How the score works
The watchability score predicts how entertaining a match is likely to be. It is not a prediction of the result.
What the number means
Every match gets a score from 0 to 10. A higher score means we expect more goals, more drama, or more at stake. A 9 does not mean the favourite wins, and a 3 does not mean a bad game: it means we expect a match with less to hold the attention.
The colour of the score follows three bands: 7 and above is high, 4 to under 7 is mid, and under 4 is low. Most matches land between 3 and 6, so a 9 or a 10 is rare but does happen.
The six signals
Each signal looks at one part of a match and scores it from 0 to 1. The score starts as the weighted average of the signals, then we stretch that average onto 0 to 10 so a season's matches use the whole scale. The stretch is fixed and the same for every match, so it does not change the order of matches. The percentage next to each signal is its weight in that average. Up to three reasons are shown per match, taken from the signals that scored highest.
Goal form 25% of score
How many goals the two teams have recently been involved in, scored and conceded together. We average the goals in each team's last five matches. Around 1.5 goals a match or fewer scores nothing; 4 or more scores full marks.
Example reason: Both teams average 3.1 goals a game
Drama form 15% of score
How dramatic each team's last five matches have been. A match earns credit for a comeback (losing at half time but drawing or winning at full time), for a big change in the score after half time (two goals or more), and for going to extra time or a penalty shootout. Both teams are averaged together.
Example reason: City came from behind in 2 of their last 5
Closeness 15% of score
How tightly the two teams are bunched in the league table. The gap in points counts for 60% and the gap in places for 40%. Teams level on points score full marks, and teams six or more points apart, or four or more places apart, score nothing. Cup matches have no table and skip this signal.
Example reason: 2 points apart
Quality 15% of score
How good both teams are, judged by their recent results and their league position. Recent results are points per game over each team's last ten finished matches, in any competition, including last season. A win in every match scores full marks. Early in the season, when a team has played few league games, its results count for most of the rating; by half a season the league position counts for half. A team with fewer than five recent results is rated by its league position alone. The weaker of the two teams sets the score, because a match is only as good as its weaker side. A reason appears only when both teams average at least two points a game.
Example reason: 2.4 vs 2.1 points per game
Stakes 20% of score
How much the match matters. In a league, stakes are zero until halfway through the season and then rise to full value on the last matchday. A clash between two top-six sides is the exception: it always carries some stakes. Full value needs both teams still able to reach the title, a relegation place or a European place. Knockout rounds have fixed values, so a final counts for the most and early rounds for less. The last group matchday of a group stage counts for a moderate amount.
Example reason: Title race
History 10% of score
How entertaining the two teams' previous meetings were. High-scoring games, recent comebacks and extra time all count for more. This signal is skipped when we have no head-to-head data for the pair, and the other signals carry the score.
Example reason: Last meeting: City won 4-3
If a signal cannot be measured, it is left out and the other weights are scaled up to fill the gap. For example, closeness is skipped for cup matches, which have no league table, and history is skipped when the two teams have no previous meetings in our data.
Where the data comes from
Fixtures, results and league tables come from football-data.org. Current seasons are refreshed every 6 hours, so a table or a recent result can be up to six hours behind.
How we measure actual drama
Actual drama is how we judge a finished match after the fact, on the track record and the rewatch picks. It is the sum of four parts:
- Goals: each goal scores 1 point.
- A comeback: 2 points, when a side that was behind at half time does not lose at full time.
- Extra time or a penalty shootout: 2 points.
- A close finish: a draw, or a win by 1 goal or fewer, scores 1 point.
A goalless draw scores 1 point; every goal adds to the total, so there is no upper limit. This is a different measure from drama form, which is a pre-match signal built from each team's recent matches.
Known limits
- No shots, expected goals (xG) or possession data. We see scores, half-time scores and tables, so a goalless draw full of chances looks the same as a dull one.
- Early season is uncertain. Goal form and drama form need at least three recent results per team, quality uses league position alone until a team has five, tables mean little after a few games, and stakes stay at zero until halfway through a league season, except a clash between two top-six sides, which always carries some stakes.
- History is often missing. Our data source only knows meetings in the competitions we follow, so most Champions League ties and many Championship matches have no head-to-head data, and the other signals carry the score.
- No squad, injury, or manager information. Team strength is judged from results and league position only.
- The score is a prediction. A 9.0 match can still end 1-0 and be a poor watch on the day.