Concept
Think about the following questions on football. Which is the best team in the World? Which team is going to win the El Classico? A Premier League team is likely to win against a lower league team. But the question is, by how many goals? What is the exact chance? Is that a value bet?
Those are probably some of the most commonly discussed topics amongst football fans and punters but it is not easy to draw a conclusion. We would have a rough idea on how good is the team but it is not easy to turn the idea into numbers.
Back in 2015, we invented a system called TeamForm Rating System to rate all football club and national teams and predict football match results. We are a small group of football enthusiasts who are determined to deriving an effective and repeatable algorithm to predict football results better than bookies.
We analyse and package data. Our aim is to empower you to reveal the real strength of football teams and predict football results better than ever. No matter you are a beginner or expert in football betting, you will benefit from our findings to overcome bookies’ advantage and improve your chance of winning. Let me explain how our system works.
What is TeamForm Rating?
TeamForm Rating (TFR) represents the form and ability of football team, and is derived by examining the past match results. Strong teams have higher ratings. The ratings can help us to predict football results and increase our winning chance in football betting.
The ratings are derived from thousands of historical football results, using a sophisticated rating system involving statistical models and mathematical formulae. Every team can be compared with each other, including both national teams and club teams.
We analysed dozens of deciding factors and went through repeated trail-and-error processes. Then we found a way to turn the ability of football teams into meaningful numbers. As a rule of thumb,
10 TeamForm Ratings points = 1 Goal
Let’s say Team A with TeamForm Rating 100 play against Team B with TeamForm Rating 90 at a neutral ground (no home advantage). Their difference in TeamForm Ratings is 10 and team A will most likely beat team B by 1 goal. There are other factors taken into account but TeamForm Ratings give us a fundamental base in predicting the outcome of football matches.
TeamForm Rating tells us how good a football team is. More precisely, rather than the absolute ability, it represent more on the domination of a football team, compared to other football teams in its own era.
How the rating system works
TeamForm Rating system is not just a ranking system to award team achievements such as winning matches and trophies. It is a rating system to evaluate the actual ability of football teams and to predict the football match results.
At the beginning, an initial rating is assigned to each team. Then predictions of outcome are derived based on a number of factors such as TeamForm Ratings, home advantage, form, past performances. After the match, the teams will exchange rating points to each other. The number of points exchanged depends on the factors such as predicted result, actual result, importance of the match, abilities of teams and so on. The team with better performance than expected will have gain rating points, while the other lose the same amount of rating points. This process repeats and the system will be settled and become effective after a period of time. The higher the rating, the better the team.
TeamForm Rating vs Quality of Team
100+: Top 3 to 5; the most dominant teams in the world and favourites to win Champions League.
95 to 100: Top 10 to 15; major contenders in Champions League; have the quality to enter semi-finals and challenge for the title.
90 to 95: Top 20; the major teams in top European Leagues; could proceed to knockout stage of Champions League; have the potential to upset top teams with good performance.
85 to 90: Top 50; decent teams in major European Leagues; fight to quality to Champions League knockout stage.
80 to 85: Top 125; average teams in top European leagues; majority of top South American teams are in this category.
The above explanations give a basic understanding on how TeamForm Rating relate to ability and quality of football teams worldwide. Generally we can consider any team with TeamForm Rating over 90 a very good team and is a major force in European football. National teams are usually about 5 points behind the club teams in same category.
Methodology
Scope
The system is sourced from over 1.7 million historical results of 700+ football competitions in 160+ countries, which covers both club and national teams. This huge scope can improve the accuracy of the rating system as data error due to missing matches is minimised.
Start values
Based on the historical results in first three season of the inter-continental club competitions (e.g. Champions League in Europe, Copa Libertadores in South America), the standard of football clubs in each country was derived and assigned as start values of TeamForm Ratings to all teams in the country.
Match predictions
Before the match start, the system will predict the number of goals scored by home team and away team based on their TeamForm ratings and the estimated home ground advantage.
With the expected number of goals scored, the system convert the numbers into happening chances of each correct score, using an enhanced Poisson distribution. Having know the probabilities of each correct score, we will be able to deduce the fair odds of betting markets such as 1X2 result, total goals over/under, both teams to score and many more.
Points exchange
After each match, both teams’ TeamForm rating will be adjusted based on the performance, with the following formula:
New TFR = Old TFR + K x (Actual Performance - Expected Performance)
Expected performance is from the match predictions.
Actual performance is measured from factors such as goal scored, goals conceded, goals timing, bookings, etc.
K is a weight index which covert the performance into rating changes and is calculated from following factors:
- Importance of the match. For example, the K factor in Champions League Final is larger than that of a friendly match.
- Rating difference between 2 teams. The K factor is smaller if the rating difference is large since the stronger team usually will not give 100% when playing against a much weaker opponent. This kind of match is not the best reflection of both teams' true ability therefore a smaller K factor is assigned.
- Consistency of starting lineups. If the starting lineup of a team is not consistent in recent matches, the performance is not reflecting their true ability and so a smaller K factor is assigned.
A larger K index will have the ratings respond to performance quicker but will suffer from more variation. A smaller K index provides more stable values but may fail to respond to the true values in time. A balanced K index is crucial for the system to be accurate and effective.
The winning team is not guaranteed to earn rating points. Let's say a strong team play against a weak team at home and the home team is expected to win by 2 goals. If the home team won by 1 goal only, they will lose rating points.
Example calculations
Man Utd vs Chelsea16 April 2017Premier LeagueTeamForm Ratings: Man Utd (94.1), Chelsea (96.0)
Expected goal supremacy = (TeamForm rating (home team) - TeamForm rating (away team)) / 10 + home advantage= (94.1-96.0) / 10 + 0.46 = 0.27
Although TeamForm Rating of Man Utd is lower than Chelsea's, with home advantage Man Utd was still the favourite to win the match according to our predictions.
Man Utd was expected to beat Chelsea by 0.27 goal. From the statistics we estimated the number of total goals scored in this match is 2.62. This number is derived from a few factors including the competitiveness of the match, past performance of both teams and average number of goals scored by Premier League teams.
From our customised Poisson Distribution we deduced the probabilities of outcomes as follows:Man Utd win: 41.4%Draw: 28.3%Chelsea win: 30.3%
Finally Man Utd beaten Chelsea 2-0, with Rashford scored in the 7th minute and Herrera scored in the 49th minute.
New TFR = Old TFR + K x (Actual Performance - Expected Performance)
Having considered the number of goals scored and goals timing, the actual performance of Man Utd was 1.12 goals better than expected while Chelsea’s performance was 1.12 goals worse than expected.
The K factor for this match is 0.89 and the TeamForm ratings of Man Utd and Chelsea were updated as follows:Man Utd’s new TeamForm rating = 94.1 + 0.89 x 1.12 = 95.1Chelsea’s new TeamForm rating = 96.0 + 0.89 x (-1.12) = 95.0