Define the three-way market
Home win, draw and away win are mutually exclusive outcomes for one accepted period. Confirm whether the market is regulation time, because qualification after extra time or penalties is a different event. IFAB Law 10 distinguishes drawn matches from methods used to determine a competition winner.
Betfair's football rules and DraftKings' soccer guide are current operator examples. Store neutral venue, listed home team, event period, abandonment rule and accepted result label.
Build one complete probability vector
For each match, estimate:
P(Home) + P(Draw) + P(Away) = 1
Do not estimate the selected outcome in isolation while leaving the alternatives undefined. A complete vector exposes mapping errors and supports consistent margin removal when bookmaker prices are used as a benchmark.
Worked accumulator calculation
Suppose three accepted decimal prices are 1.75, 2.00 and 1.90:
Combined price = 1.75 * 2.00 * 1.90 = 6.65
For GBP 10 stake:
Gross return = 10 * 6.65 = GBP 66.50
Suppose the corresponding illustrative model probabilities are 0.60, 0.52 and 0.56. Under an explicit independence assumption:
Joint probability = 0.60 * 0.52 * 0.56 = 0.1747, or about 17.47%
Raw break-even probability = 1 / 6.65 = 0.1504, or about 15.04%
The arithmetic is illustrative and does not establish that the estimates are accurate.
Dependence and price checks
OpenStax's independence condition must be justified across matches. Shared team rotation, competition incentives, weather and model errors can create dependence.
If benchmarking against a bookmaker's complete 1X2 book, record all three simultaneous prices and remove margin under a named method before treating them as probabilities. Expected value requires the model joint probability and every net payoff under OpenStax's framework.
Verification checklist
- Freeze complete home-draw-away probabilities.
- Archive accepted prices at full precision.
- Confirm regulation-time settlement.
- Map cross-match dependencies.
- Recalculate void or postponed legs.
- Reconcile final cash return with the receipt.
Next step
Use 1x2 Betting Explained for the next part of this topic.
Normalize a benchmark explicitly
Suppose simultaneous illustrative 1X2 decimal prices are 2.00, 3.50 and 4.00. Raw implied values are 0.5000, 0.2857 and 0.2500, which sum to 1.0357 rather than one. Under simple proportional normalization:
| Outcome | Raw implied | Proportionally normalized |
|---|---|---|
| Home | 50.00% | 48.28% |
| Draw | 28.57% | 27.59% |
| Away | 25.00% | 24.14% |
This is one margin-removal convention, not a claim that the normalized vector is the true probability. Preserve the raw prices and named method so another reviewer can reproduce it.
Compare the model with all three benchmark outcomes, not only the selected leg. A model can appear attractive on the chosen home win while assigning an incoherent total probability or systematically understating draws. Complete-vector calibration catches that failure.
Continue learning
- Next guide: Accumulator Calculator
- Related guide: Accumulator Tracking
Assumptions and limitations
The prices and probabilities are illustrative. Match-result forecasts are uncertain, draws are a distinct outcome, and a larger combined return does not demonstrate value.

