Everyone talks about form, yet few people define it. In statistical terms, form is simply an estimate of how a team is performing right now, based on its latest matches rather than its long-term reputation. The challenge is deciding how far back to look.
Why recent matches count for more
Teams change. Players get injured and return, tactics evolve, new signings settle in, confidence rises and falls. A match played last weekend therefore tells us more about today's team than one played eight months ago. Predicta reflects this with recency weighting: every match counts, but the most recent ones carry a larger share of the weight than the older ones.
This is a compromise. Ignore old matches completely and you throw away useful information. Treat all matches equally and you describe a team that no longer exists.
Five matches, ten matches or the season
Predicta lets you choose the history window, and each option has a different personality.
- Last 5 matches: very reactive, and it captures a sudden change. It is also the noisiest, because five games is a tiny sample.
- Last 10 matches: a balanced middle ground. It smooths out a single freak result while still reflecting the current period.
- Current season: the most stable and the least sensitive to short-term swings. Early in the season it contains very few matches, though, and late in the season it may include a stretch that no longer applies.
There is no universally correct choice. A short window suits a team that has clearly changed, for instance after a new coach. A longer window suits a team whose results have been consistent.
The trap of the winning streak
Humans love patterns. Three wins in a row feel like momentum, and three defeats feel like a crisis. But football contains a great deal of luck, and a short run can be produced by chance alone. A team can win a match while creating little, or lose a match it dominated.
A better habit is to ask what the run is made of. Were the goals scored and conceded consistent with a real change, or was it one lucky deflection and a lone red card? Looking at the same team through two different windows is a quick sanity check: if 5 and 10 matches tell very different stories, treat the estimate with extra caution.
Key takeaways
- Form estimates current performance from recent matches, not reputation.
- Recency weighting gives newer matches more influence without discarding older ones.
- Short windows react fast but are noisy; long windows are steady but slow.
- A short streak can easily be luck, so avoid reading too much into it.