Do goals from midfield matter?
Does Luka Modric, compared to midfielders like Frank Lampard, score too few goals?
If there’s been any criticism of Luka Modric during his time in England, it’s been a lack of goals. The Croatian midfielder has nine to his name in three seasons, a sub-par total for some.
The Mirror’s Darren Lewis recently wrote that he felt this low return, in comparison to some of the Premier League’s historically clinical midfielders, would have been a fair justification for Tottenham to sell Modric:
Let’s have it right: Modric can pick a pass and is a magnificent link man, but he doesn’t get you 20 goals a season like Paul Scholes did and Frank Lampard does now.
He is not as explosive with the goals from midfield like Freddie Ljungberg was at Arsenal, or Cristiano Ronaldo was at Manchester United.
So, as long as Redknapp strengthens wisely – the word is he is looking at up to four players – then what’s not to like about the decision to sell Modric?
The irony of this criticism is that, according to football-lineups.com, 56% of Spurs’ goals last season came from midfield; the highest portion of midfield goals in the division. Whilst you could still argue Modric under delivered in front of goal, more pressingly it was the team’s strikers who were disappointingly impotent.
The topic got me thinking – how important are goals from midfield?
Surely it doesn’t matter where the goals come from? Or does a steady stream from more withdrawn sources relax the strikers and power a team up the table?
One of the big issues is defining a midfielder. Lewis, above, classifies Cristiano Ronaldo as one, yet I think many would argue his role was closer to that of a striker.
Football is far more complex than these simplistic definitions, but football-lineups.com, as linked above, gives the classification a go. The following analysis is never going to be 100% accurate because of this, but hopefully it provides food for thought.
First, a simple relationship: since the start of the 2004/05 season, teams with high-scoring midfield departments have generally collected more points over the course of the season:
The problem with this chart is that there’s an obvious bias; better teams score more goals, and some of these goals must come from midfield. So do teams that score a larger percentage of their goals from midfield score more points? Not over the last seven seasons:
It’s difficult to establish what relationship you might expect. Goalscoring midfielders may reduce the pressure on a team’s strikers, who in turn score more goals, therefore reducing the portion of midfielders’ goals. Perhaps an inverted-U shape?
Either way, no relationship exists in recent Premier League seasons, and this is confirmed by modelling the effect of total midfield goals on points total, holding constant goals scored and conceded. That is, given two teams with identical goals for and against, the team that scored more goals from midfield does not necessarily collect more points over the course of a season.
However, there is a weak but significant relationship prior to the 2004/05 season, illustrated below:
In the five seasons from 1999/2000 to 2003/04, there is a significant (10% level) and positive linear relationship between percentage of goals from midfield and total points.
Using the model described earlier, it also appears midfield goals were more important during this period – a team with more goals from midfield collected, on average, more points than its competitors, given equal goals for and against.
This could be because of measurement errors (I’ve already noted the concerns with the data), as midfielders are increasingly becoming harder to define. The Premier League has also seen a shift away from universal usage of the 4-4-2, meaning there are attacking and wide midfielders often defined as forwards in the data. This perhaps makes the role of what one would consider a conventional midfielder less focused on scoring goals.
That brings us back to Modric – should he be scoring more goals? Posts on eplindex.com (here and here) highlight his efficiency with the ball, creating numerous chances and displaying fine ball retention for Spurs last season. It seems an unjust criticism when he’s clearly excelling as a creative force.
Even Manchester United’s summer target Wesley Sneijder has struggled for goals recently, with only three from open play in two seasons and none from open play in 2010.
But Modric shouldn’t be criticised for not compensating for the failure of others. Unsurprisingly, his record was not scrutinised when Spurs reached the Champions League in 2010/11.
As with all of my posts though, an open discussion is welcome – are goals from midfield important in getting an edge over rivals? Should Modric focus on improving his record in front of goal? Or does it simply depend on the personnel within each team?
5 Added Minutes 
Brilliant, and an interesting way to look at things. I particularly like the great point that the old school “English way” is being eroded away.
