Premier League Betting: Don't overplay big margins of victory

Confidence may be high after two big victories, but don't overestimate Manchester City's chances against Liverpool

Margins of victory matter when it comes to rating football teams, writes Jack Houghton, but they don't matter as much as the market seems to think, which is why Manchester City might be a bad bet to win away against Liverpool next weekend...

In a probably futile attempt to sound more knowledgeable about football when I meet up with an old work colleague tomorrow, I decided to watch the BBC Breakfast summary of recent Premier League goings-on this morning.

The main story seems to be that Jose Mourinho, who manages Chelsea (you see, I'm learning...) and has been in trouble in the past for suggesting that referees are part of a "campaign" against his team, is upset that a series of key moments in the game against Burnley were not handled well by the referee, although he doesn't say this directly, for fear of being "punished" again.

The BBC lady presenting the Premier League summary wrapped up the piece by alluding to the shifting momentum of the season: Manchester City had profited from Chelsea's misfortune with a 5-0 win against Newcastle, having previously won 4-1 away at Stoke, and are now only five points behind in the table.

It's an obvious narrative, I guess, and as someone who is attempting to perfect a system for rating Premier League teams this season, it raised a couple of interesting questions:

First, how should large margins of victory - such as Manchester City's recent successes - be treated within a ratings' system?

And second, do such large margins of victory positively affect a team's subsequent performance, or are they over-interpreted?

Well, the answer to both questions is partly related.

The way that most ratings' systems treat margins of victory - including my own - is to award more points to a team depending on the size of their victory. As an example, the World Football Elo Ratings, which I have given high praise to before, constantly increases the size of the pot on offer depending on the goal difference secured by the winning team.

The only way I have figured out to determine how much to let goal difference influence ratings is to try out different formulas and see how predictive each is in terms of past results: the formula that gets closest to actual historical results is the one to go with. What I found was that, whilst goal difference is significant, and increases in significance as the size of goal difference increases, the rate of increase slows as the goal difference increases.

This is no big surprise: if a team wins by three goals, is this particularly less significant than them winning by five? Not really.

Which brings us onto the second question. Yes, large margins of victory do affect a team's subsequent performance, which is why they receive a higher rating going into future matches as a result. However, what's less clear is whether the market overreacts to such margins of victory, above and beyond how it should.

Looking at an admittedly small data set, it looks as if it does, to the tune of around 5% on average. Take Manchester City. According to my ratings, they should be around 2.77/4 to win their next Premier League match away against Liverpool, but the market is slightly more optimistic of their chances, viewing them as 2.447/5 shots.

And looking at past Premier League results, there doesn't seem to be a particularly positive "confidence effect" if a team wins a match by a large goal difference. In a crude piece of analysis I looked at every time a team won a game by three goals or more, and whether or not they won their subsequent game. On 55% of occasions this season, teams have gone on to lose or draw their next game following such a large margin of victory, and these figures are relatively consistent with previous seasons.

Recommended Bet
Lay Manchester City at 2.447/5 against Liverpool