If you are looking for a way to beat the bookies, you can consider using rating models. Rating Models are a numerical assessment of a player or teamβs strength relative to the league where they play. There are many handicapping models used by players and bookies alike in order to determine the likely outcome of a game or match. In this article, we will discuss Elo Ratings and how to use them to predict football results in order to improve your betting strategy.
- 1 What are Elo Ratings?
- 2 Chess Elo Ratings
- 3 Football Elo Ratings
- 4 Elo Rating Equation
- 5 Converting Elo Ratings into Odds
- 6 Using Elo for Football Betting
- 7 Making your own Elo Ratings
- 8 Criticizing Elo Ratings
- 9 Other Well-Known Rating Models
- 10 ThePuntersPage Final Say
- 11 Elo Ratings in Betting FAQs
What are Elo Ratings?
Elo ratings are a numerical way to rank the skill level of various competitors πbased on their performance as well as the relative skill level of their opponents. The idea is that beating a tougher opponent should give you more points than beating an easier opponent. The better the opponent, the more points you will score. The rating system can be applied to any sport, though it may need some modifications.
Hungarian American chess player and physics professor originally developed Elo ratings as a chess rating systemπΈ in 1960. By 1970, the World Chess Federation, FIDE formally adopted the Elo Rating System to rank Chess players.
Chess Elo Ratings
The World Chess Federation FIDE maintains an official list of rankings at π, which it uses to determine the top seeds in a tournament. Currently, the leader is Magnus Carlsen, with an Elo Rating of 2881 points. He is also the record holder of the highest Elo rating ever, at 2889.2. Legendary chess player Bobby Fischer got up to 2789.7 in 1972, while Kasparov had the second-highest rating ever at 2856.7 in 2000.
Sites like also offer updated chess ratings. As the title of the site suggests, a rating over 2700 puts you in the top tier of this competitive strategy sport (which you can also bet on through sites like William Hill).
Football Elo Ratings
π While Elo ratings were originally developed for chess, they can easily be modified for football or any other sport. In fact, there are two main websites that offer Elo ratings for clubs and national football teams respectively.
World Football Elo Ratings
has an Elo Rating system for international football π°that ranks countries based on the system proposed by Dr Arpad Elo. Their Elo rating system takes into account the kind of match, the home team advantage, and the goal difference in the final match result.
In 2018, based on Elo ratingsπΌ, as many pundits had criticized their former calculation methods as not dynamic enough. To this date, they still use the system to determine seeding in international tournaments such as the World Cup.
Clubelo: Club Football Elo Ranking
While determining top seeds is less relevant in club football, since fixtures are often laid out from the beginning of the season, they can still be a handy tool in predicting match results. π― offers updated Elo rankings for all European football leagues, including Premier League and League 2. Their team Elo rating takes both national and international football matches into account.
Elo Rating Equation
The Elo Rating Equation for Football takes into account home field advantage and goal difference, as well as win/loss/draw to create a rankings system. A commonly used equation is:
Rn = Ro + K Γ (W – We)
-
Rn is the new rating.
-
Ro is the prematch rating.
-
K is the weight constant for the tournament played. (For example, Eloratings.netπ½ uses 60 for World Cup finals, 50 for continental championship, 40 for World Cup and continental qualifiers, and 30 for other tournaments, with just 20 for friendly matches. It also needs to be adjusted for goal difference).
-
W κ¦is the result of the game (a team get 1 for a win, 0.5 for a draw, and 0 for a loss).
-
We is the win expectancy.
Calculating Win Expectancy
As you can see, in order to calculate the Elo Rating, you first need to calculate the Win expectancy. To do this, use the following formula:
We = 1 / (10(-dr/400) + 1)
dr is the difference in ratings plus 100 points for a home team.
The following chart shows some examples of what a win expectancyβ looks like for difference in ratings from 0 to 800 points.
