Likelihood, price and value answer different questions

01

Compare like-for-like markets

Keep the fixture, regulation-time rules, outcome and exact line together. Home win, draw and away win are three outcomes in a standard soccer 1X2 market. Removing the draw or comparing a qualification market changes the contract. Team goals and match goals are also different markets.

An Asian Handicap line is stored relative to the home team; the away display reverses its sign. Pushes, quarter-line splits, draw-no-bet refunds and card-counting rules need their own settlement treatment. The current daily EV screener does not rank Asian Handicap pushes or quarter lines. Do not round one into a supported half-line.

02

Separate likelihood from the offered price

For decimal odds d, raw implied probability is 1 / d. It includes the bookmaker margin and is not automatically a fair forecast. For an estimated selection probability p, simple model-implied net EV per unit staked is p × d − 1 before configured costs. The model’s break-even decimal price is 1 / p.

The screener also permits explicit commission and price-buffer assumptions. It applies commission to winnings and subtracts the configured decimal-price buffer before calculating EV. These are your assumptions; bookmaker fees, stake limits, liquidity and executable availability are not verified.

03

Keep bookmaker margin and consensus visible

A complete outcome set must belong to the same bookmaker, source and contract. For 1X2, sum 1 / home odds, 1 / draw odds and 1 / away odds to see the implied total. A total above 100% is overround. A simple no-vig normalization divides each implied probability by that same set’s total; it is a margin-removal convention, not proof of the true probabilities.

FootyDesk labels bookmaker-reference calculations separately from independent forecasts. Reference-book controls let you inspect weights and required books. The offered book is excluded from its reference comparison. Agreement between a market-informed blend and market consensus is partly built into the inputs; it should not be presented as independent model confirmation.

04

Use the screener as a shortlist

Start from a daily date/competition scope or carry one exact selection from Match Research. The single-match label makes that scope explicit; “View full daily slate” removes it deliberately. Menus offer computable combinations, and each tool still checks its own source freshness and history requirements.

Choose the relevant model, then set minimum odds, quote age, history, disagreement and maximum selections per fixture. Compare the distinct likelihood, value and evidence rankings. Broad high-probability totals should not dominate merely because they happen often. Read the exclusions and review queue before changing a filter.

Open a row to inspect its source/model record, calculation and same-line bookmaker comparison. Recheck the price with the bookmaker before acting. Export the filtered CSV or full JSON when you need a record; preserve the filter settings and observation times. These are structured research exports, not a claim of validated historical performance.

05

Investigate disagreement and correlation

A large model/market gap can indicate an opportunity, stale inputs, a mismatch or model error. Inspect the source clocks, lineups, sample and settlement before deciding what the difference means. “Best available current price” means the best eligible retained quote in scope, not the most trustworthy forecast or a guaranteed executable offer.

The EV screener is primarily for straights and is not recommended for parlays. Repeated matches, related goal markets and shared team exposure can be correlated. Multiplying marginal probabilities without a defensible joint model does not produce a reliable accumulator probability or combined EV.