No black boxes. Sloop captures Singapore Pools' odds, prices every market with two independent probability models, and shows you exactly where — and by how much — the book and the model disagree. Every pick is graded in public. Here's the whole pipeline.
Sloop prices every match twice, on purpose. Two independent views catch different kinds of mispricing — and you can see which one is behind each pick.
Fits its goals model to SG Pools' most-liquid lines (1X2, totals, both-teams-to-score), so it agrees with the book where the book is sharpest — then prices the markets the book leaves looser (correct score, handicaps, half-time). Value shows up in those thinner markets.
Market-blind.It builds expected goals purely from team Elo ratings and recent form, never looking at the line, then prices the whole board. When it disagrees with the book that's a real signal — but contrarian and lower-confidence, because the market aggregates far more information than ratings alone.
A single number for each nation's strength. Win and you gain points — more for beating a stronger side, by a bigger margin, in a bigger tournament; lose and you shed them. The WC plays on neutral ground, so no home-field bonus is applied.
Team strength becomes two expected-goal numbers (xG) — one per side. A Poisson distribution turns those into the probability of every exact scoreline, and from that grid we can price any market on the board.
Bookmaker odds always imply probabilities that add up to more than 100%. That overround is the margin — the house edge. Fair oddsare what's left after we strip it out: the book's true estimate, with no margin baked in.
A bet has value when the book pays more than the fair price — i.e. the model thinks the outcome is likelier than the odds imply. Expected value (EV) is the average profit per $1 staked if the model is right. Positive EV is the whole game.
The gap between the model's probability and the book's own no-vig probability. EV is that edge expressed as money; edge is it expressed as probability.
A formula that sizes each stake by how big the edge is and how short the odds are — bet more when the edge is large. We stake a fractionof full Kelly so one bad run doesn't wreck the bankroll.
Did you get a better price than the market settled on at kickoff? Consistently beating the closing line is the single best predictor of long-term betting skill — so we track it on every pick.
We grade ourselves in the open. Each model's edges are placed as virtual bets, settled when matches finish, and tracked for P&L, ROI and hit-rate — so you can judge the system on results, not promises.
The predicted score is the single most likely exact scoreline — but in a low-scoring sport even that is unlikely on its own. The result probabilities (1X2) sum many scorelines each, so they can tell a different story. Take South Korea vs Czech Republic (xG 1.63–1.30):
Why? A South Korea win is spread across many scorelines (2-1, 1-0, 2-0…), none individually as tall as 1-1. So treat the predicted score as flavour — the 1X2 line drives the actual bet.
The same match, all the way through the pipeline — with the real numbers from the model.
Sloop's fair odds are model estimates, not advice or guarantees. The market is sharp and will often be right when it disagrees with us — that's why every edge is paper-traded and graded in the open on the Paper page. Bet within your means, and only ever bet what you can afford to lose.
