(Trigger warning for non-soccer friends: this is all soccer.)
The end of the season MLS award voting is a bit of a struggle. With rather nebulously defined categories, voters (of which I am afforded a ballot through my media role) is forced to scratch their heads a lot. We try to figure out what level of adversity justifies the “Comeback Player Of The Year” award. We read through team-supplied biographies of various charitable efforts to work out who deserves “MLS WORKS Humanitarian Of The Year”.
And “Most Valuable Player”? It’s hopeless. Player “value” is subjective to the individual person, and so what makes one player valuable to one voter may not matter at all to another. And it’s been particularly difficult in 2014, with no clear frontrunner across any of the 19 MLS teams. Yet it’s serious business: one Twitter user declared me a “fucking idiot” for not having their favorite player in my short list.
So inspired by the Grand Ginger Of Major League Soccer (who advocates for having a formula for deciding your MVP, despite Twellman’s objections), I decided to create my own algorithm this year to help me figure out my vote.
First, I had to select a player pool. Defensive metrics are nearly impossible to come by, so I limited the list to forward and midfield players, trying to get at least one per team (with a few exceptions: sorry Colorado / Montreal / San Jose!). The twenty names that ended up on the spreadsheet were:
Quincy Amarikwa (Chicago); Will Bruin (Houston); Bradley Wright-Phillips (New York); Erick “Cubo” Torres (Chivas USA); Jermain Defoe (Toronto); Clint Dempsey (Seattle); Landon Donovan (L.A.); Dom Dwyer (Kansas City); Fabian Espindola (D.C.); Ethan Finlay (Columbus); Thierry Henry (New York); Robbie Keane (L.A.); Sebastian Le Toux (Philadelphia); Obafemi Martins (Seattle); Lee Nguyen (New England); Pedro Morales (Vancouver); Joao Plata (Salt Lake); Luis Silva (D.C.); Diego Valeri (Portland); Gyasi Zardes (L.A.)
Independently from the players, I had to select what stats mattered to me, and how much each counted.
I started with the most critical thing you can do as a player: win games for your team. So game winning goals were given a weight of 4 points each, and game winning assists would earn 3 points each.
Next, I wanted to reward offensive production in general, so non-penalty goals earned 2 points (even if they were the same as the game winner). Non-game-winning assists were worth 1 point.
Then I wanted to consider what actions an offensive player could take that would damage their team’s ability to win. I could only think of two that were easily measured: missed penalties and red cards. Red cards being the more serious of the two, I subtracted two points for each red and removed a single point for a missed penalty kick.
This gave me a raw score for each player. How did it look? Here was the top ten:
- Robbie Keane (79 points)
- Lee Nguyen (71 points)
- Bradley Wright-Phillips (68 points)
- Tie, Clint Dempsey and Obafemi Martins (61 points each)
- (see #4)
- Landon Donovan (59 points)
- Tie, Dom Dwyer and Gyasi Zardes (58 points each)
- (see #7)
- Thierry Henry (53 points)
- Diego Valeri (52 points)
For most pundits, this is a cut and dry confirmation of what many had been saying: Robbie Keane’s production was ridiculous, and he was an easy vote for MVP. (Grant Wahl offered his case for Keane winning MVP today, which partially lead to this post being written.)
I, unsurprisingly, am not most pundits.
I get very hung up on the word “valuable”, and after rolling it around in my head for a while, I couldn’t overlook one thing: the pay scale in MLS is quite notoriously out of whack. This is, after all, a league where a player on one side of the locker room may be making $6,000,000 as a base salary (like Toronto FC striker Jermain Defoe), while his teammate across the room may be on the league minimum of $36,500 (like Toronto FC midfielder Daniel Lovitz).
So I wanted to apply a “salary factor”, one that would adjust the player’s performance in light of their salary. This took a little while to figure out a reasonable system for, because with the range so great, it could very easily tilt the scale too far away from designated players.
The formula I came up with looks like this:
Factor = SQRT(SQRT([Player Base Salary] / [League Median Salary]))
The league median in 2014, per articles I dug up, was $80,000. Taking the fourth root stabilized the values into a range from 0.89 (for Ethan Finlay, making just $50,848) up to 2.94 (for Jermain Defoe). Dom Dwyer sits right at $80,000, so he was the only player to not have his raw score modified by the multiplier.
Is the factor uniformly fair? It’s debatable, but my general impression is yes. The Keanes and Defoes of the world should be better than the guys making a sliver of their salary. (Again, I’m obsessing over value, and I get that’s not for everyone.)
Here, then, are the rankings of all 20 players scaled by salary:
- Lee Nguyen (58.38 points)
- Dom Dwyer (58.00 points)
- Gyasi Zardes (51.88 points)
- Joao Plata (50.66 points)
- Bradley Wright-Phillips (47.71 points)
- Ethan Finlay (43.68 points)
- Luis Silva (37.88 points)
- Will Bruin (36.49 points)
- Fabian Espindola (35.89 points)
- Diego Valeri (32.89 points)
- Erick “Cubo” Torres (30.66 points)
- Robbie Keane (28.85 points)
- Obafemi Martins (28.76 points)
- Quincy Amarikwa (24.72 points)
- Sebastian Le Toux (24.07 points)
- Landon Donovan (21.85 points)
- Clint Dempsey (21.79 points)
- Thierry Henry (20.26 points)
- Pedro Morales (15.79 points)
- Jermain Defoe (12.91 points)
There’s a lot of interesting occurrences in here, particularly when it comes to LA’s attack. Donovan (4 GWG / 5 GWA) and Keane (5 GWG, 6 GWA) both produced big numbers, but when Gyasi Zardes is in the same ballpark (6 GWG / 0 GWA) for a fraction of the salary ($125,000 compared to Landon’s $4.25M and Robbie’s $4.5M), it’s hard to argue they’re not getting tremendous value out of their academy product.
But overall, the math confirmed what my gut had been feeling: that Lee Nguyen put up incredible numbers (no one was even close to his nine game winning goals) at what is almost a criminally low salary. (Never mind that he’s not an attacker and produced those numbers as a midfielder.)
So: my first choice vote (you get two choices on the ballot) for the 2014 MLS MVP for Lee Nguyen.
I have no doubt some will want to argue that my weights are wrong, that I’ve overlooked a key metric, or that I’m just dumb for using a spreadsheet to determine who to vote for. And that’s cool, but I’m not looking to argue – it was my vote, one of who knows how many in the media vote, which will only count for 33% of the total vote scoring.
For those that would prefer to argue about this, a request: write your own algorithm. The process is illuminating about what does and doesn’t matter to you when it comes to player performance. (And if you want to lobby people with the votes that count the most, aim for a club’s sporting director/coach/GM/communications director, as those 76 votes combined count as much as the entire media vote.)
As for my second choice? Well, I can’t follow a formula all the time.