Skip to content

Hold the lead

Two goals up at half-time. Safe? "Two-nil is the most dangerous lead," they say. Set the situation and see how often a lead like this actually ends in a win.

Four minutes of stoppage time are assumed. "Chasing the game" multiplies the trailing side's scoring rate; 1 means no change.

Try an example

From a , the leading side:

Wins
Draws
Loses
Chance of winning, by the minute the lead is held from

The faint vertical line marks the minute set above.

How it works

From here, each side's goals in the time that's left follow a Poisson distribution, with its expected goals scaled down to the minutes remaining:

$$\lambda_{\text{left}} = \lambda_{90} \times \frac{\text{minutes left}}{90}$$

The lead holds if the leading side's goals from here, plus the lead, beat the trailing side's goals from here. Adding up every combination gives the chances of a win, a draw and a defeat.

In plain football

  • \(\lambda_{90}\) is the goals a side would expect over a whole match.
  • \(\lambda_{\text{left}}\) is what's left of that: a side expecting 1.2 goals has only 0.45 left with 34 minutes to play.
  • Time is on the leader's side. Every minute that passes shrinks what the trailing team has left to work with.

So is 2–0 the most dangerous lead?

Not in this model. Try the half-time examples: a 2–0 lead is won far more often than a 1–0 lead, and on the chart the two-goal line sits above the one-goal line at every minute. The saying probably comes from what happens when a 2–0 lead is cut to 2–1: the momentum feels like it has swung, and those comebacks are memorable because they are rare. Real matches agree: over 10,000 Scottish league games, sides 2–0 up at half-time won 92% of the time. The Myth or Maths investigation has the full numbers.

Where it goes wrong

  • Scoring rates are treated as steady, but more goals come late in matches, as the Exponential distribution article shows.
  • A leading side often sits deeper and a trailing side pushes on. "Chasing the game" lets you test that, but the right multiplier isn't known here.
  • Red cards, injuries and substitutions change the rates mid-match; one number per side can't.

These probabilities come from a statistical model and are for analysis and education. Football remains uncertain and model predictions will frequently be wrong.