How Do You Do Value Betting? (Part 1)
Q: Can an individual bettor do value betting?
A: Yes, but it should not be the only strategy.
For an individual bettor, it is impossible to hold accounts at every bookmaker, and it is also impossible to keep enough bankroll in every account at the same time. Value betting needs you to find valuable odds across many bookmakers, then place the bet. It does not care whether one single match wins or loses. It only cares whether the odds are valuable.
So for an individual football betting system, value betting is often better used as one auxiliary strategy, not the whole system.
Q: How do you choose valuable odds?
A: A simple way to understand value odds is this: bet on the highest price for a market.
But the highest price is not automatically valuable. After converting the odds into probability, if that probability does not match the real probability behind the market, then the odds still have no betting value.
So the real problem is: how do we model the value of a market?
Q: How do bookmakers "price" a football match?
A: When a bookmaker posts odds for a football match, it is essentially doing probability modeling.
The core logic is simple: first estimate the real probability of the match outcomes, then convert those probabilities into odds that include the bookmaker's profit margin.
In traditional finance and betting, the foundation of this model is often the Poisson distribution.
1. Core logic: goals and the Poisson distribution
The biggest feature of football is that goals are rare and randomness is high. Statistics show that the number of goals scored by a team in a match can be modeled with a Poisson distribution.
So the key to the model is predicting two numbers: the home team's expected goals (lambda) and the away team's expected goals (mu).
These two numbers are not guessed by feeling. They are decided by the team's attacking strength, defensive strength, home advantage, and the league's average goal level.
Formula:
lambda = e^(alpha_i + beta_j + gamma)
mu = e^(alpha_j + beta_i)
- alpha: attacking-strength coefficient of team i (the home team)
- beta: defensive-strength coefficient of team j (the away team)
- gamma: home-advantage coefficient, usually a constant such as 0.2 or 0.3
2. Practical example: Manchester United vs Manchester City
Suppose after backtesting historical data, we get the following coefficients. These numbers are only examples.
- Manchester United attacking strength: 0.3
- Manchester United defensive strength: -0.1
- Manchester City attacking strength: 0.5
- Manchester City defensive strength: -0.3
- Home advantage: 0.25
- Premier League average goal log: 0.35
Step 1: Calculate expected goals
- Manchester United expected goals (lambda):
lambda = e^(0.3 - 0.3 + 0.25 + 0.35) = e^0.6 ~= 1.82 - Manchester City expected goals (mu):
mu = e^(0.5 - 0.1 + 0.35) = e^0.75 ~= 2.12
Step 2: Calculate exact-score probabilities
Use the Poisson formula to calculate the probability of a score:
P(k | x) = x^k * e^(-x) / k!
- Probability of 0-0:
P(0 | lambda) x P(0 | mu) ~= 16.3% x 12.0% = 1.96% - Probability of 1-1:
P(1 | lambda) x P(1 | mu) ~= 29.7% x 25.4% = 7.54%
By continuing this process, we can calculate the probability of every scoreline, from 0-0, 1-0, 0-1, and so on to higher scores.
Step 3: Aggregate win, draw, and loss probabilities
- Home-win probability: add all score probabilities where the home team scores more than the away team, such as 1-0, 2-0, 2-1. Suppose the result is 30%.
- Draw probability: add all score probabilities where both teams score the same number of goals, such as 0-0 and 1-1. Suppose the result is 25%.
- Away-win probability: add all score probabilities where the away team scores more than the home team, such as 0-1 and 0-2. Suppose the result is 45%.
Step 4: Add overround and convert to odds
The bookmaker will not post odds directly from the real probabilities. If it did, it would have no profit.
Instead, it raises the total probability, for example to 110%. The extra 10% is the overround, or the bookmaker's edge.
- Adjusted probabilities: home win 33%, draw 27.5%, away win 49.5%. Total: 110%.
-
Odds conversion formula:
odds = 1 / adjusted probability -
Home-win odds:
1 / 33% ~= 3.03 - Draw odds:
1 / 27.5% ~= 3.64 - Away-win odds:
1 / 49.5% ~= 2.02
The market odds we see are created from this process.
Additional note
The Poisson-based model above is the foundation of the classic Dixon-Coles model.
Modern large bookmakers now also use machine-learning methods and large datasets, such as xG, player tracking data, and other market signals, to improve accuracy. But the bottom logic still cannot leave the same core problem: estimating the expected goals of both teams.