The MVP Debate: Filtering Expected Goals vs. Game Winning Goals
/by Harrison Crow (@Harrison_Crow)
Last year, if you even knew we existed then, you may have heard us mentioning Lee Nguyen for his quantitative key pass numbers. He produced a lot of key passes (51 total, 14th in MLS) and put himself on the map for people like me to watch for this season.
Well, as you know, this season he's been more than just an above average central creator for a good team. He's been someone that has gained dark horse consideration for the MVP as the MLS playoffs begin.
A lot of the conversation surrounding Nguyen centers on the fact that he has most "Game Winning Goals" in MLS with 8. Meaning he's directly responsible for a total of 24 additional points for his team, the New England Revolution. Well... okay, sure. To some degree he Nguyen has been very much influential to adding those "additional" points.
My problem with attributing those points solely to Nguyen is that it takes away from the rest of his mates who also had their hand in those matches. Winning by a single goal often involves more than an outlier event. It's a transcendent performance that comes by more than a single event and a singular player but by a battle conducted by all 11 players against a fellow 11 players.
Sure, it's often a single players performance or instance that can turn the tide of a battle and help his team gain points. But if we all believe this is the case then we are only accounting for single goal states in our analysis? We should be taking into account both negative one AND even game states for goals.
Additionally, we all understand the propensity goals have to be flukey at times. We all have seen a team perform the best and be handed a draw or even a loss despite everything. Because of that it's best to look at shots and the leverage of which we could reasonably expect a goal to be scored in that situation. That's the whole premise behind expected goals.
Now, I've talked before about how I don't necessarily agree with the denotation of "game winning goal". I believe it conveys the wrong impression behind the intention and event. The stat is only awarded to players whose teams win by a single goal and given to player who scores that final goal that ends up being the difference. If a goal is scored in the fourth minute it's the same as scoring it in the 90th minute. Obviously that goal still was the determining factor in the match. But calling it a game winning goal implies a sense of heroic nature to the event.
Most of the time if I'm talking about a game winning events my mind goes to closing minutes or moments just before the games end. It's those closing seconds that we remember the most because it seems to have the most impact. Michael Jordan with the game winning shot, John Elway with the game winning drives and Roger Clemens big last moments.
If we want to strive to really account for these late match events and properly put them into context than what we should do is develop a filter for what events to take into account and when. This will give us a method of evaluating true difference makers at the end of matches. Looking at the things mentioned above, I've implied the need to account for three specific details: shots and their placement, the current score line of the event and time the event took place.
Using those three determining factors I put together a small list of total expected goal contributors with shots and their leverage occurring from the 80th min to the last. One specific item of note is that the data set includes penalty kicks because... well, I couldn't decide whether or not to include them.
1 Clint Dempsey 2.50 2 Sebastien Le Toux 2.26 3 Mike Magee 1.79 4 Fanendo Adi 1.71 5 Lee Nguyen 1.69
I could have given you all 300 individuals who had met the following condition but I scaled it down; way way down. Why did I limit this to five examples? Well, because this is about Lee Nguyen and since he rounded out the top-5 it made the perfect complimentary talking point. Also, I don't think I could have lived with myself had Juan Luis Anangono found his way into the top-10 (which he did).
Obviously, the narrative with Lee's MVP candidacy is that while he doesn't have the stats that compare quantitatively he has shouldered the loud when it counts to be that difference maker at the end of matches. Seeing him on this list does play a bit to that thought.
However, looking further down it you can see Obafemi Martins at 18 (1.33) followed even closer by Robbie Keane (1.59) and Bradley Wright-Phillips (1.49) at 8, 9 respectively. The question then becomes if that is enough to build a run towards the leagues most valuable player award?
This isn't suppose to be a biting article with the intention to take away from Nguyen's contributions. I'm a big fan of what he's done this season and the example he is for what talent teams can unearth if they are willing to invest some time and money in the domestic market.
This is about how you contextualize statistics and properly use them to convey meaning. If you were to give another top-10 ballot Nguyen would still not make it, and my awarding Wright-Phillips my vote earlier this month only seems to be justified at this point.