Alex Egol – RSAC Writer
Leicester City in 2015-16. Man City’s improbable, epic road to the title in 2012. The English Premier League has had its fair share of miracle stories. After Watford FC began the 2018-19 EPL season with four consecutive wins, including a 2-1 victory against star-studded Tottenham, there was reason to believe that perhaps another Cinderella would go to the Premier League dance.
In the weeks since, though, Watford’s dream start has begun to look more and more like a fluke. Analyzing player and team statistics for Watford FC will soon make it clear why Watford was never a top team in the Premier League.
As irrelevant as Watford FC may seem, there is a reason why I chose to write an article about them. It has to do with soccer analytics as a whole.
Soccer analytics is a developing and less-utilized field of analytics. One of its common statistics is expected goals. To learn more, check out the Measurables podcast’s episode https://player.fm/series/measurables/expected-goals-and-expected-assists on Expected Goals and Expected Assists. Different Expected Goals metrics measure the number of goals a team is expected to score based on the quality and quantity of the chances they produce. The difference between expected goals and goals scored in reality can be useful in explaining how legit a team is.
After listening to Brendan Kent’s podcast on Measurables, I was perusing through https://understat.com/league/EPL and saw Watford’s expected goals (from now on we will use “xG”) and match results. To me, Watford seemed like the perfect example of one of many difficult problems in soccer analytics: prediction.
Imagine it is September 2nd, 2018. By now, Watford has beaten Brighton, Burnley, Crystal Palace and Tottenham. You want to predict whether Watford is going to continue to play at this level, or crash and burn (maybe you are betting on the outcome of Watford’s season). You open up https://understat.com/team/Watford/2018 and look at the data below.
https://understat.com/team/Watford/2018 ; Watford 2018/19 Match Results.
** xG is under each score.
The xG data pretty clearly indicates that Watford did not deserve to win all four games. If the first two matches were simulated thousands of times, with the same chances that actually happened, Watford would have scored an average of 3.34 goals. They scored five goals. In the last two matches, Watford’s opponent was projected more xG than Watford, but Watford won both. Watford got lucky.
This data holds a lot of predictive power. If you know that Watford’s real results are inconsistent with their expected results, you would anticipate that Watford would eventually begin playing as well as the data says they should. This is why you would bet against Watford.
Sure enough, Watford have lost 5 out of their last 9 matches, and currently place in ninth overall. I expect that Watford will go even further down the table because most Watford players are scoring more goals than xG, and, for nearly every kind of chance, Watford scores more goals than xG, and concedes fewer than expected. I will unpack this.
Watford has 22 players on its roster, sixteen of which are contributors to the club’s goal scoring. Out of these 16 players, twelve have exceeded their xG count. Expect these Watford players to score less as the year goes on. Now, we will look at Watford at the larger team level.
On open play, corner, set piece, and direct free kick chances, Watford has exceeded their xG count. When defending against corner, set piece, direct free kick, and penalty chances, Watford have given up fewer goals than the xG of their opponents would predict. Expect Watford to score fewer goals and allow more goals as the season goes on. On both sides of the ball so far this season, Watford FC has gotten lucky. Don’t expect their luck to last.