'Elo Again

A few months ago, possibly longer, someone enquired as to whether I still had my articles on Elo Ratings in Football available. These were initially published around 2012 on the Betting Expert website, but don't appear to be available any longer.

At the time I responded that I didn't have them, since I couldn't find them anywhere, but while researching the logarithmic method of removing bookmakers' margin from prices at the weekend, I came across the Elo Ratings articles.

Originally written as a four part series, these are now published as one post here, and while they are somewhat dated now, it's interesting to look back on and see how some of my basic ideas from back then are now, albeit with significant enhancements, now mainstream with the xG revolution.

Why was I researching the logarithmic method of removing bookmakers' margin from prices you ask? A very good question, since the "Margin to Proportional Odds" (MPTO) method I typically use is far more simple, and for the odds ranges I'm usually interested in for Draws, results in almost the same fair odds anyway, but it was a good exercise for my aging brain. If you're interested in this topic, there is no finer readily accessible resource than Joseph Buchdahl's The Wisdom of the Crowd publication. 

For the Budeslayga System I use the Equal Margin (No Bias) method as this gives a worse price for the odds-on lays which is usually very close to the Lay price on the Exchanges. 

An example of this difference is illustrated by the winning Lay of Bayern Munich at Augsburg last weekend, where the price was 1.20 using the Equal Margin method, but 1.18 using MPTO. 

Over time these small differences add up of course, so since 2012 the 'official' ROI is 5.12% whereas using MPTO it would be 6.65% (from 3,947 selections). 

No 'real' football this weekend, but for anyone following the Bundeslayga Systems, here's a table of all the clubs who have featured 80 or more times, and their ROIs.

Borussia Dortmund top the list in terms of units won, but interestingly only 0.18 units (ROI 0.6%) are from laying them as the Home club. The ROI on laying Dortmund as an Away selection is 25.4%.

For Bayern Munich, it's the opposite. When opposing Batern as an Away selections, the ROI is a losing one at 5%. 

Only three clubs are losers both Home AND Away, with Borussia Moenchengladbach the worst by some distance, but Schalke 04 and Hertha Berlin are also unhelpful to us. On the positive side, 11 clubs are profitable both at Home and Away, although Ingolstadt 04 were relegated last season and will no longer be candidates, at least for this season. 

Quite why Moenchengladbach should be such an outlier is interesting, but clearly the market perpetually underestimates them in these matches. Fortunately they do not look likely to be a qualifier next weekend!