When Benitez and Mourinho came to the league, (in 2004/2005), the game changed. Wenger (up to then) used a standard 4-4-2 in most games (the invincible’s had a fairly fluid 4-4-2, but nothing like the variations of 4-5-1/4-3-3 we see now).
And also, the successful United sides we had seen in the domestic league, played a 4-4-2. That midfield of Giggs, Keane, Scholes, and Beckham was seen as an almost “perfect” combination.
But the influence of the European game (and European managers) has meant that other teams have had to adapt, and the more successful sides have done very well. Ferguson and his United side play a very fluid formation, and can easily switch things around with the players they have. Same at Arsenal, Chelsea, and now that Kenny is back, at Liverpool too.
Roles are not as defined as they used to be (certainly in midfield). And measuring a player by the number of goals they score/create isn’t the best way of identifying the best midfield player. Ball retention, defensive duties, their overall impact and influence on the game is key now.
I look forward to reading a lot more of your blogs, and keep it up.
Thanks. That was exactly my thinking re: Benitez/Mourinho, I found it really interesting that there was quite an obvious split pre- and post-2003/04. Fair to say they’ve also been golden years for Premier League football since, too. For more intricacies brought into the English game by foreign managers.
I think it matters for some ‘midfielders’ more than others, If Van der Vaart isn’t scoring, does he bring much else to the team. Modric is more of a Pirlo than a Fabregas or a Silva so the goals-expectation should not be as high. As you point out – players like that bring attributes that can’t necessarily be quantified on the scoresheet.
Exactly. Any criticism of Modric’s goalscoring record is, for me, a bit of a lazy criticism. Doesn’t benefit from the expectations he sets as an outstanding midfielder.
Reminds me a little of Barry Ferguson, particularly in the 09/10 season. I have heard criticism of him (well okay from one guy), that he didn’t score enough goals. But in that season he played more passes than any other player, was instrumental in keeping the ball, making the team around him look good, etc. Ferguson didn’t get any assists and the only goal he scored that season was in the FA Cup. But he was an important part of Birmingham’s 9th place finish that season.
http://www.football-lineups.com/footballer/1274/?t=378&s=185
Long winded way of saying “not all midfielders have to score from midfield” I suppose, and Modric is definelty one of those who does enough good stuff for Spurs without having to score.
Great stuff as ever Omar. Difficult subject to quantify but you’ve given it a really good (and interesting) go
Thanks 🙂
I am not sure I get this article. Shouldn’t the title be: “Does goals from midfield matter more?” As far as I can tell, your data seems to indicate that if there is such an effect at all, it is small.
Obviously though, goals from any position are a positive. Modric has 9 goals and 11 assists in 88 league games for Spurs. That’s a goal or assist in every four games. It’s very descent, but maybe a bit shy of what you expect from a player of his reputation. As far as I can remember, neither does Spurs do any better with him on the field. That might indicate he is overrated. There are also problems with chances created stats. The number of assists for every chance created, varies wildly between players. They have different styles of play. If your teammates are scoring on 3 of 63 you create, then you aren’t creating very big chances.
Simplistically, it’s whether a midfielder should be judged on his goal return. It’s also whether having a team with a significant number of midfield goals tends to do better. For example, if a team has the majority of goals coming from its defenders, you might expect it to struggle (lack of creativity/reliability on set pieces etc) – I’m trying to expand on that idea.
But yes, a goal is a goal in the grand scheme of things. Not sure I agree on chance created stats – it’s not the fault of the creator if the attacker cannot convert numerous chances.
It’s a very interesting idea looking at the goals from defenders compared to the rest of the team indeed. In the comparison of this article though, it will basically come down to whether it is preferable to score many goals from midfield or the forwards. That doesn’t seem to matter much. Relatively few goals are scored from defense, so that wouldn’t have much of an effect on the data. Another interesting thing, would be to look at the proportion of goals from set-pieces. I would think a high percentage of goals from set-pieces, predicted bad results, or at the very least few goals scored.
