeit on average the shots that he took on were fairly routine.  Of course, this canny choice of shot selection is a skill in itself but his ExpG Eff of “just” 1.09 means that he didn’t shoot the lights out.  I would therefore conclude that his accuracy was at least partly attributable to his shooting positions.

He looks like he will be a frustrating player to support and play with due to his enormous amount of lost possessions.  Having 2.5 times more turnovers that shots on target is a pretty poor figure.

Shooting Positions

Let’s look at his shooting positions:


The key at the foot of the chart should help explain things, but in summary green = goal, blue = save and red = miss or block.  The solid colours are shots and the checked dots are headers (he took no direct free kicks).

He appears to favour the left side of the pitch with a lot more of his shooting action coming from that left side.

7 of his 10 goals came from inside the 6 yard box, with the other 3 goals also coming centrally from well within the confines of the penalty area.

5 of his 10 goals were headed attempts.  In fact, heading appears to be one of his strengths as his 5 goals came from just 14 headers.
This superb conversion rate of 36% was the highest across the Big 5 leagues last season for players with at least 14 headers.
To provide some context; Postiga converted 19% of his headers last season, Kießling 16% and Negredo achieved 13%.  OK, so Costa’s headers are from prime heading positions but he obviously has a decent technique to achieve those accuracy and conversion rates.

As was expected from the shooting stats we observed at the top of this piece, his shot locations were smart with only 3 shots of the 50 he took coming from outside the penalty area.  In terms of his choice of shooting locations, Diego Costa certainly has an entirely different modus operandi to Luis Suarez.
If you were being a little critical you could argue that he didn’t score with any shots that weren’t straight in front of the goals and within fairly close range so perhaps he may lack that little spark that great strikers possess.

Shot Placements


I like his shot placements.
The majority of his on target shots were kept low and he wasn’t afraid to try to find the bottom right hand corner.  The total absence of shots that would hit a stationary goalkeeper’s torso or head is a nice aspect of his shooting.


If lost possessions could be ignored, Diego Costa is efficient but quiet.   However, they can’t be and he seems to be extremley wasteful in terms of lost possessions.
Purely from a shooting point of view, whatever he does he does well.  He also has the benefit of  providing a significant heading threat to any defence he comes up against.

However, I can’t help but think that he is missing a little something that would make him special.  His lack of involvement in an attacking sense is somewhat disappointing and, personally, I would have liked to see some more fireworks from his shooting, and certainly less lost possessions.

But perhaps, as we are talking about Liverpool an efficient, quiet striker is what is required for the new season, albeit they would probably prefer one that is less frustrating than their current striker.

EDIT – I have created a short piece where I justify why I’m not looking at Costa as a winger

  • Bob

    Another nice piece Colin. At a time when so many analytical wannabes are now jumping onto chance conversion as a means to rate forwards (without even considering sample size or other attributes), you’re miles ahead of the curve.

    • Colin Trainor

      Thanks for the kind comments Bob, they’re most appreciated.

  • Toshack

    Thanks Colin and I can only echo Bob’s comment. Nice work!

    Being a Liverpool fan myself, yes I would have like for a less “frustrating” player (judging by the stats at least), but his ExpG stats looks good anyway. As does his shot placement.

    Liverpool could use some heading power although the “tiki-taka” style generally focuses on playing on the ground. I haven’t see Costa play myself so don’t know how “static” he is – I guess not since Rodgers want to buy him…


  • Terry

    Think you’ve made a fundamental error there mate. Costa doesn’t play in the same position as the players you’ve compared him to. They’re all strikers whereas Costa played almost entirely on the wing.

  • Colin Trainor

    I was tempted to edit the article, but I decided against that and will instead add something in the comments.

    A few people have said that the comparison I have made is unfair as he isn’t an out and out striker.
    His average position (according to who use Opta data) was higher up the pitch than both Ba and Torres (I looked at them for my piece yesterday) and only slightly behind Lukaku.