Ratings Difference | Higher Rated team | Lower Rated team |
---|---|---|
0 | 0.5 | 0.5 |
10 | 0.514 | 0.486 |
20 | 0.529 | 0.471 |
30 | 0.543 | 0.457 |
40 | 0.557 | 0.443 |
50 | 0.571 | 0.429 |
60 | 0.585 | 0.415 |
70 | 0.599 | 0.401 |
80 | 0.613 | 0.387 |
90 | 0.627 | 0.373 |
100 | 0.64 | 0.36 |
110 | 0.653 | 0.347 |
120 | 0.666 | 0.334 |
130 | 0.679 | 0.321 |
140 | 0.691 | 0.309 |
150 | 0.703 | 0.297 |
160 | 0.715 | 0.285 |
170 | 0.727 | 0.273 |
180 | 0.738 | 0.262 |
190 | 0.749 | 0.251 |
200 | 0.76 | 0.24 |
210 | 0.77 | 0.23 |
220 | 0.78 | 0.22 |
230 | 0.79 | 0.21 |
240 | 0.799 | 0.201 |
250 | 0.808 | 0.192 |
260 | 0.817 | 0.183 |
270 | 0.826 | 0.174 |
280 | 0.834 | 0.166 |
290 | 0.841 | 0.159 |
300 | 0.849 | 0.151 |
325 | 0.867 | 0.133 |
350 | 0.882 | 0.118 |
375 | 0.896 | 0.104 |
400 | 0.909 | 0.091 |
425 | 0.92 | 0.08 |
450 | 0.93 | 0.07 |
475 | 0.939 | 0.061 |
500 | 0.947 | 0.053 |
525 | 0.954 | 0.046 |
550 | 0.96 | 0.04 |
575 | 0.965 | 0.035 |
600 | 0.969 | 0.031 |
625 | 0.973 | 0.027 |
650 | 0.977 | 0.023 |
675 | 0.98 | 0.02 |
700 | 0.983 | 0.017 |
725 | 0.985 | 0.015 |
750 | 0.987 | 0.013 |
775 | 0.989 | 0.011 |
800 | 0.99 | 0.01 |
Adjusting for goal difference
K πis a number you need to adjust goal difference in the game.
-
2 goals increase it by 50%
-
3 goals increase it by 75%
-
π·4 or more increases it by 3/4 + (Goal difference β 3)/8
Example
Let us look at a hypothetical exampleκ©΅. Belgium is the current Elo leader (April 2020), with 2084 points.England is currently at 12th place, with 1956 points.
π If the two were to meet for a friendly, in Belgium, we can calculate their respective win expectancies before the match, as well as how the match result would affect their respective Elo ratings.
First, letβs calculate their win expectancy, using the formula We = 1 / (10(-dr/400) + 1):
π · The ratings difference is 2084 β 1956 +100= 228.
-
Win expectancy for Home team Belgium = 1 / (10(-228/400)+ 1) = 0,789
-
Win expectancy for away team England = 1 β0.789 = 0.21
Let us assume that Belgium beats England again with 2 β 0, just like in the . How would this affect each teamβs Elo Ranking?
Belgium
Using the formula Rn = Ro + K Γ (W – We), we reach the following result:
Rn = 2084 + (20 x 1.5) (1 β 0.798) = 2084 + 6.06 = 2090
-
2084 is Belgiumβs previous ranking
-
20 is the weight constant for friendly matches
-
β5 accounts for a 50% increase for a 2 goal difference
-
1 is for a win
-
798 is the win expectancy
England
Using the formula Rn = Ro + K Γ (W – We), we reach the following result:
Rn = 1956 + (20 x 1.5) (0 β0.21) = 1956β6.3 = 1950
-
1956 is Englandβs previous ranking
-
20 is the weight constant for friendly matches
-
β5 accounts for a 50% increase for a 2 goal difference
-
0 is for a loss
-
0.21 is the win expectancy
Converting Elo Ratings into Odds
While Elo ratings rely on Win expectancy for predicting how well a team will perform, Win Expectancy does not always equal the probability of winning. Football is a sport with three possible outcomes: win, loss and draw.π₯ In order to get more or less accurate odds based on Elo ratings, you will need to be a bit creative.
Here is a possible system for converting Elo ratings into Odds:
π Calculate the difference in points between the two teams.
Give the home team a 7% advantage.
πΉ The highest-ranked team has a base chance of 35% to win. For every 10 Elo points they are higher, add 1%.
πΉ Odds of the strongest team winning are 1 divided by chance of winning.
π To calculate the odds of a draw, assume 30% as a base chance. Ignore the first 100 Elo points difference, but for every 20 points above 100, take away 1%.
π΄ Odds of the draw winning are 1 divided by that chance.
π For the odds of the weaker team, simply subtract the chance of strong team win and the chance of a draw from one.
ΰΌ Odds of the weaker team winning are 1 divided by that chance.