As far as chances created goes, it’s obvious the quality of the chances are not equal for every player. I think wingers generally create higher quality chances, for example. An example would be Theo Walcott, who had a high number of assists, even though he had few key passes. Fabregas would be another player with consistently good numbers in that department. With his 1 league assist and 1,9 chances created per game last season, Nasri was just terrible though. My impression is that the variation is massive. I would love for you to look at a larger data set though, and find out the correlation between the two. Preferably over several seasons, I guess. You could also compare the ratio of chances created by position.
Soccer by the Numbers (http://www.soccerbythenumbers.com/) has done a few posts on set pieces before so do check that out. And yes, it doesn’t matter much if goals come from midfield or strikers – but that shouldn’t detract from the contribution of certain midfielders such as Modric.
Absolutely, that’s the key flaw of the statistic, though Opta now collect data on ‘clear cut chances’. I’ve done a couple of posts in June/July that look at chance creation if you haven’t already checked them out. Problem with looking at assist/chance ratios (I think that’s what you’re talking about?) is it doesn’t account for who was on the receiving end of each chance. It’s an interesting point though, will have a think of where I can perhaps go with it.
That’s true, although I am skeptical of how big the effect is. Checked the stats for Arsenal. I found a correlation between assists and chances created of 0,47 for the top ten players on numbers of created chances per game.
I shallowly looked at the other top teams though, and the correlation there was obviously much higher. For the top 6 in City last season, the correlation was 0,76 for example. That seems to suggest that Nasri should do quite a bit better there with his assists numbers. Arsenal might actually be somewhat of an outlier, which might have blinded me. I guess it shouldn’t come as a surprise Arsenal is fucked up even in that respect. The machine known as Manchester United for example produced a correlation of 0,71 for its top 10. The overall correlation in my small sample of 26 players was 0,60. Statistics is good fun indeed. Have to be a bit less lazy though, and check larger samples.
That strikes me as much an observation of how clinical strikers are (i.e. how many chances they need to put one away) as the quality of chances created (which is unobservable in the data).
West Ham, for instance, created nearly as many chances as Man United last season (https://5addedminutes.files.wordpress.com/2011/06/plchancecreation.jpg) – but does that tells us if the quality of chances were low or their strikers were hopeless?
Good article, but I think it needs a qualifier – it depends on what the ambition of the team is. If you are aiming for the title itself, then goals from midfield can be very important. The winning team in every season for the last ten years has scored an average of around 80 goals per season (with only one season being won by a team scoring less than 70 goals – 2009). The average for a 4th place team over ten seasons is 65 (rounded up), with two seasons where the total goals for the 4th place team was less than 60. So if a team is looking to win the league, they have to find 70+ goals from somewhere. Assuming the popular 4-2-3-1, with the “3” being the attacking midfielders and the “2” being considered part of the defense, the attacking 4 (3-1) will be responsible for the majority of the goals. Modric’s league goal rate of 1 goal in every 10 games (approx.) for Spurs is not bad, but it means the other 3 players in the attack have to score around 66 goals between them – that’s 22 goals per player on average. So Modric’s contribution specifically is okay, but it could be better. Had he scored another 5 goals last season, (and let’s say those goals contributed winning scorelines instead of draws), that would have added an extra 10 points to Spurs final total – 10 points that would have put them in the Champion’s League and 3rd in the final table. Having said that, of course we can’t predict the distribution of those extra 5 goals, but it’s merely an example to illustrate a point – if your ambition is survival, then any amount of goals from midfield will be good. If your ambition is top 4, then a 1:10 ratio of goals to games will probably suffice as long as your strikers are good and your fellow midfielders contribute equal numbers of goals. However, if you want to win the league, you will need at least one midfielder with a very good goals to game ratio, and probably more than one (usually wingers as well as attacking midfielders). So goals from midfield DO matter, depending on your ambition as a club. The lower the ambition, the lower the importance.