The product the casino actually wants you to play
Every sportsbook product manager has the same data on their dashboard: parlay handle is a small fraction of total wagering, but parlay revenue is the disproportionate driver of profit. New Jersey Division of Gaming Enforcement 2024 data: parlay handle 22% of total bets, parlay revenue 41% of net win. Why? Because the vig on a four-leg parlay is roughly four times the vig on a single bet. Same bettor, same game choices, four times the cost of admission.
Sportsbook UI design follows this economic reality. The "parlay builder" appears prominently. Same-game parlay tabs surface on every game page. Marketing emails feature parlay specials. The customer is being optimized — gently, persistently — toward the highest-margin product.
The math of compounded vig
# Each leg at -110 (decimal 1.909, implied 52.38%) 1-leg: implied 52.38%, fair implied 50.00%, vig 4.55% 2-leg: implied 27.44%, fair implied 25.00%, vig 8.89% 3-leg: implied 14.37%, fair implied 12.50%, vig 13.05% 4-leg: implied 7.53%, fair implied 6.25%, vig 17.04% 5-leg: implied 3.94%, fair implied 3.13%, vig 20.83% 6-leg: implied 2.07%, fair implied 1.56%, vig 24.43% 7-leg: implied 1.08%, fair implied 0.78%, vig 27.86% 8-leg: implied 0.57%, fair implied 0.39%, vig 31.13%
The bettor who places a 6-leg parlay is paying roughly 24% of stake to vig in expectation — equivalent to playing roulette with a non-standard wheel. The "huge payout if you hit" framing masks the fact that the expected return is brutally negative.
Payout calculation walkthrough
# Three-leg parlay example, $100 stake leg_1: Lakers -3 (-110) → decimal 1.909 leg_2: Chiefs ML (-150) → decimal 1.667 leg_3: Celtics +7 (+100) → decimal 2.000 parlay_decimal = 1.909 × 1.667 × 2.000 = 6.366 parlay_american = +537 # since decimal > 2 payout on $100 = $636.60 (profit $536.60) # Fair true-odds parlay (no vig) fair_p_LAL = 0.500 fair_p_KC = 0.600 fair_p_BOS = 0.500 parlay_fair_p = 0.150 (15.0% chance of winning all three) fair_decimal = 1 / 0.150 = 6.667 fair_payout = $666.67 # EV gap = $30.07 = 4.5% of fair payout # Equivalent to playing a 4.5% vig product
The bettor sees +537 American odds and feels like they're getting a great payout. The bettor isn't; they're paying compounded vig on top of base vig.
Same-game parlays — the worst vig in retail
The same-game parlay (SGP) is the highest-vig common product in US sports betting. Books offer SGPs because:
- Legs are correlated (book gets to price the correlation, usually unfavorably to the bettor).
- The product is uniquely engaging — bettors build narratives around a single game.
- Vig is invisible because there's no comparable "fair price" for the bettor to consult.
Industry data: SGP theoretical hold averages 18-28% across US books. Compare to single bet (4.5%) or even traditional cross-game parlay (8-17%). The bettor placing a 4-leg SGP on, say, "QB throws for 250+, RB rushes 80+, WR catches 7+, game total Over 50" is effectively paying ~22% to the book in expectation. Equivalent products: keno, slot machines with bad pay tables.
The pattern in 2026: DraftKings and FanDuel report 35-45% of recreational customer revenue comes from SGPs. ESPN Bet's product strategy is explicitly SGP-forward. Books are racing to build same-game-everything products because the margins are the best in the industry.
When can parlays be +EV?
Three conditions, all required:
- Each leg has documented +EV. Not a hunch. Not a hot take. Documented model edge.
- Legs are independent (or positively correlated in a way books haven't priced).
- Cumulative EV exceeds the parlay's vig load.
Worked example. Bettor has 53% true probability on Lakers -3 and 53% on Chiefs ML; both bets carry 1% edge each at -110.
single_LAL EV = +1.0% single_KC EV = +1.0% # Parlay at +264 parlay_p = 0.53 × 0.53 = 0.2809 parlay_payout = 3.636 # decimal at +264 parlay_EV = 0.2809 × 3.636 − 1 = +2.13% # Parlay EV is roughly double the sum of single EVs after vig drag. # Only positive when both legs are documented +EV.
