Using xG and Poisson Maths to Estimate Score Draw Chances

Expected goals (xG) has moved from a niche analytics term to something quoted on Saturday night highlights shows, but most casual use of it stops at “this team created more good chances than their score suggests.” Pushed one step further, with a bit of basic probability theory, xG can actually estimate how likely a specific scoreline is — which is exactly the question a treble chance coupon needs answered.

What xG Actually Measures

Expected goals assigns a probability value to every shot a team takes, based on factors like distance, angle, and the type of chance (a close-range tap-in scores close to 0.9, a tightly angled long-range effort might score 0.03), and sums these values across a match. A team accumulating 2.1 xG created several good chances, even if they only converted one of them into an actual goal that day. Over a single match, this is just a measure of created chance quality; it becomes genuinely predictive when averaged across several matches, smoothing out finishing luck in either direction.

From xG to a Probability Distribution: The Poisson Model

Goals in football are relatively rare, discrete events, which makes them a reasonable fit for the Poisson distribution — a statistical tool used for modelling the probability of a given number of events occurring in a fixed window, when those events happen independently at a roughly constant average rate. Applied to football, a team’s average expected goals for a match becomes the Poisson “rate,” and the distribution tells you the probability of them scoring exactly 0, 1, 2, 3 or more goals.

The Poisson probability formula for exactly k goals, given an average rate λ (lambda, the team’s expected goals for that match), is:

P(k) = (λ^k × e^−λ) ÷ k!

A Worked Example

Take an illustrative fixture where pre-match analysis suggests the home side has an expected goals rate of λ = 1.3 for this match, and the away side a rate of λ = 1.1. (These figures would typically come from blending each side’s season-long attacking xG with the opponent’s defensive xG conceded — the specifics of building that estimate are a separate exercise, but the numbers themselves are the inputs we need here.)

Step 1: Probability of the Home Side Scoring Exactly 1

P(1) = (1.3¹ × e^−1.3) ÷ 1! = (1.3 × 0.2725) ÷ 1 ≈ 0.354, or about 35.4%.

Step 2: Probability of the Away Side Scoring Exactly 1

P(1) = (1.1¹ × e^−1.1) ÷ 1! = (1.1 × 0.3329) ÷ 1 ≈ 0.366, or about 36.6%.

Step 3: Combine for the Joint Probability of a 1-1 Draw

Because the two teams’ scoring is treated as independent in this simplified model, the probability of both specific outcomes happening in the same match is the product of the two individual probabilities:

P(1-1) = 0.354 × 0.366 ≈ 0.130, or roughly 13% for this specific scoreline.

Step 4: Repeat for 2-2

P(home scores exactly 2) = (1.3² × e^−1.3) ÷ 2! = (1.69 × 0.2725) ÷ 2 ≈ 0.230, or 23.0%.

P(away scores exactly 2) = (1.1² × e^−1.1) ÷ 2! = (1.21 × 0.3329) ÷ 2 ≈ 0.201, or 20.1%.

P(2-2) = 0.230 × 0.201 ≈ 0.046, or roughly 4.6%.

Adding Up All Score Draw Possibilities

For this illustrative fixture, the model suggests just over a one-in-four overall chance of a draw of some kind, with 1-1 the single most probable individual scoreline — useful, specific information that a simple “both teams tend to draw a lot” observation wouldn’t give you.

Where the Model Falls Short

This simplified approach treats home and away scoring as fully independent, which isn’t quite true in reality — a match that turns cagey tends to suppress both sides’ scoring together, while a chaotic, end-to-end game inflates both. More advanced models adjust for this correlation, and also factor in red cards, weather, and tactical context mid-match, none of which a basic two-input Poisson calculation captures. Treat the numbers above as a structured estimate, not a guarantee.

Using This Practically

1. Gather recent xG-for and xG-against figures for both sides, ideally from their last six to eight matches, split by home and away performance.
2. Blend the two into a working expected goals rate for each side in this specific fixture.
3. Run the Poisson calculation for 0, 1, and 2 goals for each side.
4. Multiply matching pairs to estimate each specific score draw probability, then sum them for an overall draw likelihood.
5. Compare fixtures across your shortlist using this consistent method, rather than gut feeling alone.

A Grounded Conclusion

Even a rigorous calculation like this one still only produces a probability, not a certainty — a roughly 13% chance of a specific 1-1 draw means it fails to happen the overwhelming majority of the time. Use this method to compare fixtures against each other more rigorously, not to convince yourself any single match is a sure thing. Stake only what you’ve budgeted, treat the pools as entertainment for over-18s, and seek confidential support from UK gambling support services if staking ever feels like it’s getting out of hand.