How we make predictions
Every pick combines a statistical model with manual analysis. Here is what stands behind the numbers on a card and how we track our own results.
Four steps to a pick
- 01DataTeam and player stats, form, xG, fixture schedule, team news.
- 02ModelEstimates outcome probabilities and compares them with market odds.
- 03Manual analysisWe check what the model can't see: motivation, rotation, tactics.
- 04PublishingThe card goes to Telegram and Threads, then to the site. The result goes to the archive.
What the numbers on a card mean
How likely the main pick is according to the model. 60% means that in similar matches this outcome happens about 6 times out of 10. It is not a guarantee.
How solid the data behind the pick is: whether the model and manual analysis agree and whether there are unknowns in the line-ups. High probability with low confidence is a reason to be careful.
The chance that the match goes completely off script: derbies, squad rotation, weather, motivation.
Probabilities of a home win, a draw and an away win. They always add up to 100%. The outcome the pick points to is highlighted.
Badges
How we track our own results
- We delete nothing. Every published pick stays in the archive with its result.
- Odds are recorded at the moment of publishing — [odds source].
- Accuracy is the share of winning picks among settled ones. Void picks (postponed matches) are excluded.
- ROI is the profit at the same 1-unit stake on every pick, divided by the number of bets.
- Past results do not guarantee future ones.
Where the data comes from
[List your sources: match statistics provider, odds source and when you record odds, team news sources. The more specific, the more trust.]
FAQ
We only publish where the model sees an edge over the market. Better to skip a match than post a weak pick.
About the project and contact
[Who is behind yarmo — 2–3 sentences: experience, why you started, what matters to you.]
We do not accept bets and are not a bookmaker. Content is for information only.