Sharp bettors who do bet parlays do so on this principle: independent +EV legs compound into bigger relative EV per dollar. But: ① most "edges" on routine markets are not independent; ② sharps' parlay handle is a tiny fraction of their book volume; ③ books detect parlay accounts and limit them faster than single-bet accounts.
The promotional parlay-boost trap
Books occasionally offer "parlay boosts" — Lakers/Chiefs/Celtics parlay paid at +400 instead of standard +377 (boost of ~6%). On the surface, this looks like rebated vig. The catch: ① boost is often only on parlays that include legs the book wants action on (one-sided heavy public side); ② boost amounts rarely exceed embedded vig; ③ boost forces the bettor to play parlay structure they otherwise wouldn't.
A pro-grade analysis: only take parlay boosts where the boosted price actually exceeds your fair-price estimate. Most don't.
Cross-leg correlation — books are aware
Many natural parlay legs are correlated. If QB throws for many yards, the team total is more likely to go over. If a team blows out the opponent, the spread and ML are correlated. Books embed correlation adjustments in SGP pricing, usually making the parlay pay less than the math suggests.
| SGP combination | True odds estimate | Book payout (typical) | Hidden vig |
|---|---|---|---|
| QB >250 yds + WR1 >75 yds | +250 | +180 | ~14% extra |
| Team A wins + Team A team total >25 | +150 | +105 | ~9% extra |
| Over total + both teams >25 | +275 | +200 | ~12% extra |
| Team A wins + opponent QB <225 yds | +135 | +115 | ~5% extra |
The "savings" implied by correlated legs disappear in book pricing. Sharps who do bet SGPs do so when their correlation model disagrees materially with the book's — rare, and limited to specific market segments where books are less calibrated.
The cognitive bias engine
Why do recreational bettors love parlays despite the math? Three documented cognitive biases:
- Cumulative probability blindness — humans systematically underestimate compound probability. Four 60% events feels like ~50% combined; actual = 13%.
- Anchoring on payout — the +1200 number anchors expectation; the 7% win rate doesn't.
- Narrative coherence — building a parlay around a single game's story (QB throws TDs, RB scores, game goes over) feels more knowable than independent picks.
- Loss aversion vs. variance-seeking — facing certain small losses, bettors switch to high-variance lottery-like products. The classic prospect theory finding.
Books design UIs explicitly to amplify these biases. The parlay builder UI shows the building payout number prominently; the implied probability is hidden or absent. The "boost!" badges anchor the bettor's attention on price improvement, not vig load.
The professional bettor's stance on parlays
Industry consensus:
- Avoid parlays entirely if you don't have documented edge on each leg.
- Avoid SGPs regardless of edge — the correlation pricing is rarely favorable.
- If you must bet parlays, build them from independent +EV legs and verify cumulative edge vs. cumulative vig.
- Use parlay boosts only when the boost price exceeds your independent fair-price estimate.
- Track parlay results separately from straight bet results to honestly assess if your parlays carry edge.
Sources & further reading
- New Jersey Division of Gaming Enforcement — Sports Wagering Annual Reports 2022-2024 (parlay handle and revenue breakdown).
- American Gaming Association — "Sports Betting State Reports 2024" (parlay product market share).
- Paul, Rodney J., Weinbach, Andrew P. "Sportsbook profitability and bettor preferences in the sports wagering market." Journal of Gambling Studies, 2014.
- Buchdahl, Joseph. "Why parlays are a bad bet (most of the time)." Football Data Blog, 2017.
- Pinnacle Betting Resources — "Why bookmakers prefer parlay bets" (open documentation).
FAQ
How is parlay payout calculated?
parlay_decimal = decimal_A × decimal_B × decimal_C × .... Example: three legs at -110 each (decimal 1.909). parlay_decimal = 1.909^3 = 6.957. Payout on $100 stake = $695.70. Compare to true odds: at -110, fair win probability is 50.0%; three independent legs at 50% = 12.5% win rate; fair decimal = 8.00 (+700). The parlay pays 6.957 instead of 8.00 — the $104.30 difference per $100 is stacked vig. Three-leg parlay theoretical hold ≈ 13%, four-leg ≈ 17%, five-leg ≈ 21%.