Half-Time / Full-Time Analysis: A Two-Phase Goal Data Structure
Half-time / full-time analysis splits a football match into two phases — the first half and the second — and assigns a probability to each of the nine possible combinations of half-time leader and full-time result (1/1, X/1, 2/1 … 2/2). The nine scenarios always sum to 100%, and even the tallest cell — typically home-leads-and-wins — rarely clears 20-28% in a balanced match, while comeback cells like 1/2 and 2/1 usually sit at just 2-4%. This page explains how 11Stat derives the grid from per-phase goal expectancy, how lead transitions read as data, and where the distribution's limits lie. It is football data analytics, not betting advice.
What half-time / full-time analysis actually is
A match result is not a single line — it is the composition of two phases. Half-time / full-time analysis cuts the match at the interval and asks two questions at once: who leads at half-time, and what is the final result? Crossing the three interval states (home ahead / level / away ahead) with the three full-time outcomes yields a 3×3 grid of nine scenarios: 1/1, 1/X, 1/2, X/1, X/X, X/2, 2/1, 2/X, 2/2. The nine cells sum to 100% at every instant.
The critical nuance: even the peak cell is modest. In a balanced match 1/1 typically carries 20-28% and X/X 8-12%, while the comeback cells are squeezed to 2-4%. The "most likely scenario" therefore usually does not happen — the grid is a snapshot of a distribution, not a verdict. The underlying vocabulary lives in the glossary.
Two-phase goal expectancy: first-half vs second-half rates
Goals are not spread evenly across 90 minutes: in the major leagues roughly 44-45% of goals arrive in the first half and 55-56% in the second. Fatigue, score pressure and substitutions systematically raise second-half tempo. Splitting a single 90-minute xG figure in half misses that asymmetry; the correct structure gives each phase its own goal expectancy.
11Stat decomposes each team's goal production into per-phase rates and builds a separate Poisson matrix for each half. The Dixon-Coles correction, which reweights low-scoring cells, matters even more in the first-half matrix, because 0-0 is football's most common interval state and pure Poisson shows systematic bias exactly there.
Building the nine-scenario grid from two matrices
The first-half matrix delivers the distribution of interval states: home ahead, level, away ahead. A conditional second-half matrix is then attached to each interval state, and the composition of the two phases is played out thousands of times by Monte Carlo simulation. A scenario like X/1 is the probability of the compound event "the half ends level and the home side moves ahead after the break to win".
The output is a 3×3 grid whose rows are the half-time state and whose columns are the full-time result. Derivative readings — the probability of 0-0 at the break, or of more goals after the interval — come from the same grid, and they must stay structurally consistent with the score matrix: that coherence is the test that the model does not contradict itself.
Lead transitions: how often a half-time lead survives
The historical data is blunt: in the major leagues, the team leading at half-time goes on to win roughly three matches in four; holding for a draw and losing from ahead share the remaining quarter. When the half ends level, all three results stay live — roughly 35-40% of matches reach the interval level, which is why X/1 and X/2 are the most analytically interesting cells in balanced fixtures.
The comeback cells (1/2 and 2/1) sit in the 2-4% band in most leagues: the trailing side must both equalise and move ahead within 45 minutes — two conditional events stacked. Their smallness is not a model weakness; it is football's structure. Once the whistle blows, the grid hands over to the minute-by-minute live win probability.
How 11Stat reads the grid on an analysis card
At 11Stat the phase distribution is not the monopoly of a single engine: the Poisson/Dixon-Coles core engine, the Elo-form engine and the advanced bivariate engine each produce their own distributions, and the scenario reading on an analysis card is derived from their consensus, then calibrated against historical performance.
The output deliberately declares no single scenario; it shows the 2-3 most probable scenarios with their percentages. When the gap between listed scenarios is small, that is an honest signal the match is genuinely uncertain at scenario level — the job of analysis is to display uncertainty, not to hide it.
Why "certain HT/FT" claims collapse under the data
HT/FT is one of football's highest-variance compound outputs: the intersection of two separate phases carries far wider uncertainty than a single match result. Using certainty language over a distribution whose peak hovers around 25% is mathematically indefensible — in a nine-cell structure, "some other scenario" is always far more probable than the top one.
The honest quality test is calibration: do scenarios assigned 10% occur roughly one time in ten over the long run? 11Stat holds itself to that yardstick and uses no certainty language. The HT/FT grid is an analytical instrument, not a promise of a result. Analytical model — no guarantee of outcomes. This is not a betting service.
Frequently Asked Questions
Is half-time / full-time analysis a betting tip?
No. It is data analytics that reports the probability distribution across nine scenarios; no cell is a recommendation, a steer or a guarantee. 11Stat provides no betting service.
Which HT/FT scenario is the most common?
In most leagues 1/1 (home leads at the break and wins) is the tallest cell at roughly 20-28%, followed by X/X at typically 8-12%. Values shift with the league and the strength balance of the fixture.
Why are comeback scenarios (1/2 and 2/1) so rare?
Because the side trailing at the interval must both equalise and move ahead within 45 minutes — a composition of two conditional events. Historical frequency sits around 2-4% in most leagues; rare, but never zero.
Are more goals really scored in the second half?
Yes. Roughly 55-56% of goals in the major leagues arrive after the break; fatigue, score pressure and substitutions raise the tempo. That asymmetry is the entire reason per-phase goal expectancy exists.
How does the HT/FT grid differ from live win probability?
The grid is a nine-scenario distribution computed before kick-off; live win probability is the home/draw/away trio updated minute by minute with score, time and live xG while the match is played. One is a pre-match map, the other an in-play compass.
If the most likely HT/FT scenario doesn't happen, was the model wrong?
No. A 25% scenario failing to occur is the expected outcome three times out of four. The correct yardstick is long-run calibration: stated probabilities should match observed frequencies, not single-match hits.