Football is a statistician’s nightmare. There is no single metric you can look at and confidently say which team was better. Goals, the scoring mechanism of the sport, are highly noisy too. What this means is that if you a pick up any shot in a game and play that same situation repeatedly the average player is just as likely miss or score.
Looking at trophies won is not a great idea either. Arka Bhattacharya attempted to do this in his brilliant article here. But here’s the problem, in his list of big clubs in India – 3 teams aren’t even top-flight teams anymore. Greatness comes from winning trophies but not all great teams win trophies. Even in a league, and especially in India, I don’t think final positions are a great indication of a team’s overall quality. What’s up Burnley and Minerva?
Okay, so what about points won? Again, not a great idea because a team will get the same points whether they beat BFC or NorthEast United. Another weakness with using a points based system is how do you reward teams that win crunch games? Mumbai City started the season by picking up points but when it was the time to perform, they didn’t.
Therefore, I present to you – ELO RATINGS. This is a system most widely used in Chess to ensure that newbies don’t get paired against Grandmasters but all it does is quantify how good a player is. This system is also used by FIFA to calculate the Women's National Team rankings.
Here’s how it works,
Every Team starts the season with 1500 points
The game is zero-sum. If Team A defeats Team B, Team A steals points from Team B
The number of points stolen is dependant on the Goal Difference in the game and the strength of the two teams
Team A will get more points if they beat a strong team 4-0 rather than beating a weak team 4-0.
In this system, teams that go on winning will have score above 1500 and teams that lose a lot will drop below the 1500 mark. Assume that a team with 1500 points is the average team in the table.
Have some feedback? Catch me on Twitter - @sgtsaltnpeppa