How picks are made

1. Finding candidates. Three times a day, around 9:00, 15:00 and 21:00 Finnish time, we search public sources for matches starting at least one hour after publication and before 9:00 the next morning – so every pick settles by the next morning. Promising matches further out are re-analysed on match day. The evening edition covers the North American leagues and late European matches (starting at 21:45 Finnish time or later), whose line-ups are already confirmed. Typically 10–12 candidates across several sports. We only consider competitions with reliable line-up and statistics coverage – top divisions and the tier below, not lower leagues or challenger events.

2. Analysis. Each candidate is analysed separately: statistics, line-ups, schedule congestion and sport-specific mandatory checks – for example the starting goaltender in ice hockey and load management in basketball. Figures are quoted only when found in a source. If a mandatory check cannot be completed, the pick is rejected.

3. Probability estimate. We assign a probability to every outcome, summing to 100 %. The starting point is the market's odds with the bookmaker's margin removed; we deviate only for a named, sourced reason. Without a reason our estimate equals the market's – and the pick is not a value pick. A match is proposed as a value pick only when there is a fresh, named reason to doubt the market price – for example a confirmed key absence or a goaltender decision.

4. Edge and filtering. Edge = our estimate × odds − 1. Value picks are published only when the edge is positive; implausibly large edges are rejected because they are more often an error than a find. Likely picks are favourites at odds 1.30–1.75 that we rate at least as likely to win as the market does – their edge is usually near zero or negative. At long odds we require stronger reasons, and value picks are capped at odds 2.60. Each lane carries at most three picks; on days with a large match supply the cap rises to five. On some days the edition is short or empty: that means we thought the market was right. The administrator may also add a pick the model rejected – these are labelled "Sorceress's pick" and tracked separately. Specials are a fourth lane: an individual player's statistic or a special market – points, goals, assists, shots, power-play points or saves in the NHL; points, goals, assists or shots in Liiga; points, rebounds, assists or threes in the NBA. We publish only markets that are settled from the official box score. A pick is published only with a confirmed line-up and voided if the player does not play. Odds are 1.60–2.20, our estimate may differ from the market by at most ten percentage points, and one pick per player and statistic. At most three picks in the afternoon and overnight editions.

Over/under is a third lane: we call the total goals or points in ice hockey, basketball and football. Odds are 1.70–2.20 and our estimate may differ from the market by at most eight percentage points. In ice hockey overtime and the shootout count; in football only regulation time. The same lane carries spreads in basketball and American football: we call the winning margin – Lakers −4.5 wins if the Lakers win by five or more. Half-point lines only, with the same thresholds as totals. At most three picks per edition.

5. Tracking. Every pick is settled the next morning and the statistics are public: hit rate, flat-stake singles return (a voided pick counts as odds 1.00) and a breakdown by lane. Hit rate and return are calculated over all published picks; editor's picks are also shown separately. Last 10 means the ten most recently played settled picks. Internally we also track calibration – whether, say, our 60 % estimates win about 60 % of the time. Closing line. We record each pick's odds at publication and again just before kickoff. If the odds shortened – the market moved toward our pick – the pick 'beat the closing line'. This is the key metric professionals use: it shows whether we find mispricings before the market does, independent of single-match luck. Our reference is Pinnacle, whose odds are considered the sharpest.

6. What this is not. Editions are opinions, not promises or advice. Betting always carries the risk of losing your stake, and past results do not guarantee future ones. The service is for people aged 18 or over. If gambling feels out of control, free help is available from Peluuri: peluuri.fi, 0800 100 101.