StatsBombIQ StatsBomb Data
January 15, 2014

What Does a Great Defensive Midfielder Look Like?

By Ted Knutson

Liverpool is the next team on tap for a transfer shopping piece, but before we get to that, we need to do some more baseline work.

With regard to Liverpool specifically, they need

a)      A creative defensive midfielder.

b)      A good left back and maaaaybe a good right back too.

For today, I’m going to ignore the second need with a promise that I will come back to it. The problem here is…

There are many, many midfielders in Europe. They play all sorts of different roles, and the key performance indicators are much more diverse and complex than what we would pick for forwards or attacking midfield players. Hell, even the ones who play defensive midfield vary in role from destroyer to regista to box-to-box. Sooo…

What the hell does a good creative defensive midfielder or regista look like?

pirlo_fashion

No no, not like that.

I mean what do they look like statistically?

*crickets chirping*

Okay then, that’s what we’ll figure out today.

Defensive Midfield KPIs
What stats do we care about for defensive midfielders? Tackles and Interceptions are a given (the defensive part of “defensive mid”), as is passing accuracy. Giving away the ball from a deeper position is bad, so high passing accuracy is a must.

Beyond that, things that might be important are offensive passing contributions (key passes, throughballs, but probably not dribbling), and then defensive context factors like how often a player is fouling, or being dribbled past. I haven’t done any detailed analysis to figure out exactly which of these contextual stats matter, but I will glance over them when looking at player stats to see if anything stands out as impressive or strange.

One stat that does matter for this role, perhaps now more than ever, is the ability to hit accurate longballs. Part of this is a tactical response to teams that press – you need a player in a deep-lying role who can hit your forward men with passes and start counterattacks. Another element is simply the need to be able to make longer, accurate passes due to field position.

Andrea Pirlo
Pirlo is the modern game’s prototypical regista. In fact, I find his stats somewhat difficult to wrap my head around. He does just about every single thing incredibly well. Except maybe run.

pirlo_stats

Tackling and Interception stats show he is moderately involved in breaking up the play, but his real value shows itself in the passing stats. He’s averaged 11 accurate long balls a game for as far back as I can see, while adding 2.5 Key Passes per90, .85 throughballs a game (down a bit this year, but Juve are even better than years past. In general, .85 will be near the top of all players in Europe), and completes passes at about an 87% clip. He even had a 13 assist season at age 32. The guy simply does everything incredibly well.

I ran a very rough similarity score over the data I have access to, and depending on what filters you use, seven or eight unique names show up as players who have come close to replicating Pirlo’s production in a season since 09-10.

The only name to appear on the list for three seasons is… Steven Gerrard.

Right. So Liverpool are basically replacing their own Andrea Pirlo (who is himself approaching 34).

No problem!

Another was Zonal Marking favourite, the incredibly underrated David Pizarro. He is unfortunately  too old for consideration, as is the retired Madrid legend Guti. Current Arsenal player Mikel Arteta also throws up a flag as a potential candidate, which is both surprising and fairly impressive. Finally, you get Luka Modric’s last season at Spurs before he moved to Madrid and

A couple of guys we will talk about as potential transfer targets for Liverpool tomorrow. Patience!

Sergio Busquets
If Pirlo is the prototype for the modern regista, than Busquets is the mould for the cultured midfield destroyer. He is the engine that made those amazing Barcelona teams function under Guardiola. His primary skills are breaking up attacks, recycling the ball to Barcelona’s attackers, and almost never making mistakes.

This is what Busquets’ destroyer production has looked like over the last 3.5 seasons.

Busquets_stats

Tackles + Int averaging around 5.5 per90, 4.5 completed longballs per game, and a passing accuracy that… well, that is probably fueled by the crazy Barcelona skew. Normal people on normal teams don’t complete 90% of their passes or more. At least I think they don’t. Let’s find out what the data says.

Not wanting to break the model concept, I plugged in the main Busquets KPIs of tackling, long balls, passing accuracy and the occasional key pass and found that there are many more Busquets type players in the data than there are Pirlos. Useful information, that.

Names who show up include Yaya Toure before he left Barcelona, Cambiasso at Inter, Michael Carrick and Paul Scholes, Mikel Arteta (again?), Thiago Alcantara, and…

Joe Allen?

*checks data furiously*

Not just once, but twice? First at Swansea and then at Liverpool?

whoa

Tomorrow: I’ll run through a transfer shopping piece for players that Liverpool should be interested in, that play mostly in the Pirlo role.

Apparently Liverpool already have their Busquets.

Stats Appendix
P90 – Total minutes played in a season divided by 90.
NPG – Non-Penalty Goals
NPG90 – Non-Penalty Goals Per90
ShAcc – Shooting Accuracy (multiply by 100 to get the percent)
Sh90 – Shots per 90
SOT90 – Shots on Target per 90
GConv – Goal conversion rate. (Multiply by 100 to get the percent)

A90 – Assists per 90
GA90 – Non-Penalty Goals + Assists per 90. A better measure of overall scoring contribution.
Drib90 – Successful dribbles per90
KP90 – Key Pass per 90
TB90 – Throughballs per 90
Pass% – Passing percentage.

Tack90 – Tackles per 90
Int90 – Interceptions per 90

 

 

 

Article by Ted Knutson