Using Elo for Football Betting
We have said this in many other articles, but we will say it here again: the most important thing any punter needs to do is search for high-value bets. When you want to place a betπ on the outcome of any football match, you should first consider whether the bet has any value.
κ§ Elo ratings can be a valuable tool for predicting what a match result will be, based on each teamβs past performances. To use Elo for football betting, follow these steps:
Look at the difference in Elo rating.
Calculate the win expectancy with the formula We= 1 / (10(-dr/400) + 1)
β (dr is the difference in ratings plus 100 points)
Create your own probabilities and odds with the method above.
Compare your probabilities and odds to the bookieβs odds and implied probability.
If you see the value, you can consider betting.
Keep in mind that points 1, 2 and 3 are all indicators of a teamβs relative strength.
Making your own Elo Ratings
While many sports have official rankings of their own, not all use Elo ratings to rank their players or teams. In the case of football or chess, there are readily available Elo ratings. On the other hand, many smaller sports have no official or unofficial ratings at all.
If you want to bet on a smaller sport, you could consider creating your own Elo ratings. To do this, you will first need a starting point.
-
You can assign each player or team an arbitrary number of points – for example, 1400.
-
κ§Using the formulas above, you can calculate Elo ratings as players compete against each other.
-
βAs the season develops, your personal Elo ratings will become an increasingly accurate tool for predicting match outcomes.
In individual sports, such as boxing, it is best to start your rating system at least a few years back to have as accurate data as possible. In team sports, this can be relevant too; however, the makeup of teams usually changes a lot between seasonsπ°, making it harder to rely on data from previous seasons.
Criticizing Elo Ratings
Elo Ratings were primarily designed as a rating system in order to create seeds for a tournamentκ©΅. Of course, they are not entirely accurate in predicting who will win a specific match or tournament.
π Belgium held the highest Elo ranking from 5 November 2015 to 7 April 2016. They also regained the spot on 20 September 2018 and still hold it (as of April 2020). However, the squad did not do as well as expected in the 2016 European Championships under Marc Wilmots, losing to Wales in the quarterfinals. In 2018, they did better under former Everton Manager Roberto Martinez, beating England for third place.
κ§ Even though Belgium was surpassed by both France and Croatia for the World Cup final, France still holds only third place, behind Brazil (at the current time of writing, April 2020). So it is easy to see how the system has flaws and that there is room for improvement.
Of course, the Elo formula allows you to create different modifications and weightings. You can easily play around with the K factor to adjust for all kinds of variables.
Other Well-Known Rating Models
κ©΅ Besides Elo, there are several other popular rating models. We will discuss the most prominent ones.
Glicko Rating
The were invented by Mark Glickman as an improvement on the Elo system. They were designed with games of skill in mind, such as chess and Go. They are currently in use on various online game serversꦿ and are popular for video game rankings. These include Pokémon Showdown, Free Internet Chess Server, Online Go Server, Counter-Strike: Global Offensive, Team Fortress 2, Dota Underlords, Guild Wars 2, and many more.
It introduced the concept of RD, Ratings Deviationβ, which indicated how accurate the score is. For example, a player with 1500 points and an RD of 50 has a real strength between 1400 and 1600. It also takes recency and time not played into account.
TrueSkill
TrueSkill was developed by Microsoft to match players in Xbox Live games. It takes team games into account and also accounts for how accurate the score is. It uses Bayesian statistics to make predictions.
Custom Ratings
There can be thousands of rating models; some have more official status for seeding tournaments, while others are custom creations by individual handicappers. Mathematical models from video gamesπ are some of the most accurate, since they have been tested on millions of players and billions of games.
ThePuntersPage Final Say
Elo Ratings can be a handy tool for finding value. Of course, they do not offer guarantees. If you want to take betting on football or chess seriously, you can find a wealth of Elo ratings that are ready to use. πFor smaller sports or leagues, you can consider creating your own Elo ratings, which may be a handy way to beat the bookies.
Elo Ratings in Betting FAQs
π²In terms of international country football, the top 10 countries have an Elo Rating between 1967 and 2084. In European club football, the top 10 clubs have a rating between 1848 and 2020.
ཧIn chess, tournament players often have an Elo rating between 1500 and 1600. An experienced player with 1 to 2 years of experience has an Elo rating of 1100 to 1200. A rating of 1200 to 1400 puts you at a competitive level at Class D, category 4. A grandmaster has an Elo between 2500 and 2700. Magnus Carlsen, the current world champion, is the record holder of the highest Elo rating ever, at 2889.2.