World Cup 2014: Spain's pain is rating followers' gain

Spain before the coin toss. 1.9 anyone?

In his third instalment, Jack Houghton, a self-confessed know-nothing when it comes to the beautiful game, shows how ratings are not infallible, but by converting them into fair match odds, you can make sure you're on the right side of any bet.

This series of blogs aims to demonstrate how ratings can be used to better inform the armchair punter and, so far at least, things are looking promising. In the first instalment I showed how even relatively crude rating systems can be used to identify teams who are over- or under-valued, drawing attention to the relative hyping of African teams, who may not necessarily deserve the plaudits they receive.

So far, the pattern of the last World Cup has been repeated: the five African representatives have played nine games and won two; and whilst Ivory Coast's victory over Japan, and Nigeria's victory over Bosnia, saw dips in the betting bank, they have been more than made up for by the other seven losses.

In the second instalment I showed how more sophisticated rating systems - some of which are freely available - can further help identify value bets. Put forward in that blog were Mexico as likely qualifiers from Group A (who are now 1.618/13 to do so) and Germany and Argentina in the outright market, who have both had positive starts to the tournament. Unfortunately, tipping Greece and England as value to qualify from their groups has backfired. And this leads on neatly to the focus of this blog...

On Friday I received this from a friend, who wants to remain anonymous:

"If ratings are so good, how come England have bombed? And how come Spain, second-best ranked in the world, are already out?"

Fair questions! However, they miss the point of using ratings. No system is going to perfectly predict the future. If they could, then the World Cup could be played out in a studio with a spreadsheet, with reaction provided by Des Lynam and Jimmy Hill for old time's sake. Football matches are affected by a number of variables which will affect the performance of any team on a given night, and so, theoretically, any result is possible.

Ratings, though, allow you to assess who is more likely to win a match and, more importantly, what the fair odds should be for that match. If you are then able to bet at odds that are higher than those fair odds, you should, if your ratings are any good, make a profit in the long-run. Or, to quote the oft-used example, if you are offered 1.910/11 about a coin-toss, you know that you will lose money in the long-run if you take the bet.

Take Spain's opener against Holland. Going into that match, Spain held an Elo rating of 2085, compared to Holland's 1957. In other words, Spain were rated as 128 Elo points superior.

Now, converting Elo points into match odds can be a mathematically involved affair, and for anyone who is interested, an internet search will lead you to plenty of debate on the subject and hours of sometimes fruitless spreadsheet experimentation; but, for the sake of making it straightforward, I can tell you that, according to my spreadsheets, every 100 Elo points of superiority roughly equates to a team having a 10% increased chance of winning a match, starting from a figure of 35%.

Spain, then, according to their rating, stood a 48% chance of winning against Holland. By simply dividing 1 by 0.48, this tells you that their odds for winning should have been 2.0811/10, roughly a coin-toss. So anyone who took the 1.910/11 available was on the wrong side of the coin toss. Sound familiar?

Over the coming games, try this formula as a way of identifying value bets:
1. Find the difference in the Elo rating between the two teams.
2. Take the strongest team as having a 35% chance of winning, and then add 10% to that figure for every 100 Elo points they are superior (170=17%, 268=27%).
3. Convert that into their odds of winning by dividing 1 by their chance of winning (e.g. 0.56). You now have an idea of what price they should be to win the match.
4. Now assume 30% as the chance for a draw. Take away 5% for every 100 Elo points that the better team is superior.
5. Convert that into odds of a draw in the same way as described in 3, above.

There you have it! There is obviously a margin of error in this method, as it is a fair bit more complicated on the spreadsheet, but, for the friend who emailed me, it might help explain why I don't care that Spain are out.

Did I mention that I laid them against Holland?