Scorecasting Economists

Echoes – what happens when football is played behind closed doors?

Football is returning! People are divided in the UK about whether that’s a good thing or not, but the Bundesliga in Germany will start again on Saturday (we’ll make forecasts tomorrow and here’s our end-of-season forecasts).

But it’s going to return without spectators – or “behind closed doors”, in the football vernacular. Will that affect anything at all?

Regardless of what we might think, we Scorecasting Economists like to look at the numbers to answer questions like this. They’re interesting from a sporting perspective, but also more broadly. Economists care about outcomes, and care about distortions in outcomes away from what we might think of as their natural level. They care about undue influences on outcomes, discrimination in decisions, and other external events that might affect how things turn out.

Football yields great opportunities to investigate things like this. Why do home teams win really often? Economists have proposed that it’s the referee that are actually the mechanism through which fans help the home team to win. So if fans aren’t present, they can’t influence outcomes via the referee. We’ve started to investigate this.

How many matches have taken place behind closed doors historically? It turns out, a small number – about 0.1% of all matches played across England, Italy, Spain, France and Italy, as well as the Europa League and Champions League. That’s 192 matches 131,229. It’s not many, but we can reduce that sample down a little, since there’s no point looking yet at Germany or England, since there have been no closed door matches ever in these two country (until now).

We focus on the matches in the plot below:

What differs between them? In the following two charts, we look simply at the difference in mean outcome between “normal” matches, and “closed-door” matches:

Home teams score fewer goals, and win fewer matches. Referees award away teams almost half a yellow card less. More shots on target are saved. Less injury time is awarded, though this effect is insignificant.

Hence this points towards home advantage being removed by playing without spectators. And it suggests the mechanism may be referee related, since the strongest effect by some distance is the yellow card effect, which is robust to all sorts of controls.

Many of the effects are insignificant. This could be because they are insignificant. But it could be because we have a small sample of closed-door matches. Hence while the general public is highly ambivalent about football’s return, as scholars we are delighted that we will soon have a lot more closed door matches to use in this work. We’ll be updating in the coming weeks…

How will Eredivisie end?

The final in our current series of looks at how European football seasons will end (England, Italy, Spain, Germany and France) is the Dutch top league, the Eredivisie.

As things stand, Ajax and AZ Alkmaar are separated only by goal difference (45 to 37), tied on 56 points. The tightest of the leagues we’ve looked.

Despite that, our projections give Ajax a 53.5% chance of winning the league to Alkmaar’s 37.7%, reflecting that goal difference, and also that Ajax have scored more goals (68 to 54). Feyernoord maintain an outside chance at 7%.

How will Ligue 1 end?

We’ve looked at English leagues, at Serie ALa Liga, and the Bundesliga. Now we look at how France’s top division might have finished, had it been played out from the point the suspension began.

Twelve points clear, it’s fairy obvious PSG will win Ligue 1 this season. We have them at 99.9%, with Marseille having a 1 in a thousand chance of winning. There’s a lively battle for the remaining European places, with realistically only Marseille being guaranteed, at this point, European football in whatever form it happens next season.

There’s a 50% chance Rennes play in the Champions League, a 48% chance Lille do.

At the bottom, Toulouse are 98.5% likely to be relegated.

How will the Bundesliga end?

We’ve looked at English leagues, at Serie A, and La Liga, now we move to the Bundesliga in Germany.

Football has stopped everywhere (apart from Belarus). Will it restart? How? Behind closed doors? Before player contracts expire? The possibility of seasons being aborted seems all too real now. But how will end-of-season honours, European qualification, relegation, and so on, be determined?

In Germany, it’s been another season where Bayern haven’t dominated, but have moved into position when it mattered. Four points clear when the season was postponed, they are 77% likely to win the Bundesliga. Dortmund and Leipzig maintain a 10% chance each, Moenchengladbach and Leverkusen trivial likelihoods (1.3% and 0.8%).

How will La Liga end?

We’ve extended our end-of-season projections, which we did for English leagues before the COVID19 shutdown of sport, to Italy yesterday. This matters since it appears that increasingly, UEFA and leagues are moving towards inventive solutions to complete leagues before the end of June, when a lot of contracts expire.

So we’ve done the same for all of the other big five leagues, plus the Eredivisie in the Netherlands.

Here, we present La Liga. It’s hard to look beyond Barca. Although they are only two points clear, they have a 70% chance of winning La Liga. This reflects the different fixtures lists facing the two clubs, as much as anything else – our simulations run through all the remaining games of the season and count up how often each team wins particular games. Real Madrid have a 28.3% chance of winning, leaving 1.7% split between four teams – no chance, practically.

How will the Serie A season end?

A couple of months ago now, we began projecting forward our forecasts to the end of the season, inserting actual match outcomes into matches according to the forecast expected goals for each team.

We didn’t realise at the time how important this might come to be, as the coronavirus pandemic took hold, and has led to the postponement of all the major football leagues around the world, except a tiny handful (including Belarus).

We’ve shown the Premier League a few times – Liverpool are 100% likely to win the title, but everything else is up for grabs.

Across Europe, talk continues about cutting seasons short – but how could that happen? How could it happen when things remain so uncertain? For example, in Italy Juventus are only a point above Lazio, while Inter’s game in hand could still keep them in the race. How could that be ended now? Who would be awarded the title?

Well, we can use our forecasts projected to the end of the season to get some idea of the uncertainty attached to outcomes. It’s clear that Juventus (100%) and Lazio (99.8%) will both be in the Champions League, and it’s almost certain Inter (96.3%) and Atalanta (81.9%) will be, too.

But the title? Juventus are 52.6%, Lazio 42.2%. Inter are 4.7%, and even Atalanta have the most outside of outside chances at 0.5%.

Championship, R38 (13-15 Mar, 2020) — Lots of people will catch Covid19, and some football will be played

The table below gives the Model’s forecasts for Round 38 of the English Championship, along with forecast probabilities estimated from current online bookmaker odds.

EFL Champs, R38 2020, RED
  • Expected Goals: the forecast average number of goals the Model expects for Home and Away teams
  • Outcome Probs: the model predicted % chance of either a Home or Away win, with 100 minus these two numbers being the % chance of a draw.
  • Score Picks: the Most likely forecast scoreline outcome, as well as the most likely conditional on the most likely result outcome happening.
  • Home wins / Draws / Away wins: the predicted % chance of various potential scoreline outcomes of the match.
  • Mean odds: Estimates of online bookmakers’ average probability forecasts for the Home and Away teams to win, which could be compared with Outcome Probs.

Champions League Last 16 Second Legs

We enter into a rather unique spell in European football at the moment, with domestic football in Italy suspended for the foreseeable future due to coronavirus, and many other matches being played behind closed doors – including two this evening, in Valencia in Spain, and in Paris in France, two countries with over a thousand cases of the virus.

Our Model is a bit under the weather it seems, unable to determine which four matches take place this week, and which ones next week. So there are some early forecasts for the games in Manchester, Barcelona and Munich (assuming they take place next week).

As always, we go up against Jean-Louis Foulley, whose forecasts are posted in the second table. We agree which team is favourite in each match, but disagree a little on the strength of Atalanta as they travel to Valencia.