How accurate is the independence approximation for BTTS?

It is a useful approximation but not exact. Goals are negatively correlated because teams adjust tactics based on the score. The model tends to overestimate BTTS in matches where one team is likely to keep a clean sheet and underestimate in high-scoring, open games.

Can I use team Over 0.5 odds directly to estimate BTTS?

You can, but only after removing the bookmaker's margin. The implied probabilities from odds sum to over 100%, so you must normalize them to get fair probabilities. Then multiply the two fair probabilities to estimate BTTS.

Why does the 1–0 scoreline matter for BTTS betting?

Because 1–0 is the most common football score, occurring in roughly 25–30% of matches. This means that in a large fraction of games, only one team scores, violating the independence assumption and making the simple product model less reliable.

Both Teams to Score (BTTS) explained

Updated on 2026-07-27 · 814 words

What it is

Both Teams to Score (BTTS) is a bet on whether both teams will score at least one goal in a match. The outcome is binary: yes (both score) or no (at least one fails to score). It is one of the most popular markets because it depends on the flow of the match and does not require picking a winner.

From a probability perspective, BTTS is a compound event: it happens when the home team scores and the away team scores. The chance of both occurring is the product of each team’s probability of scoring, assuming independence. In reality, goals are not independent — a team that concedes may push forward, increasing its own scoring chance — but the product model is a reasonable first approximation and widely used by bettors.

The math

Let p(H) be the probability that the home team scores at least once, and p(A) that the away team scores at least once. The estimated probability of BTTS is:

p(BTTS) ≈ p(H) × p(A)

To estimate p(H) and p(A) from odds, you can use the team’s odds of scoring over 0.5 goals. For example, if the home team has odds of 1.36 for Over 0.5 goals, the implied probability (before margin) is 1 ÷ 1.36 = 0.735 (73.5%). Similarly, if the away team’s Over 0.5 odds are 1.55, implied probability is 1 ÷ 1.55 = 0.645 (64.5%).

If the sum of implied probabilities for Over and Under 0.5 exceeds 100%, you must remove the margin to get fair probabilities. Suppose the margin is 5% on each market. After normalizing, p(H) = 73.5% ÷ 105% ≈ 70.0%, and p(A) = 64.5% ÷ 105% ≈ 61.4%.

Then p(BTTS) ≈ 0.70 × 0.614 ≈ 0.430, or 43.0%.

The corresponding fair odds for BTTS Yes would be 1 ÷ 0.43 ≈ 2.33 (American +133). If the bookmaker offers odds above that, the bet has positive expected value (before margin). In practice, the bookmaker’s odds for BTTS Yes already include their margin, so you must compare with the line after removing margin.

Worked example

A match between two mid-table teams in the English Premier League. You estimate each team’s chance of scoring starts at about 65%. Using the independence formula:

p(H) ≈ 0.65, p(A) ≈ 0.65
p(BTTS) ≈ 0.65 × 0.65 = 0.4225 (42.25%)
Fair decimal odds: 1 ÷ 0.4225 ≈ 2.37 (American +137)

If a bookmaker offers BTTS Yes at 2.50 (American +150), the implied probability including margin is 1 ÷ 2.50 = 0.40 (40%). After estimating the margin at about 5% for this market, the fair probability is 40% ÷ 1.05 ≈ 38.1%. Your estimate of 42.25% is higher, indicating a potential value bet. But remember that the actual probability depends on the accuracy of your team scoring estimates and the independence assumption.

To check, you could compute expected value: (0.4225 × 2.50) – 1 = 1.056 – 1 = +0.056, or +5.6%. That is a positive EV, but the margin means you need to be confident in your inputs.

When not to use it

Goals are not independent events. The product formula assumes that home and away scoring are unrelated, but in reality they are negatively correlated. When one team scores, the other often pushes forward, increasing its own scoring chance, but also the opponent may sit back. This correlation is strongest in low-scoring matches. The model underestimates BTTS probability in open games and overestimates in tight, defensive matches.
Scoreline 1–0 is the most common result. Many matches end with only one team scoring. Historical data shows that around 25–30% of matches finish 1–0, which directly contradicts the independence assumption. The model tends to overestimate BTTS because it ignores the possibility of both defenses holding firm simultaneously.
Early red card or injury breaks the model. A sending off after 20 minutes drastically changes both teams’ scoring probabilities. The independent estimates derived from pre-match odds no longer apply. The product formula becomes unreliable because the conditional probabilities shift in ways the simple multiplication cannot capture.
Team estimates are hard to get right. p(H) and p(A) themselves are uncertain. Using team Over 0.5 odds from the bookmaker simply recovers their margin—you need your own independent assessment. Without a robust model for team scoring probability, the BTTS estimate is only as good as those inputs, and errors compound multiplicatively.
Sample size for validation is small. Testing a BTTS model requires hundreds of matches. Short-run variance can make a bad model look good. Even with 100 matches, the margin of error on BTTS hit rate is around ±9%, making it hard to distinguish skill from luck.

For a precise comparison of your BTTS estimate against the market odds, use the bet analyzer tool.

Gamble responsibly. Must be 21+. Risk of financial loss. This content is educational and does not guarantee profits.

Frequently asked questions

How accurate is the independence approximation for BTTS?: It is a useful approximation but not exact. Goals are negatively correlated because teams adjust tactics based on the score. The model tends to overestimate BTTS in matches where one team is likely to keep a clean sheet and underestimate in high-scoring, open games.
Can I use team Over 0.5 odds directly to estimate BTTS?: You can, but only after removing the bookmaker's margin. The implied probabilities from odds sum to over 100%, so you must normalize them to get fair probabilities. Then multiply the two fair probabilities to estimate BTTS.
Why does the 1–0 scoreline matter for BTTS betting?: Because 1–0 is the most common football score, occurring in roughly 25–30% of matches. This means that in a large fraction of games, only one team scores, violating the independence assumption and making the simple product model less reliable.