What A Weekend? Is The Unusual Really That Unusual?

The weekend just gone was like none before in Premier League
PINC
history!

At least, that’s what a short memory will tell you.

Yes, it was extraordinary. The current EPL champions Liverpool were humbled 7-2 by Aston Villa, who escaped relegation by the skin of their teeth on the final day of the last season, while Manchester United capitulated to a 6-1 home defeat by Tottenham Hotspur, who are coached by United’s former coach, Jose Mourinho.

Just how extraordinary was it, though? Well, to answer this, we need to measure extraordinariness. One way to do this is to look at what was expected beforehand.

For a number of years now, I’ve been producing forecasts before each game week in English football. Scoreline predictions but also match outcome forecasts. We could use these. Our model gave Aston Villa a 17.6% chance of beating Liverpool, let alone scoring seven goals (for that we gave them a 0.3% chance, 7-2 being a 0.07% chance).

So one measure of unlikely-ness is to look at the difference between what we thought would happen (Villa to win = 0.176) and what happened (Villa to win = 1), and get a measure of 0.824 for shock value. We could then average these over a game week, and ask just how unusual was this weekend? The answer is actually not that unlikely. Of the last 67 game weeks for which we’ve produced forecasts, this one was the 40th most unusual.

Arsenal beat Sheffield United, Southampton beat West Brom, Wolves beat Fulham, Chelsea beat Palace, Everton beat Brighton and Newcastle beat Burnley, after all.

In fact, by this metric, the most unusual Premier League weekend in the last three years was actually the weekend before last.

Another metric would be not to do any forecasting, but instead rely solely on historical patterns. How often have there been 7-2 scorelines in recorded football history? The answer is, according to all match results reported on Soccerbase, 420 occasions since 1872, in over 600,000 matches.

The volume of goals though is remarkable, right? Since 1992, goals in the Premier League have solidly averaged about 2.6 or 2.7 per match, and this season so far, the average is 3.8.

Again, not necessarily. Historically, September and October have been more goal-crazy, especially relative to January, February and March. On average, controlling for years and leagues, using all match results reported on Soccerbase, there are statistically significantly more goals per game in September, October and November – almost 0.1 of a goal. Below is a plot of the seasonal pattern:

September, October and November are the months with more goals. The seasonal pattern tends to be that teams are more open at the start of the season, before things begin to tighten up in the middle of the season, perhaps as teams learn how to play against each other, before at the end of the season things open up again, perhaps once some teams no longer have anything to play for.

And there’s only been 38 matches of the season gone so far. Exactly 10% of the season. We could look at all runs of 38 matches historically in the Premier League, and find out how unlikely this run of goals is. That’s plotted below since 1888, so the first observation is the goals per game in the first 38 matches of the 1888-1889 season of the English Football League:

So these most recent 38 matches are unusual since the end of the 1960s – a 60-year period. So, pretty unusual then. Will this carry on though given the trends for Septembers and Octobers?

Also, is this phenomenon more broad than just the Premier League? In the Table below, average goals per game so far in the current season (2020-2021) are presented alongside average goals per game last season (2019-2020). The Premier League is much higher, but the same isn’t really true anywhere else either in England, or across the elite leagues in Europe.

It looks like it’s just a Premier League thing. It looks like a once in sixty year occurrence. It’s not obvious it’s a closed-door thing, either, since research I’m involved in suggests no real effect on goals scored. Could it be a Covid-19 thing? Two of Liverpool’s star players are out due to Covid-19 – could others in the team be suffering?

Whichever way, it’s probably not worth banking too much on this goal-fest continuing…