How to convert Ratings into Betting odds
Ratings systems; the most favoured methods amongst punters for predicting the outcome of sporting events. This is because of the easy in which one can be made. But even with this easy, most punters will still run in to trouble once it comes to knowing when to place a bet. Why? Because knowing the relationship between these ranking scores and the betting worlds odds aren’t as straightforward as they seem.
This article will show you how to convert your rating scores into usable odds or percentages. And yet even with this information still most people reading this will have no success. Basically because their rating system has no correlation to that of the chance a team or person has of winning. It is scary some of the poorly thought-out rating systems I’ve read over the year, some of which come from well informed betting journalists.
The one thing you should always keep in mind when converting rantings to odds
Bad data in, equal’s bad data out
There is no method of converting ratings that is going to get around this. If you want reliable outputs in which to place a bet you will need to make sure your rating system has some actual bearing on the game or match outcome. Obviously this is easier said than done, and will take plenty of work on your part. And I’d highly advise paper testing your set of odds or comparing them to historical odds before diving in and splashing your cash about.
Elo ratings A respected rating system
For any of you that have taken the time to research ratings systems, you couldn’t have failed to have come across Elo ratings or rankings. And although this system will need tweaked depending on which sport its used for, it’s a very good foundation for anyone looking to start building their own ratings. Therefor we will use this rating system for the bases of this article.
What is Elo ratings
Elo isn’t an abbreviation for anything, like many assume, it is the surname of the man behind the system – Arpad Elo. A chess master, that devised the ratings for ranking chess matches, to determine each person’s chances of winning. Used solely for chess raking for many years before sports punters realised it’s benefits to predicting an array of different head-to-head events.
The maths behind Elo ratings is beyond the scope of this article, I’ll be showing you how to convert any ratings into odds, so if this is something you wish to use I’ll leave it to you to research.
Converting ratings to odds
Let’s take a look at a real life scenario for this example. We are going to use Elo ratings for the English Premier League. As we are using Elo just to demonstrate how to convert ratings to odds, I’ll not be compiling my own Elo ratings (I actually use my own rating system so don’t have an Elo one at hand). We will be using ratings compiled by ‘Since a win’ who have slightly adapted Elo to take into account home and away team advantages.
Below is a table of how the ratings currently stand at the time of writing this article (03/09/15).
With these rankings on their own it would be impossible to know the match odds of any team against another. A formula would be needed to convert the team ratings to odds. Unfortunately there is no one size fits all formula; but once you have calculated the formula for your set of ratings you should be good to use this for all future games.
The method used to calculate the formula is regression analysis. Now don’t hit the back button, there is no need to run away. Regression analysis sounds a lot worse than it is. In fact if you have a basic comprehension of Microsoft Excel this is very easy and our example will be using Excel.
Regression analysis is used for estimating the relationship between variables. In our case we want to know the relationship between home and away team ratings and odds. For this you need to compare your ratings against a previous set of odds or those for the upcoming games. We will use odds from week 4 in the Premier League. And once this is done we will calculate the odds for week 5.
Week 4 Premier League odds
With these odds and Elo ratings we will now show how to do a regression analysis for home team odds, we’ll leave it to you to repeat the process for away team and draw odds. First covert the odds to percentages (1/odds * 100) and then add these to excel along with the home and away team ratings.
Now just simply run Excel regression (make sure you have the analysis ToolPak installed). Below is how it looks on my addition of Excel. I have selected the team percentages for ‘Input Y’ values and the home and away ratings for ‘Input X’ values
Once you click ok, all the work is done for you. All pretty simple so far. But knowing what the output values mean and what to do with them isn’t as obvious. Let’s look at the data this regression analysis returns and highlight the important parts.
Running through the highlighted cells from top to bottom.
- R Squared – The closer to 1 the better (our 0.91 means we have a very good fit). It means 91% of variation in ‘percentage chance’ is explained by the variables ‘home ranking’ and ‘away ranking’.
- Significance F – This is the statistical significance. You need a value less than 0.05 here for the significance to be relevant.
- Coefficients – This is the important bit. These are the values we need to build our formula. For any future game, Percentage chance (for the home team) = Intercept + (X Varibale 1 * Home Rankings) + (X Variable 2 * Away Rankings).
Compile odds from rankings
Ok, that’s the entire boring bit out the way, and thankfully that will only have to be done once. We now have the formula that we need and to calculate the odds for week 5’s games all we need to do is fill in the data to that formula. So let’s do that and see how closely it relates to the real odds.
As you can see, for several of the games the converted ratings have been pretty good at getting the correct odds, but a few have been pretty poor. Of course this wouldn’t be good enough to be spending your hard earned cash. A couple of reasons I’ll give are I have only done the regression analysis on one round of games. The more games you do it on the better. Also these are not my ratings, so I have no idea how long Elo maths have been running on it or how promoted teams have been dealt with (as they have entered the league with no history).