By Harrison Crow (@harrison_crow)
It’s no surprise that expected goals is finally being talked about in the fantasy sports realm. This is great and it’s really entertaining to me because, as you might expect, it’s where we at ASA often use it the most. It’s an incredibly useful tool that can provide some quick tools for judging players when needed.
Now, let’s talk about how we’re using it.
Expected goals is, as we have well documented over the years, a measure of the opportunities and chances created by a player and their team. Porting that to the fantasy soccer realm there are terms and conditions on this that we need to consider.
Expected goals isn't a one-stat-fits-all for all metrics. Rather it’s a sum of many parts. Looking over at NYCFC and the fact that they’re killing it with the highest expected goal differential is great! But—realizing how they’re doing is even more important as that speaks to the sustainability of their success.
Expected goals has a high correlative nature with predictive success but there are a few pitfalls that come with it. Recognizing those pitfalls and taking the next step is where you can have success in identify the sustainability in the quality of those chances.
A good example of this is breaking down why NYCFC has the expected goal differential that they have, whether or not that’s good. My friend Rob Lowe went on a little bit of a tweet rant on this subject last week and it’s a great example of this thought process.
New York City is fifth overall in shots taken inside the 18 yard box a game. BUT… they’re pumping balls into the box via crosses and are eighth in total key passes by crosses and second in completed crosses per game!
NYC has an attack that, despite being led by David Villa, has largely done it’s attacking WITHOUT him to a tune of 66% of their total xGF coming from other players. Historically only a few teams have shown full season success that way (the primary examples are 2012 SJ Earthquakes and 2015 Columbus Crew though there are couple others).
This isn't to say that NYC is bad at attacking. It is just a legitimate question mark on whether or not that attacking can continue to sustain scoring goals using this direct approach. Additionally, their defense over the last two years has been incredibly suspect but this season xGA ranks them first in MLS. Why, and what are they doing different?
First, their shots against are way down (11 per game opposed to 13 in 2016 and 15 a game in their inaugural season). Reducing volume is a huge step to becoming a better defensive team but that’s not all of it. They’re only allowing six percent of total shots from inside the six yard box, good for sixth in MLS, and they're sixth in forcing shots from outside locations (38%) rather than straight down the middle of the pitch.
Basically, they are limiting shots and pushing those shots into less dangerous positions. It’s a very good defense and viably sustainable through the season. Are they truly the best team just because xGD says so? No, probably not (but maybe!). But they’re easily in the top third and I’d even put them in the top three right now considering their performance in Dallas this week.
Because of all this Sean Johnson is a great fantasy keeper to have as he’s got a better probability to post a clean sheet just based upon his surrounding defense. Likewise, grabbing Villa just because his team is scoring goals isn't necessarily a great idea by itself as he’s hardly a focal point. I’d take CJ Sapong and his near 5.0 xG, which is 49% of his team's xG. But if you’d just rather call me crazy please feel free.
Another useful tool in fantasy but probably not for the reasons that you think is G-xG. Goals minus expected goals can tell you about the viability of an individual “finishing” their chances or… at least what we might interpret it as such.
We might look at it and see someone scoring more goals than their xG and think “they’re clinical” or “beating the system” due to an undiscovered finishing talent. However, as shown over the past years (part 1 here and part 2 here), goal scoring is streaky and most individuals will work their way back to the mean.
This is largely because most individuals take less than 100 shots in a season (there have only been 12 unique players in the last three season which have surpassed this plateau) which makes using this tool on a season to season basis rife with noise. As players will have the propensity for under or over performing in those short windows with such small data sets.
What it is useful is to identify some of the players who are perhaps underperforming and attempting to diagnose why they are under performing. One such case in point would be Cristian Coleman who has -1.08 goals G-xG this season. The book on him because of him coming up empty has been he’s terrible at finishing goals - he’s had all of 11 shot attempts in less than 300 minutes (or less than four games). But with such a tiny sample size it’s so hard and difficult to really pin down if he’s a “failure” or not. But if we were to open this up to CCL he’s scored two goals on eight shot attempts during 289 minutes on the pitch.
If we combine the two he has two goals on 19 shots in 580 minutes. That’s not great but it’s not terrible either, and his xG p96 comes in at 0.37 just a smidge below Maxi Urruti’s at 0.42. I’ll say I think he’s been underwhelming for what we expected out of him but I I’ll give him at least 1,000 minutes to prove to me he can’t consistent create chances.
The bigger (and probably more subtle) issue however is the fact that Urruti’s doing the majority of his own work creating over 40% of his shots unassisted while Colman has been starved for assistance with only 20% of his shots coming on his own.
Another important aspect to these numbers is how penalties is applied. Obviously, in fantasy, it’s incredibly important to know who is taking team PK’s but with that it’s important to consider how many they've taken and the fact that this has likely either boosted and/or stunted their expected goals relative to their peers.
We here at ASA control for the lack of predictability in penalties for teams by using the occurrence or probability of the event occurring again rather than the conversion ratio for the league. This doesn't happen for players which in turn should only further emphasize the “buyer beware” stamp on the predictive or repeatability of expected goals in the short/near term.
One specific example is Erick Torres and his commanding lead in Comeback Player of the Year and the slew of feel-good pieces concerning his resurgence. Not to rain on the parade but his goal scoring has mostly come by way of set pieces (two goals) and penalty kick goals (four!) accounting for a two thirds of his league leading eight goals. This is A) not sustainable and B) absolutely a reason to avoid him in the short term in fantasy.
Over the last few years Cubo needs a play maker and while he’s got some help with amazing wingers there is no one player behind him in the midfield that is especially adept in helping him create chances (and before you shout Alex through your computer screen’s at me, he’s only responsible for setting-up two of Cubo’s shots for grand total of .23 xA).
Despite Houston taking advantage of things that make him particularly useful and still helping them score goals this isn’t going to likely pay out in the long run and I would absolutely bet against him scoring 15 goals at this point, despite being half way there, unless Houston finds a player that can find ways of putting him in better scoring position.
Using expected goals is a great new tool a lot of folks are discovering to be helpful in selecting players for their fantasy teams. Likewise as handy of tool as it is, it’s equally as important to understand how to apply these numbers and identifying when, where and what approach we need to have with them as they can be dangerous and misleading as they are useful and insightful.