Both teams to score looks like the easiest question in football. Two attacks, two defences, ninety minutes: surely the teams that score a lot and leak a lot produce yes, and the tight, toothless ones produce no. Every pundit segment and every stats graphic treats it that way.
We ran the question through ten thousand matches. The answer is stranger than the intuition.
The test
We took 9,982 finished fixtures from our own corpus — fifteen leagues, three seasons, every match where both sides had a full data trail. For each fixture we froze what was knowable before kick-off: each team's chance creation and chance concession over its previous ten games. No hindsight, no leakage — the exact tape a reader of the app would have seen that morning.
Then we asked: which pre-match signal actually separates the matches where both teams scored from the matches where they didn't? Four candidates, the obvious ones: how good the weaker attack was; how good the stronger attack was; how tight the better defence was; how leaky the worse defence was.
The result: everything is nearly flat
Across all 9,982 matches, both teams scored 55.6% of the time. Sort the matches by any of the four signals and look at the extremes: the strongest signal in football's most argued-about market — the weaker attack's quality — moves the outcome by nine percentage points across its entire range. From "slightly worse than a coin flip" to "slightly better than a coin flip". Everything else moves it less.
Stack the extremes on top of each other and it barely improves:
Matches pairing a bottom-third weak attack with a mean-tight defence produced BTTS 47.4% of the time; matches pairing a top-third weak attack with a leaky defence, 61.7%. The full sweep of everything the tape can tell you, combined, spans fourteen points around a coin flip.
For comparison: in our earlier ten-game study, a team's chance creation carried into its next ten games with a correlation of 0.63. The BTTS outcome correlates with its best pre-match predictor at 0.07.
Why the tape can't grip it
The reason is arithmetic, not mystery.
BTTS asks the lowest question in football: one goal each. Not two, not "who wins" — one. A genuinely weak attack, one expecting a single goal's worth of chances in a match, still scores at least once about two times in three. That's how Poisson arithmetic works at the bottom of the scale: the difference between a poor attack and an average one is enormous over a season and almost invisible inside the question "will they get one tonight?"
Nearly every professional team lives on the flat part of that curve. The tape differences that decide league tables get compressed into a few percentage points the moment you ask the one-goal question. What's left is dominated by the thing no tape can hold: the variance of a single goal — a deflection, a set piece, a keeper's afternoon.
How to read a BTTS probability
This is why, in Broovo and anywhere else honest, a both-teams-to-score probability almost always sits between roughly 0.45 and 0.65. That's not a model hedging its bets — it's what the flat chart above looks like when it's translated into a single number. A market whose outcome lives close to the coin flip produces probabilities that live close to the coin flip.
So the useful skill isn't hunting for a big number — big numbers structurally don't exist here. It's reading the small ones correctly:
- 0.50 means genuinely nothing. Not "the model can't decide" — the data has decided, and it says this match is a coin.
- 0.55 is a whisper — one signal leaning gently.
- 0.60+ is as loud as this market gets — and now you know what it takes to get there: the weekend's strongest weak attack meeting its leakiest defence, every slow signal aligned the same way. A few points from the flip is the ceiling, because in ten thousand matches, nothing ever pushed past it.
Read this way, a board full of 0.5x stops looking indecisive and starts looking like what it is: an honest map of a market where one goal each is a low bar, and variance clears low bars.
Related reading
What is xG? — the metric all of this is built on. Two titles, two different teams — where the tape does grip: chance creation, and what persists.
Method: 9,982 fixtures across the 15 leagues in Broovo's corpus, 2023–2026. Pre-match profiles are each team's previous ten matches with full data, frozen strictly before kick-off. BTTS = both sides scoring at least once in ninety minutes. For information only — not betting advice. 18+
