Monday, April 21, 2014

Rise of the MLS Diamond

The always excellent "March to the Match" podcast's most recent show focused on the rising use of the "diamond" midfield in 2014.  The episode (and MttM in general) are must-listens for MLS fans.  To the uninitiated, the "diamond" refers to the shape of the midfield.  Real Salt Lake runs a classic version of this formation, also called a 4-1-2-1-2.

Curious, I decided to look at what formations are the most popular in MLS and if there had indeed been a shift to the 4-1-2-1-2 diamond formation.  One thing to note is the data used (courtesy of Opta) refers only to how teams start a game and does not account for any in-game tactical changes.

It is still very early in the season, but there has indeed been a noticeable shift towards the 4-1-2-1-2.  This trend will certainly be worth watching as the season goes along.

Monday, April 14, 2014

The Portland Timbers' (Drawing) Problem + Alternate Table Realities

In 2013, Caleb Porter led a remarkable turnaround for the Portland Timbers, garnering coach of the year honors in the process.  The team only lost five times in 34 regular season games (now seven in 40 counting 2014).  Despite this enormous success, there is some vague discontent in the Rose City.  The reason?  Of the 33 games the Timbers have gotten a result from, the team has only 14 wins compared to 19 draws.  This is highly unusual.  Even if Portland drew an above average number of times (let's say 35%), you would expect the team to have 19 wins and only 14 draws from those 33 results.

A reason for optimism? In 2013 Portland did have the best goal differential in the league, and this has been proven to be among the better predictors for year over year performance (at least in the EPL, so not quite apples to apples).  Also, there are no indications that a high draw rate is repeatable from season to season.

All indications are that the abnormally high draw rate being experienced in Portland is not sustainable and will regress toward the league's average draw rate in due course.  Caleb Porter is still an excellent coach and team seems very likely to be in the hunt for the Western Conference crown.  

Alternate Table Realities

There was some discussion on twitter about MLS' high draw rate (38%) in the first month of the 2014 season.  Ravi Ramineni, of the Sounders, made an interesting comment.

I went ahead and looked at this alternate NHL-scoring style reality (forgetting the NHL overtime/shootout rules) for both MLS and the EPL.  As expected, the results didn't really move the needle all that much, though it would have delivered the 2013 Supporters Shield to Portland instead of New York.

Wednesday, April 9, 2014

Let's Talk About Wil Trapp

One of the feel-good early story lines of the 2014 season is the dynamic play of the Columbus Crew.  Playing in an outdated stadium on the outskirts of one of the smallest markets in MLS, first year coach Gregg Berhalter appears to have something special brewing.  Although winning three of their first four games is great, what is really intriguing about this Crew team is how they are achieving their results: possession-based (highest % in MLS), attractive soccer. The parallels with Caleb Porter's 2013 Portland Timber squad are so obvious that pundits are already trying to coin their style, only slightly in jest, as "Greggball" or "Berhalterball".

Apart from the coach, Columbus' management deserves a lot of credit for putting together a strong squad.  The addition of US international Michael Parkhurst is a big boost to the backline's ability to play out of trouble.  Then, there is the ever-sterling performance by the lesser-known Higuain brother, Federico.  In my opinion, Higuain is the most underrated player in MLS and it is a joke he is not perennially in MVP/All Star discussions.  See here for more evidence.

Higuain's favorite passing partner in midfield is 21 year old Wil Trapp.  The Crew's homegrown former US U-20 star has been at the heart of Columbus' 2014 transformation.  Despite his fast-rising star, Trapp does not get much attention from the national soccer press, as exemplified by his exclusion from the 2013 MLS 24 under 24 list. Yes, he plays for a small market team and it is true he is small and not physically imposing (listed at 5'8 at looks as if the wind might blow him over).  But he possesses a level of maturity and technical skill not often seen in the young American player pool (see this US Soccer spotlight).  

Individual soccer stats can be very misleading, but sometimes you see some numbers that make you question former-held assumptions.  

How important is the number of passes you complete in a game?  Or how many tackles you win?  Or how many long balls you play?  In truth, none of these statistics on their own are very useful.  But, sometimes you get enough data points that you begin to ask some questions. All players reach the point where vague assertions like the one I made above ("possesses a level of maturity and technical skill not often seen...") must be confirmed with output.  In the first weeks of the MLS season Wil Trapp is blowing right through that barrier with such a force that we just cannot ignore him anymore.  

Thursday, April 3, 2014

Stats: Michael Bradley is Insane

When Michael Bradley decided to move to Toronto FC this winter, there were a number of skeptics.  Why would a player on top of their game leave a European power (Roma) for a MLS doormat?  Before the season we took a look at what Toronto could expect from their new midfield maestro, trying to downplay his attacking qualities.  Now, just three weeks into his MLS career, it is clear that whatever expectations we may have had for Bradley are being blown out of the water.  Michael Bradley is already re-defining what an elite MLS midfielder looks like.

