Match every field before comparing
Two rows are comparable only when event, market, selection, settlement period, currency, price format and capture time align. “Manchester win” may mean regulation time on one page and qualification on another. Promotions, boosts and exchange prices can also have different rules or costs; Pinnacle's current bet-placement documentation provides one first-party example of price-change handling.
Six-step workflow
- Assign a stable fixture and market ID.
- Capture all source prices within a declared time window.
- Record availability and maximum intended executable stake.
- Convert accepted quotes to decimal odds.
- Calculate EV with one unchanged probability estimate.
- Preserve price changes, rejections and partial acceptance.
Pinnacle's bet-placement documentation illustrates that a selected price can change before placement. The accepted bet record, not the earlier screen quote, belongs in performance results.
Worked price comparison
Assume p = 0.48 for the same regulation-time selection.
| Source | Decimal price | EV = (0.48 * d) - 1 | Status |
|---|---|---|---|
| A | 2.05 | -0.016 | Available |
| B | 2.12 | 0.0176 | Available |
| C | 2.16 | 0.0368 | Price changed before acceptance |
If B is accepted, use 2.12 and 0.0176 in the decision record. Do not report C's 2.16 unless it was executable for the intended stake.
For a £20 illustrative stake, gross return difference between 2.12 and 2.05 = £20 * (2.12 - 2.05) = £1.40 when the bet wins. The expected-value difference per unit = 0.0176 - (-0.016) = 0.0336 units under the same 48% estimate.
Why source choice still needs evaluation
A historical study found meaningful forecast-quality differences among ten bookmakers and six leagues (Štrumbelj and Robnik-Šikonja, 2010). Price comparison should therefore retain source identity; it should not assume every source offers an equally informative probability benchmark.
Operational costs
Line shopping requires account access, identity verification, deposits, withdrawals, data capture and time. Those costs may matter even when they do not fit neatly into the one-bet EV formula. Record them separately and avoid encouraging users to open many accounts without considering consumer protection and personal affordability.
Verification checklist
- Same market and settlement terms?
- Prices captured close enough in time?
- Promoted and ordinary prices separated?
- Accepted rather than merely displayed price stored?
- Exchange commission included where relevant?
- Every unavailable or rejected price preserved?
Report common coverage
A fair source comparison starts with the intersection of events and markets available from every included source. Report results on that common set before adding source-specific coverage. Otherwise an operator covering easier-to-price leagues can look better merely because its sample differs.
Keep three counts for each source: displayed quotes, executable quotes at the intended stake, and accepted bets. Also report median capture-time difference and price-change rate between selection and acceptance. The resulting table tells readers whether a higher average quote was routinely usable or was concentrated in a small, stale, or low-limit subset. Do not fill missing prices with the worst available quote; missingness is an outcome of the operational test.
Next step
Use Odds Comparison Football for the next part of this topic.
Continue learning
- Next guide: Positive EV Betting
- Related guide: Value Betting Spreadsheet
Assumptions and limitations
The prices, probability and stake are illustrative. Availability, limits, jurisdiction and account terms vary. A better price improves the payoff conditional on the same outcome, but it does not validate the model probability or guarantee positive returns.
GambleAware
GamCare
Gambling Therapy
GAMSTOP