Parlays compound margin against you when individual legs lack edge. Same-game parlays add a second problem: bookmaker correlation pricing that captures most of the natural correlation value before you see the price. The disciplined approach is to model each leg independently, apply explicit correlation adjustments, and only bet when the combination beats the bookmaker SGP price after correlation is accounted for.
A single bet at 5% bookmaker margin gives you 95% of fair price expectation. A 4-leg parlay of similar bets compounds to (0.95)4 = 81.5% — your edge has been eroded from −5% to −18.5%. The same mathematics that lets a small bookmaker margin become unbeatable across many bets makes parlay margins structurally hostile to bettors who do not have a positive-EV edge on every single leg.
Before considering any parlay, model each individual leg as a single bet. Use the Poisson calculator for goal markets, your xG model for AH and 1X2, and player xG per 90 for player props. Calculate the fair probability and fair price for each selection independently. If any single leg does not show positive EV against the bookmaker single-leg market, do not include it in the parlay — it will only add negative EV to the combination.
For independent selections (different matches): combined fair probability = P(leg 1) × P(leg 2) × P(leg 3) ... Combined fair odds = 1 / combined probability. Compare against the bookmaker parlay odds. Positive EV = (combined fair probability × bookmaker parlay odds) − 1. Only consider parlays where this number is positive after accounting for the additional bookmaker margin compounding.
Same-game parlays cannot use simple multiplication because the outcomes are correlated. Quick correlation adjustments: home win + over 2.5 goals → multiply combined probability by 1.10–1.15. Home win + home star scorer → multiply by 1.20–1.30. Two same-team scorers → multiply by 1.15–1.25. The bookmaker price already includes their own correlation model — you are looking for cases where the bookmaker has under-correlated (priced two correlated legs as more independent than they actually are).
Take the correlation-adjusted combined probability and convert to fair odds. Compare against the bookmaker SGP price. Positive EV: bookmaker SGP odds × correlation-adjusted probability > 1. Most SGPs fail this test — bookmakers know correlations and price them in. The genuinely undervalued SGPs are rare but tend to come from specific correlations: home star scorer + home win, BTTS + over 2.5 goals (highly positively correlated), goalscorer + first-half goal in their match.
Parlays and SGPs have higher variance than singles — losing streaks compound and bankrolls deplete faster. Reduce your standard Kelly fraction by 50% for parlays and 75% for SGPs. Stake = (edge / odds − 1) × bankroll × 0.125 (for parlays) or × 0.0625 (for SGPs). Many disciplined bettors avoid SGPs entirely because the variance-adjusted edge rarely exceeds the additional margin.
Use these multipliers to adjust the simple-product probability for correlation. A value above 1.00 means outcomes occur together more often than independence would suggest. Below 1.00 means anti-correlated.
| Combination | Correlation | Adjustment | Verdict |
|---|---|---|---|
| Home win + Over 2.5 goals | Positive | ×1.10 to ×1.15 | Bookmakers usually price this in fully. |
| Home win + Home BTTS | Mixed | ×0.95 to ×1.05 | Roughly neutral — home team scoring helps win, but conceding hurts. |
| Home win + Star home player to score | Positive | ×1.20 to ×1.30 | Strong — home wins are usually goalscoring performances by the favourite's key player. |
| Both teams to score + Over 2.5 goals | Strongly Positive | ×1.20 to ×1.35 | Highly correlated — BTTS implies at least 2 goals already. |
| Two same-team players to score | Positive | ×1.15 to ×1.25 | Goalscoring performances cluster — high-scoring matches benefit multiple players. |
| Home win + Away player to score | Negative | ×0.85 to ×0.95 | Anti-correlated — away goals reduce home win probability. |
| Under 2.5 goals + No goalscorer prop | Strongly Positive | ×1.30 to ×1.40 | Same underlying low-scoring match outcome. |
| Different match: Match A win + Match B win | None (independent) | No adjustment | Multiply probabilities directly. This is a true parlay, not an SGP. |
Suppose your model says: Arsenal to beat Bournemouth at home — fair probability 65%. Over 2.5 goals — fair probability 60%. Saka anytime scorer — fair probability 50%. Independent multiplication: 0.65 × 0.60 × 0.50 = 0.195 (19.5%) → fair odds 5.13.
Apply correlation: Home win + Over 2.5 = ×1.12. Home win + home star scorer = ×1.25. Combined effect is not simply ×1.12 × ×1.25 — correlation compounds non-linearly. Use a conservative ×1.20 overall. Adjusted probability: 0.195 × 1.20 = 0.234 (23.4%) → fair odds 4.27.
Bookmaker SGP price: 4.50. Compare: 0.234 × 4.50 = 1.053 → +5.3% EV. This is on the margin — at this small edge, variance dominates. Apply quarter-Kelly fractioned by SGP variance discount: stake = 5.3% / (4.50 − 1) × bankroll × 0.0625 = 0.094% of bankroll. Tiny stake.
Most SGPs fail this test entirely. The point of the framework is to filter out the 95%+ that are negative EV after correlation, not to find "winning" SGPs to bet aggressively.
A traditional parlay (accumulator) combines multiple selections from different matches — bookmakers price these by simply multiplying the individual decimal odds together. A same-game parlay (SGP) combines multiple selections from a single match — for example Arsenal to win, over 2.5 goals, and Saka to score. SGPs require special pricing because the outcomes are correlated (Arsenal winning and the match going over 2.5 goals are positively correlated). Bookmakers price SGPs using proprietary correlation models that always include extra margin on top of the implied probabilities.
Same-game parlays carry significantly higher bookmaker margin than equivalent single bets. A typical SGP carries 12–25% margin compared to 5–8% on single markets. The reasons: (1) Bookmakers add their own correlation adjustments — usually conservatively in their favour. (2) SGP markets are popular with recreational bettors, allowing the bookmaker to price wider. (3) Liability is harder to manage for SGPs, so the price reflects that risk. The exception is when correlation is genuinely under-priced — but this requires a model that prices each leg independently and a correlation matrix you trust.
A parlay has positive EV only when each individual leg has positive EV. Combining negative-EV legs compounds the disadvantage — a 4-leg parlay of 5% positive-EV bets compounds to roughly +21.6% combined edge, but a 4-leg parlay of 3% negative-EV bets compounds to roughly −11.4%. The mathematics: parlay EV = (Π fair_prob × parlay_odds) − 1. For SGPs, you also need to account for the correlation adjustment that bookmakers apply, which can reduce or eliminate genuine correlation value.
Correlation is the statistical relationship between two outcomes within the same match. Examples: Arsenal to win + Over 2.5 goals = positive correlation (a goal-scoring Arsenal performance makes both more likely). Liverpool to win + clean sheet = positive correlation (winning teams concede less). Two players from the same team to score = positive correlation (a high-scoring performance benefits both). Negative correlation: home team to win + away player to score (anti-correlated). Bookmakers price SGPs by adjusting individual leg probabilities for correlation — so the SGP price is rarely the simple product of independent prices.
Use the accumulator and EV calculators to test combined-leg fair prices before placing parlays or SGPs.
