Tuesday, March 5, 2013

A statistical look at MLS Week 1

Before we get started: If you haven't already, do yourself a favor and lope on over to Tempo Free Soccer and poke around. Alex Olshanksy is doing man's work with statistical analysis there, and if you want to be a smarter, better human, the choice is obvious.

In depth statistical analysis has a home in soccer in the same way it has a home in baseball. What I think will always be the case is stat-heads will occupy a shaded corner of the field, and the rest of us can venture in that direction when the need arises. Will that shade ever stretch over the entire pitch? No, I don't think so. But the thought that it might is what typically fuels the unnecessary backlash against it. Don't worry -- these dreaded numbers aren't robbing the game of its beauty. It's helping you define it in the interim periods between kicks. Put the knife down.

In that vein, I'd like to take a brief look at his first 2013 MLS Tempo Free Table of the season and extrapolate some truths from the first weekend of the season. First thing that needs to be established is the metric used for defining "possession" in this case, which is one of the most important determinants for dominant sides. To do this, Olshansky takes Tackled, Possession Lost (TPL) totals, subtracts them by Clearances (C) and then adds that total to Attempts on Goal (AOG). What you then have is a possession, thereby creating the brand new Possessions Per Game (PPG) category.

Without further ado, some things that stood out to me.

- Montreal's 1-0 win over Seattle was probably best known for Davy Arnaud's stunner (though I'll still take Valeri's all day every day), but know this: Montreal and Seattle came out of that game one-two in PPG. Montreal and Seattle each had 157 possessions, which works out to an incredibly high rate of industry over 90 minutes. Quick, attractive soccer, crisp movement and a very, very fast pace. What made that so remarkable is that it's the first game of the season. I think the adrenaline of the moment might have spurred this on, as might have the enormous crowd in Seattle. Indeed, the Sounders were much more composed. They sit in second in passes per possession (3.48) while the more flustered Impact are 14th with 2.45. What that tells you from Seattle's end is largely what the final score tells you -- Fredy Montero is already missed. Lots of passes in build-up but unpolished finishing on the final leg.

- After a deep probe in the playoffs last season, a lot's been heaped on the shoulders of Ben Olsen's growing DC United side. But, in a 2-0 loss to Houston on Saturday, DCU was statistically and aesthetically piss poor. They played a hopelessly direct style, finishing 18th out of the 18 teams to play over the weekend (NE was off) in passes per possession with 1.96. They're currently the only side in MLS with less than two, including meek performances from Chivas and Chicago. Their raw possession data is identical to Houston's (both had exactly 142.50 possessions) but their profligacy with the same amount of looks on the ball was shockingly all-consuming. Moreover, Houston's penalty area might've well been the Sahara. Through the opening weekend, DCU is currently dead last in attempts on goal (4.91), expected goals (0.26) and shots on goal (0.00). Not to mention DC is in the bottom three in each of the major defensive metrics -- defensive attempts on goal, defensive expected goals and shots on goal surrendered. It's a pitifully small sample size, and Houston is the two-time defending conference champs, but they'll need to shore up a lot if they hope to compete in the playoffs.

- The game of the weekend was probably a 3-3 draw between Portland and New York. Not only for its excitement and the long-awaited season-opening Portland tifo, but because it matched up a pair of rookie coaches, fresh meat for the analysis till. After a woeful first half, Caleb Porter's possession-based formula finally came to bear. Portland completed nearly four passes per possession against New York, tops in the league. It attempted more shots than anyone in the league and was behind just LA in shots on goal and the seasonally-adjusted expected goals. Mike Petke's style is a bit harder to nail down. New York was miserable defensively, (18th or 19th in all three metrics), and the only thing they can be said to have done even reasonably well was putting a decent amount of shots on target. Everything else was fairly nebulous. Both teams got a point, but I'd say Porter won in terms of establishing an early base from which to build on.

- If the clear across-the-board statistical winner from MLS' opening weekend were the Galaxy, the loser was the team on the other end of LA's 4-0 blowout win. The Chicago Fire were uniquely terrible. Not only are they in the bottom half of the league in literally every category, they rank first in luck, which is points per match subtracted by expected points per match. Godspeed, boys.

- Will Parchman

4 comments:

Tony M said...

Thanks for the tip on Tempo Free Soccer. That site is awesome.


Oh, and you did good here, too

dikranovich said...

i think the real takeaway from week one should be that if you dont get shots on target, you cant score goals, and if you dont score goals, it is hard to pick up three points.

dc had zero shots on target, chicago had one, and chivas had two.

interestingly kc had 4 out of 7 shots on target and got three goals. nice production.

Jay said...

The other thing to extrapolate from a single game's worth of data is that the data isn't going to hold up, and thus extrapolations at this point are... dumb. I can't be the only person to notice that LA's expected points per game are 3.88, or that Chicago ranks first in luck only because they have negative expected points. Some might say "WTF, dudes, them numbers are stupid bullshit," which, you know, isn't wrong, but it also represents a data set that isn't seeded with the previous season's numbers the way Ken Pomeroy's college b-ball rankings are. Let's see what this tells us in June.

Mikey B said...

Is there a stat that measures the likelihood that Chivas USA can actually break the 7,500 attendance mark? Or A stat measuring the chance of a league-enforced sale?

I keed, I keed.