I brought up a point 2 weeks ago that's worthy of further discussion. Back then, I argued that David Estrada plays a different game in midfield than as a forward, and that his performance as a mid was a good skill set to bring in against Chivas USA (I also hinted that Lamar Neagle plays better as a forward than on the wing, this year). The example, along with others I discuss below, makes it clear that on-field role substantially modifies player statistics. If we're going to use statistics as a means of player evaluation, then a change in role must be clearly recognized (and, ideally, normalized), much the same way as we need to characterize how performances may be impacted by easier (or harder) opposition.
Lately, we've been discussing how the Sounders can get their strongest lineup on the field, even to the point of playing key players out of position. I feel that is probably the wrong argument for us to engage in - demands of health and rotation make the best XI a best XV. I am wondering, instead, how to move the Sounders XV around the field to where they are most effective (along with the team). There is no better way to demonstrate the impact of on-field role than to pit a player against himself. We'll evaluate Lamar Neagle and Brad Evans below, and hopefully get some idea of how Seattle should arrange late-season tactics.
Neagle has clearly been most productive in 2013 as a forward - all 8 goals and 3 assists have come while playing in a nominally advanced position. Some of this has been luck, and some of it has been genuine differences in play. As a forward, he generates a shot (either his own or assisting someone else's) about 4.49 times per 90 minutes played, and 2.86 times as a winger. Breaking down his offensive and defensive rates...
The similarity in Neagle's play between wing and forward has, in my opinion, been the issue. He has been similarly aggressive in both roles, but out wide that has caused him to give up possession in more stressful areas for the defense, and low-percentage passes in the middle third are less likely to generate dangerous second chances on offense than giveaways around the box. Displaying lost possession:
That having been said, I haven't given up on Neagle as an effective winger in MLS. He has the speed and acceleration to separate from a defender, but not enough deception. He has the ability to send dangerous crosses into the box at significant range, and his lack of production from the flanks may be partially attributed to bad luck and poor finishing from his target. I believe that he needs to work on his technique in taking on defenders, and send in crosses more often than he has, instead of consistently cutting in towards the box.
Following a successful run with the United States men's national team, Evans' strongest position may be right back... but Yedlin's emergence has tied that up. There have been several examples this season of the team playing with only 1 defensively-minded midfielder, and playing quite poorly (Houston). I believe it's important to have Evans in the game (or, failing that, Moffat, and then failing that, Rose) against talented opposing midfields- but where?
Any judgment of passing is difficult without the kind of analysis I attempted in comparing Evans to Erik Friberg back in 2011.
Evans is a high-percentage passer at CM, but tends to favor back and lateral passes, usually after spending time on the ball (he is not the 1-touch passer the Sounders get in Shalrie Joseph). Some have pointed to recent assists suggesting his play at CM has become more creative this year, and that would be an interesting argument to test in the numbers. However, I would argue that playing him wide does not prevent the sort of circumstances that led to those assists, and indeed encourages them.
What impact does his play out wide have on defensive and offensive rate statistics?
Evans out wide averages 6 fewer passes and 5 fewer defensive actions per 90 minutes, compared to his play in the middle. Almost the entire difference in passing rate is attributable to one ridiculous game: the season opener against Montreal, where he racked up 80 non-throw passes in 90 minutes. Excluding that game leaves him at roughly 39 non-throw passes per 90 minutes in both positions, and 4.4 fewer defensive actions out wide. Adding together assists and key passes, his "shot assists" are nearly identical between the positions (1.21 wide per 90, 1.16 center). All 3 goals have come playing as a CM.
When I began this exercise, I honestly expected the numbers to be slightly more in favor of wide play on offense, and to see a lesser difference on defensive rates. However, before I conclude I've falsified my hypothesis (again), I'll note that Evans has in fact been two different players out on the flank. Both these players pass at more or less the same rate (in the range of 0.35-0.55 per minute), but one defends like EvansCM (0.12-0.14 actions per minute), while the other does not (0.5-0.8). Evans can either defend on the wing like David Estrada, or be slightly less involved in the defense than Mauro Rosales. Why?
Who were the CMs?
4/20: 1-0 W away, Alonso/Joseph
5/11: 4-0 W home, Alonso/Carrasco
5/18: 4-2 W home, Alonso/Carrasco, Alonso leaves for Joseph at 31'
8/25: 1-0 W home, Alonso/Joseph
5/8: 1-0 W away, Alonso/Carrasco
5/26: 0-4 L away, Alonso/Carrasco
7/3: 2-0 W home, Carrasco/Rose
7/6: 0-2 L away, Carrasco/Joseph
There are a couple of common trends here. Evans had more defensive involvement in wins, possibly defending leads. It's also noteworthy that Joseph played a substantial amount of the game in the middle on 3 out of 4 occasions where Evans took more defensive responsibility while playing wide. 3 out of 4 of the starts in which his defense slackened involved 2 defensively-minded CMs. Where does he play in such a role?
This argument becomes academic at a certain point. I'm trying to put together a Evans/Alonso/Dempsey/Rosales midfield. Dempsey and Evans will occupy similar spaces no matter their nominal positions - both players stray inside even when playing wide mid. However, I think it's important to stress that Evans should try not to occupy Dempsey's space in attacking midfield on offense, and also important to confirm that he is still able to meet defensive obligations lining up, primarily, on the outside.
What is missing in soccer statistics?
Mean defensive and offensive rates of Sounders midfielders. Error bars are 1 standard deviation.
How would one normalize individual numbers with respect to role? This is a daunting problem for any side-by-side comparison of players without exhaustively studying their behavior and overall team performance game-by-game. If the problem is going to ever be tractable - if soccer statisticians are ever to create an objective summary statistic of quality... to come up with a "WAR," as Jeremiah put it on Nos Audietes - then they will need some consistent means of quantifying opportunity and difficulty.
Arranging the Sounders
Given that Evans is better-used out wide, and that Neagle is significantly more effective at forward, my "ideal" XI becomes:
Dempsey plays behind the other forwards in the center, as an attacking midfielder. The formation is asymmetric due to his need to get forward into the box and Rosales' runs on offense, coupled to Evans providing defensive cover.
- If Dempsey is needed at forward or is unable to play, either Evans or Moffat or Joseph come in at CM (depending on availability and the nature of the opponent) and more offensive responsibility is put on the wings.
- Neagle is the first substitute or alternate for Martins, EJ, or Mauro
- If Rosales is unavailable and the Sounders see good matchups for set pieces or long crosses, Burch becomes a potentially useful role player.
Having named 4 subs playing significant minutes and bringing important skills to the team, my Ideal XI indeed becomes an Ideal XV.