Both Teams to Score: Turning Expected Goals Into a BTTS Price

What counts and what does not

Settlement rules for a both teams to score bet are short, but the edge cases catch people out.

The own-goal line matters more than it looks. If a defender turns the ball into his own net, the opposing side is on the scoresheet, and a yes bet that needed that team to score is still alive. The shoot-out line matters in cup ties: a 0–0 after 90 minutes that goes to penalties is a winning no bet, whatever happens from the spot.

From expected goals to a probability

The standard way to estimate the chance of each team scoring is the Poisson distribution, the same model behind our goals model tool. It assumes goals arrive at random at a steady average rate. Under that assumption, the chance that a team with expected goals λ scores nothing is e^−λ, so the chance that it scores at least once is 1 − e^−λ.

If you treat the two teams' scoring as independent, the BTTS yes probability is the product of the two:

P(BTTS yes) = (1 − e^−λ1) × (1 − e^−λ2)

where λ1 and λ2 are the expected goals for the home and away side in this match. BTTS no is simply 1 minus that figure.

A worked example

Say your numbers for a match are 1.45 expected goals for the home side and 1.20 for the away side. These are illustrative inputs, not data from a real fixture.

1. Home side fails to score: e^−1.45 = 0.2346. So it scores at least once with probability 1 − 0.2346 = 0.7654, or 76.5%.
2. Away side fails to score: e^−1.20 = 0.3012. So it scores at least once with probability 1 − 0.3012 = 0.6988, or 69.9%.
3. Both score: 0.7654 × 0.6988 = 0.5349, or 53.5%.
4. Fair price for yes: 1 ÷ 0.5349 = 1.87.
5. Fair price for no: 1 ÷ (1 − 0.5349) = 1 ÷ 0.4651 = 2.15.

The fair price is the point where a bet breaks even in the long run. It contains no margin, so a sportsbook will always quote something shorter on at least one side.

Comparing the model with the market price

Suppose the expected price on offer for both teams to score is yes 1.80 and no 2.00. Convert each to an implied probability: 1 ÷ 1.80 = 55.6% and 1 ÷ 2.00 = 50.0%. Together they add up to 105.6%. The extra 5.6 points is the margin, the sportsbook's built-in edge on this market. You can check any pair of prices this way with the odds converter.

To see what the market really thinks, scale the two figures back to 100%: 55.6 ÷ 105.6 = 52.6% for yes and 47.4% for no. So the market, once its margin is removed, rates BTTS yes slightly lower than our model's 53.5%.

That gap is not enough. The price you can take is 1.80, not the margin-free figure, and our model says yes is worth 1.87. The expected value of a £10 stake at 1.80 is 10 × (0.5349 × 1.80 − 1) = −£0.37. The no side is worse: 10 × (0.4651 × 2.00 − 1) = −£0.70. Neither bet is worth placing on these inputs.

A sensible rule is to want more than the fair price, not just the fair price, because your expected goals are estimates. If you want a 5% cushion over 1.87, you would look for about 1.96 before betting yes. The best odds betting sites page compares where margins tend to be thinner.

How the probability shifts with the inputs

The formula has a property worth knowing: the weaker attack drags the whole figure down. Lower the inputs to 0.9 and 0.8 expected goals and the calculation becomes (1 − e^−0.9) × (1 − e^−0.8) = 0.5934 × 0.5507 = 0.3268, a 32.7% chance and a fair price of 3.06 for yes. Two sides who rarely concede, or one side who rarely scores, push the market strongly towards no.

The same logic explains first-half BTTS. If you assume goals arrive at an even rate across the match, each side's first-half expected goals is half its full-match figure. With 0.725 and 0.60, the chance of both scoring before half-time is (1 − e^−0.725) × (1 − e^−0.60) = 23.3%, a fair price of 4.30. The even split is a simplification; treat the result as a starting point.

Both teams to score with the result or the total

Many sites that offer it list combined markets: BTTS and match result, or BTTS and over/under 2.5 goals. These are where the Poisson model is most useful, because the parts are linked and cannot simply be multiplied.

Using the same 1.45 and 1.20 inputs and summing every scoreline the model produces:

The first three rows add back to 53.5%, as they should. The last two do the same: 41.2% plus 12.3%.

Now look at what happens if you treat the legs as independent. BTTS yes is 53.5% and over 2.5 is 49.4% in this model, so multiplying gives 26.4%, a price of 3.78. The true joint chance is 41.2%. The two outcomes move together: in this model 41.2 of the 53.5 points come from scorelines with three or more goals, and the only BTTS result under 2.5 is 1–1. That correlation is why sites that offer a bet builder price linked legs as one bet rather than letting you multiply them.

The home-win combination goes the other way. Multiplying 53.5% by the 42.9% home-win chance gives 22.9%, a price of 4.36, but the joint figure is 19.8%, a fair 5.05. A home win in which the away side also scores needs the home team to score at least twice, which is harder than simply winning, so the two outcomes partly work against each other.

Where the model goes wrong

Poisson treats each team's goals as independent events. Real matches are not like that. A side that falls behind attacks more, which raises its chance of scoring and its chance of conceding a second. A side protecting a lead may sit deep. For both teams to score, these game-state effects tend to make the simple model miss on the tight, low-scoring scorelines that decide a BTTS no bet.

Three practical checks before you trust the number:

• Team news. A missing striker or first-choice goalkeeper changes λ more than most bettors adjust for.
• Motivation. A side that only needs a draw plays differently from one that needs a win.
• Sample size. Expected goals from a handful of matches are noisy. A model is only as good as the inputs, and small samples swing.

For the goals-total side of the same question, see the over 2.5 goals guide. For individual scorelines, the correct score guide goes further into the grid.

BTTS in-play and with document checks

Prices for both teams to score change fast once a match starts. The first goal moves the market sharply, because from then on only the side still on zero needs to score. If you bet in-play, the same formula works with the expected goals remaining for each side, which shrink as the clock runs down. The live betting guide covers the timing.

The market itself has no link to identity checks, but winnings do. Sites that let you register without documents can still ask for them later, most often before a large withdrawal or when account activity looks unusual. A steady BTTS bettor making many small stakes rarely meets those triggers; a single big win can. The fast withdrawal comparison sets out what each site's terms say about payouts and checks. Set your stakes before the match, not after a goal, and see responsible gambling if betting stops feeling like a choice.

Our ranking of no verification sportsbooks shows which operators name a sum for ID checks and which leave it open.

Handicaps, live betting and cash out are explained in the other betting guides.

Bitcasino.ioSportsbook

US$2,500When ID can be asked

Open