Even though he sits behind Toronto's attack, for the most part, Bradley has proven to be one of the most valuable shot creators in MLS in the first three weeks, accounting for a remarkable 48% of Toronto FC's shots. (SH/90=Shots per 90     KP/90=Key Passes per 90)

Although Toronto has really struggled with possession in their first three games (they pass a league low 1.5 times per possession),  Bradley has been the focal point of almost every sustained spell of possession.  His indispensability is best exemplified by his outrageously high (best in MLS) pass usage rate, which is simply the number of passes a player attempts divided by their team's total.  Bradley's rate of 23.5% implies that he is either passing or receiving a pass on an absurd ~47% of Toronto FC passes.

All of these stats are great, but only when taken in context with the arc of his career can one truly appreciate that the level Bradley is playing at is truly exceptional.  In particular, it is truly rare that a player lead their team in shots contributed (key passes + shots) AND in defensive actions (tackles + interceptions).

He may not always play like the best player in the world, and his ridiculous early season form might digress, but it is self-evident that Michael Bradley is making a strong case as the best midfielder in MLS and, perhaps, its best player.

MLS Tempo Free Table (2014)

Tuesday, April 1, 2014

Want to be a Professional Soccer Coach? Apply to MLS

A recent twitter discussion sparked an interesting discovery:  MLS coaches do not have very much professional head coaching experience.  Like, at all.  In fact, as of this writing a full 74% (14/19) of MLS head coaches are at their very first professional head coaching job (I include lower division head professional experience but not head college experience).  Quite honestly, this is an astoundingly high number for any professional league.  In fact, of the five coaches who held previous professional positions, only the Columbus Crew's first year coach Gregg Berhalter got his professional start outside of MLS (Hammarby).

It is important to understand that MLS is a very unique league in the world of professional soccer, and many foreign coaches (Gullit, etc.) have struggled with concepts like the draft, allocation money, and salary budget constraints, in addition to the cultural nuances of the North American sporting culture.  Therefore, it is not necessarily a bad thing from a team's perspective to hire someone who understands MLS' unique framework.  However, I think it is fair to say that eventually the league will need to adapt in its coaching hiring practices if it wants to be one of the leading leagues in the world.  Obviously, much of this adaption will be down to how much money teams are willing to spend on veteran coaches, which is something the majority of teams are neither required nor willing to do.  Nevertheless, the comparison between MLS and other leagues is quite jarring.

Wednesday, March 12, 2014

Splitting EPL Possession % into Offense/Defense

This article originally appeared in Stats Bomb

Does Possession % Matter?

Any analytically inclined soccer fan (aka you) is probably well-aware of the limits of possession % as a meaningful metric.  In fact, its faults are so numerous and well documented that many people bring it up solely for ironic purposes.  I understand the collective derision, but if we look at the metric in a deeper way can we glean some interesting information?  I think so.

One thing that I think does need to be stated is that, despite its limits, there is a relationship between possession % and points (at least in the EPL – see graph below).

The causes of this relationship are complex and difficult to disentangle, but probably the best way to think of possession % is as a symptom of playing winning football as opposed to the cause.  Devin Pleuler has an interesting take on possession as a defensive weapon.

How is Possession % Calculated?
Based on some good work a couple years back by Graham Macaree, we know that the possession % that the majority of media outlets use is really just a pass ratio.  The pass ratio approach is pretty simple: team possession % = team’s total passes / both teams’ total passes. This methodology was confirmed to me by an Opta employee.  We can debate the merits of this approach until we are blue in the face, but for many sensible reasons I think it is probably the best proxy.

Splitting Possession % into Offense/Defense
Despite these pass ratios being converted into possession %, not all pass ratios are created equal.  For example, let us assume that an average EPL match sees 900 passes on average between the two teams (450 for each team).  On this match day Arsenal outpasses Swansea 600-400 (60%/40%).  Across town, West Ham outpasses Crystal Palace 300-200 (60%-40%).  Both Arsenal and West Ham have the same possession %, but they have achieved them in vastly different ways.  By comparing their passing #’s to the league average, we can essentially allocate Arsenal and West Ham’s 20% possession advantage (60%-40%) to an offensive and defensive component, as demonstrated below.  You start by comparing how many passes each team attempted and allowed and compare them to the league average.  Arsenal, in this example, was 150 passes above an average offense (600-450).  West Ham, by contrast, was 150 passes below an average offense (300-450).  But, West Ham allowed 250 less passes than an average defense (450-200).  

That was a hypothetical, but what does this approach look like for this year’s EPL? (stats are a week or two old)

Talk about a tale of haves/have nots.   The difference between the #1 possession team (Swansea) and the #10 team (Chelsea) is closer than the difference between Chelsea and the #11 team (Newcastle)!  Another thing that jumps out is the comparison between Southampton and Arsenal; both have similar possession #’s, but achieve it in a very different fashion: Arsenal with offense and Southampton with defense.  You also might notice is the larger variance in the offensive component compared to the defensive component.  This makes sense, as a team might face a variety of passing styles over the course of the year, but their offensive style is more persistent.  Running some regressions (based on past five years of EPL data – 100 teams) backs this up, as the offensive component has a much stronger correlation with total possession differential than the defensive correlation.  Interestingly, while you would expect a relationship between the offensive and defensive components, the R2 was only 0.49, which demonstrates that this exercise of decoupling possession into offense/defense has some merit.