BetAssist uses information from selected sharp bookmakers and exchanges to estimate fair probabilities, check the quality of the evidence, and compare those estimates with the available line.
Grades: value and fair chance
A grade describes estimated value. Fair chance is the estimated probability of winning, shown separately beside each selection. Grades use unrounded inputs, so a threshold can be crossed even when the displayed percentages look the same.
- A: estimated edge of at least 3% and fair chance of at least 35%.
- B: edge of at least 2%, fair chance of at least 30%, and a net decimal price of 5.0 or shorter.
- C: edge of at least 1%, fair chance of at least 20%, and a net decimal price of 5.0 or shorter.
- D: outside the A–C value rules, including lines that do not meet the chance, price or evidence requirements.
- F: an estimated edge below −3%.
- —: insufficient reliable evidence for a grade.
A is checked first, then B, then C. Match result, goals and both-teams-to-score markets use these criteria. Evidence checks can limit a grade to D or leave a line ungraded.
Selections and check times
Recommended underdogs and new accumulator legs must qualify as A, B or C. A fixture’s overview selection is not automatically a recommendation. You can also build your own selection list, keeping one selection per fixture.
The checked time shows when the assessment was observed. Refreshing the website does not fetch bookmaker prices. When a new assessment is captured, the board updates; a changed grade does not change a price you have already taken.
How the record is calculated
The club record combines historical backtesting through 22 September 2026 with recorded pre-match assessments from 23 September 2026 onward. Each settled selection is counted once. New results are added automatically to the same grade table.
Backtesting applies the grading rules to saved historical estimates. Those rules were developed using this history, making that part of the record in-sample rather than predictions published at the time. Forward testing uses assessments observed and published before the match. Combining the two does not turn backtested results into forward-tested performance.
For tracked results, we use the latest eligible assessment within 35 minutes before kickoff for the same market, selection and goals line. An unavailable assessment or missing close is excluded. Match results use regulation time, excluding extra time and penalties. Tracked results keep the grade recorded at the time, including when a market’s eligibility changes.
Hit rate is wins divided by settled selections. Flat ROI is profit divided by one unit staked per settled selection, using recorded net prices. Pending, unscored and void selections are excluded. A+B and A+B+C combine the corresponding grade rows; they are not additional selections. Multiple selections from one fixture are correlated.
Backtest data and validation
The historical dataset covers 1,272 club fixtures and 19,080 market lines. Grades use saved rounded estimates and exclude nonvalue evidence routes from A–C. Historical freshness is approximate, and the archive cannot reproduce every current evidence check. Some historical selections fall outside the tracked record’s 35-minute capture window. Returns use recorded profits rather than prices reconstructed from rounded inputs.
Backtesting and ongoing forward testing help assess performance. Neither guarantees future results, and actual price availability and customer costs can change realised returns. Accumulators retain their original entry terms and are not combined with single-selection returns.