How Many Goals on Saturday?
Goals are rare, sudden and unpredictable — yet the number of goals per match follows a well documented pattern in statistics. Count last season, predict next weekend, check on Monday.
Where this idea comes from
Start here — this is the source that inspired this exploration.
Introduction
Nobody can say when a goal will come. But count the goals in every match of a season and a stubborn pattern appears: lots of 1s and 2s, occasional 0s and 3s, rare avalanches. Statisticians know this pattern well — it shows up wherever rare events happen at a steady average rate, from goals to lightning strikes to misprints. David Spiegelhalter used it to predict Premier League scorelines for Plus magazine. Your job: dig the pattern out of real results yourself, use it to predict a coming weekend, and then face the question every forecaster faces — were your misses bad luck, or a bad model?
Guiding Questions
- Collect a season of results for a league you follow and tally the goals per match: how many 0-goal games, 1-goal games, and so on. What's the average, and what does the shape look like?
- If goals were completely random in time but at that average rate, how often SHOULD each total happen? Work out predicted frequencies and lay them over your tally.
- Use your fitted pattern to predict next weekend's fixtures — most likely scoreline and the chance of a goalless draw — then check the results. How did you do?
- Your model treats every match alike. Decide which real difference between matches matters most, split your data accordingly, and test whether the pattern shifts the way you expected.
- When your predictions miss, how would you tell bad luck from a bad model? Design and run that test honestly.