EPA is a concept that originated in American football in the 1980s and aims to provide a more nuanced and objective view of how individual goals contribute to a team’s eventual point total.

It is entrenched in the idea that not all goals have the same impact on the outcome of a match. Scoring a decisive goal two minutes before the end of a tied game typically turns the scorer into an instant fan hero, since such a goal is highly likely to have secured two additional points for the team. On the other hand, if a player scores his team’s fifth goal in the same minute to make it 5–0, he likely won’t receive much acclaim, since the three-point gain had already been sealed.

Using a database of Czech First League matches and the sequences of goals scored, we built a probabilistic model based on occurrences between 1994 and 2017 that takes into account:

– the match score at the time the goal was scored,
– which team did the scoring (at home or on the road), and
– the minute in which the goal was scored.

Here’s a specific example to make the concept easier to grasp:

Imagine the 67th minute of a match has just ended, and the score is level. By checking our match database, we find that this scenario had occurred 1,820 times since the 1994/95 season (as of 2016/17 – note of CF). Based on the final outcomes of those matches, we can calculate the statistical probabilities of each possible result and, from that, the expected number of points for each team.

In this case, the home team won 526 times (28.9%), the match ended in a draw 1,022 times (56.2%), and the away team won 272 times (14.9%). Therefore, the Expected Points (EP) for the home team is:

3 × 0.289 + 1 × 0.562 + 0 × 0.149 = 1.429 points.

Now suppose the home team scores in the 67th minute, taking a one-goal lead. Performing the same calculation for this new scenario (home team leading after 67 minutes), we find the home team went on to win in 1,111 cases (80.6%), the match ended in a draw 227 times (16.5%), and the away team won 40 times (2.9%). The expected number of points for the home team:

3 × 0.806 + 1 × 0.165 + 0 × 0.029 = 2.583 points.

The ultimate EPA value of the goal scored in the 67th minute then simply constitutes the difference between these two values above, added to the scorer who’s signed underneath it:

2.583 – 1.429 = 1.154 EPA.

So, based on past data from the Czech top flight, a goal scored by the home team in the 67th minute of a tied game brings an additional 1.154 expected points. For comparison: if the goal had extended a lead from 2:0 to 3:0 in the exact same minute, its EPA value would have been just 0.059 — a significantly lower contribution to the points total. This is intuitive: a two-goal lead with 23 minutes left is rarely lost, so the third goal adds little in terms of securing points.