    As stated in the article, Falcao may have been Top Dog at Atletico but I stand by my assertion that his shooting numbers are poor for someone with such a high average position.

    • Toshack


      I knew I had read something about Costa somewhere and finally found it.
      “This is the player whose development made it past time for Falcao to go do his thing somewhere else. Assuming he can stay healthy (and he has had some injury issues), Costa is ready to become a major player for Atletico, as he assumes their primary forward role.” That’s what Ted said in early June. Which seems to support your stance on Costa’s possibilities (and role).

      And speaking of “frustration” Ted ended his piece with “Okay, I’d also be slightly worried about his discipline. A forward that wracks up 11 yellows in 24 starts? Who does that?” Interesting times ahead – if Liverpool really do sign him…

  • Colin Trainor

    OK, I created a very short piece to highlight Costa’s average positions. It can be found here:

    • Toshack


      I’m with you on this one. I don’t think it matters how people perceive where Costa was playing or not – even though you have shown in your screen shots he did have a “striker” position. If Liverpool does sign him it will be for the potential he carries with him, as pointed out by Ted in his analysis.

      Winger, wide forward, striker – or not. According to Ted’s analysis “He (i.e. Costa) and Falcao had similar scoring rates, despite the fact that Costa clearly was not the primary option.”

      Enough said, I think?

  • Terry

    Not meaning to be a pain here but can I ask how your “heat map” works. Is it an average position over 90 mins?

    If it is then I need to point out that Costa switches wings regularly throughout his games. Wouldn’t that average out as a central position?

    • Colin Trainor


      The average positional map is created by Opta, and it is the average position of all the touches that the player had during the match.

      And yes, if he hugged the right touchline for the first half and the left touchline for the second half (and had an equal amount of touches in both halves) then the average position would be right up the middle.

      Are you saying this is the case? Not just that he switched wings between games, but that he continually switched wings during games also?

      Also I had a look at Squawka (as they show the team formations as provided by Opta) and in most of his games he is shown as 1 of 2 front men, not as a winger.

  • Terry

    He would usually work both wings at various times during a game.

    Watch this video mate. It’s DC against Celta. He pops up all over the place.

    • Colin Trainor


      Atletico’s last game of last season:

      Go look at the Action Areas for Costa. More than 60% of his touches are up the middle.

      Yes, he played some of his time on the wings, but the amount of time he spends in the middle justifies much better shooting numbers IMO.

      • Terry

        Why have you picked out one of only 4 league games where Falcao wasn’t in the team?

        I’m gonna leave it there. I watched at least a dozen Atletico games last season and I know what I saw. Maybe in future you should watch a player play before you try to use stats to define and judge him.

  • Terry

    ooops – just posted a link. Not sure what happened there. Sorry.

  • Elston

    It’s not so much a question of where he plays but what his job is. If you look at his creative stats, like key passes, which are really good, you can see that his primary job was as a creator, not a finisher.

    Ozil has central and high position on these average position diagrams (though admittedly not as high as Costa), but if you put him on this list it would be absurd, so long as you didn’t include key passes, clear cut chances created, etc. Lukaku and Ba are 95% simply out there to score goals. That’s not Costa’s role.

    It also doesn’t make much sense to calculate your “frustration measure” without including other positive contributions, especially key passes. When you’re comparing pure 9s like Lukaku and Ba, the difference probably doesn’t matter much, but I guarantee (in fact just look at Mixedknuts’ post when he profiled Costa for the relevant stats) that his numbers would look a lot better here if you looked at the bigger picture, and didn’t analyze him as if his job was to be Ruud van Nistelroy.

    • Bob

      Well it would be absurd to put Ozil in the list because his primary function is to create. That isn’t the case with Costa, who created a chance every 54 minutes last season (50th best in La Liga). His creative stats are not ‘really good’. Far from it in fact, they are worse than Aspas’

Improve Performance and Productivity in Your Club:
State-of-the-art Football Analytics