On Wednesday, Opta, the official data supplier for MLS, reported that Earthquakes goalkeeper Daniel Vega had the highest goals prevented in the month of April via their @OptaJack Twitter account:
But what is “Goals Prevented” exactly? Effectively, Goals Prevented is a post-shot Expected Goals (xG) metric calculation that applies to goalkeepers. While American Soccer Analysis (ASA) has had a goalkeeper post-shot xG metric for several years which extends back to 2011 for MLS, the Goals Prevented metric was introduced by Opta this past January along with their post-shot xGOT (Expected Goals On-Target) shooter metric. While there has been great confidence in the xG metric from a shooter and team perspective (as it has proven over time to be more reliable predicting future performance than goals and goals against), even making regular appearances on MLS television programming, there has been less confidence in the post-shot versions of the metric as seen by the lack of support for them from the stats companies. More recently, however, StatsBomb threw their hat in the ring, indicating that the metric is now getting a foothold in soccer analytics.
While xG looks at every shot, including blocked shots, misses and shots off the post, post-shot xG (Opta’s xGOT) looks only at shots on target (aka, shots on goal). Opta and other statistical companies use xG to measure the percent chance a shot will be a goal based on thousands of previous shots taken from the same location. Generally speaking, the closer to the goal the shooter is, the higher the xG given to their shot. For example, here is the Opta xG chart for all the shots taken in the San Jose vs. Portland game back on April 6th. While we can see the overall xG number for each team and also for each shot, we unfortunately don’t have the details on how these companies, such as Opta and StatsBomb, do their xG calculation. This means we are left to our own devices to try to understand the data behind Vega’s April Goals Prevented number. Fortunately, ASA shares their methodology and using their generously-contributed data we can look at to understand this post-shot xG value for goalkeepers.
First, we can see that public data from ASA agrees with Opta that Vega had the highest post-shot xG prevented value of any MLS goalkeeper for the month of April. While Opta has the number at 5.71 goals prevented, ASA has it as a more conservative 4.55.
On the ASA website, this metric is called G-xG which is shorthand for goals minus expected goals. The greater the (negative) difference between the goals which should have been scored, as reflected by xG, and the goals which were actually scored, the better the keeper is considered to have performed (i.e., over-performed). If the value is a positive, the worse the keeper is considered to have performed (i.e., under-performed). For comparison, Andrew Tarbell had an -0.34 G-xG, 27 SOT against and 10 goals against last April – effectively over-performing by less than a goal according to ASA.
Given there were five games in April – how did Vega over-perform by around a goal a game?
As seen in the graphic above, one reason is that the Quakes stopped giving as many chances directly in front of the goal and within the six-yard box. Several of those March goals came from simple low crosses from the baseline across the face of goal.
In March, San Jose gave up nine goals in the run-of-play (noted here as “Regular”), however in April they only conceded twice. Although the number of on-target run-of-play shots increased from 19 (four games) to 21 (five games), Vega made 19 saves from these shots. This doesn’t tell us too much about the difficulty of the shot, however we can look at the shot xG against (xGA) for this information.
The average on-target shot for April by San Jose opponents was 0.33 (a 33% chance of a goal being scored according to xG), as compared to 0.47 in March. This is one indicator that the defense has been improving despite surrendering approximately six shots on target per game in both March and April. The quality of opponent shots was not as high in April as it was in March. In April, opponents scored on only 17% of the shots on target, which is effectively half of the 33% expected. Not only did the defense improve a bit, but the goalkeeping significantly improved earning Vega Opta’s honors for most Goals Prevented for the month of April. This catapulted him to third in the league overall at -3.21 according to ASA’s G-xG, behind perennial top performers Stefan Frei and soon-to-be-headed-to-Manchester City Zach Steffen – excellent company to be in.
Goals Prevented using xGOT or G-xG gives us a good, if imperfect, picture about the expected shot blocking performance of a goalkeeper. The more data available, the more complete the picture will be. Only nine games into his first MLS season, it’s difficult to say if Vega has turned a corner or just has the hot hands right now. However, with Jon Busch having the best mark of any San Jose goalkeeper this decade with -6.35 G-xG in 2011, Vega is currently on pace to surpass him. Only time will tell if Vega in his first MLS season at the age of 34 will end up with an all-time great goalkeeping season for the San Jose Earthquakes.