Saturday, October 01, 2016

Scottish Football: Thinking About Elo

The American data journalism site fivethirtyeight.com uses Elo ratings pretty heavily in its sports coverage, but the ratings are actually pretty popular worldwide. A few points about Elo:
  • Elo was originally created as a chess rating system. It underlies the go ratings I've blogged about.
  • I used to think Elo stood for something, but it's actually named after its creator, Arpad Elo.
  • The difference in ratings is used to predict the probability of an upset.
  • The outcome of a match is used to adjust the ratings, with points shifting from the winner to the loser.
  • There are endless variations, including adjusting for score.
It is unsurprising that this system has been repurposed for soccer. I decided to take a look at the Scottish club rankings, as posted by clubelo.com.

This gives a pretty clear answer to something I've wondered -- is anybody else in Celtic's vicinity? No. Aberdeen has consistently been the second-best team, which I had realized, but I didn't know it would come across so strongly in the graph.  St. Johnstone, Hearts and Inverness form the next grouping of three.

I have been keeping an eye on the Scottish Premiership tables. It's very early -- less than 20% done, but the European places are currently occupied by Celtic, Aberdeen and Hearts -- plus the Scottish Cup winner, or St. Johnstone, if the Cup is won by one of the top 3.

So the top 3 all look good, but St. Johnstone is the club that had a streak of four weak years in the Europa League snapped by not appearing this year. If they're back, it's probably not a good sign for Scotland's UEFA coefficient.

Another thing that clubelo.com does is use the Elo rankings to project the rest of the Champions League group stages. According to them, Celtic has a 75.5% chance of doing worse than Borussia Mönchengladbach, a 97.4% chance of doing worse than Manchester City, and 100% chance of doing worse than Barcelona. In other words, a slim-to-none chance of qualifying for the knockout stage, but a 20-25% chance of dropping to the Europa League. I'll buy that.

Another thing the site tells you is the peak Elo rank of each club. Despite the fact that Celtic played minnow to Man City's whale in this week's draw, note that as recently as 2012 Man City was the #3 club in Europe, their best ever rating, which they held for a total of 33 days. But the #1 club back in 1972, when I was born -- Celtic, a position they held for a year and a half.

On the one hand, it's tempting to see that as a historical artifact. On the other, as recently as 2014, Celtic out-drew Man City (by an average of 4 fans!). While the TV money isn't comparable, there has to be some way of turning those resources into European competitiveness (and there was last week in Glasgow!). I don't think putting them into the English pyramid is the right answer, but maybe some sort of Atlantic League, where Celtic and Rangers could mix with top teams from other smaller European countries, would do the trick.
Post a Comment