SGP Correlation Analyzer
Calculate the true probability of your Same Game Parlay by accounting for correlation between legs. NFL QB+WR stack? NBA points+assists? See how correlation affects your EV before you place the bet.
SGP Legs
Strong correlation can mask EV. Verify it with historical data — if QB+WR correlations are actually lower, you may be overestimating.
Correlation Guide by Sport
🏈 NFL
Strong +: QB pass yds → WR rec yds (0.6-0.8). If Mahomes throws for 300+, Kelce had a big day.
Strong −: RB rush yds → QB pass yds (negative). Run-heavy game = fewer passes.
Best SGP: QB+WR stack on same team, fading the rush.
🏀 NBA
Strong +: Star pts → team total (0.5-0.7). Tatum 30+ = Celtics likely over team total.
Moderate: Pts + ast (0.3-0.5). High usage = both.
Best SGP: Star player props + team win, factoring blowout risk.
⚾ MLB
Strong +: SP strikeouts → team win (0.4-0.6). Elite SP = more Ks = higher win prob.
Negative: HR props → Under total (neg). Home runs = runs = over more likely.
Best SGP: Pitcher Ks + team ML, avoiding correlated outcomes on both sides.
Why SGP Math Is Different: The Correlation Tax
Standard parlay math multiplies leg probabilities: P(parlay) = p₁ × p₂ × … × pₙ. That multiplication is only valid when the legs are independent — outcomes in different games that cannot influence each other. In a Same Game Parlay, independence fails almost by definition: every leg is drawn from the same game script. One efficient offensive night pushes the QB's yards, his top receiver's yards and the team total up together; one defensive slugfest drags them all down together.
Positive correlation raises the joint probability above the naive product. If Mahomes going over 250 passing yards makes Kelce's receiving over far more likely, then the parlay of the two hits more often than multiplying the individual probabilities suggests. This tool models that with a simple linear blend: it slides the true probability from the independent product toward certainty in proportion to the correlation you set. It is a first-order approximation, not a copula — but it is enough to show how violently correlation moves EV, which is the decision that matters.
Here is the flip side, and the reason books love this product: because correlated legs win together more often, the fair combined odds are shorter than the multiplied single-leg odds. Multiply the singles and you get a payout no book will ever quote on an SGP — and for positively correlated legs, that multiplied number genuinely overstates the fair price. The book's discount to that number is partly legitimate correlation repricing and partly extra margin. That extra margin is the correlation tax: SGPs typically hold 15–25%, versus 4–5% on singles. The problem for the bettor is that you cannot see where the legitimate repricing ends and the tax begins — unless you model it, which is what this page is for.
The practical workflow: estimate each leg's true probability (strip the vig from the standalone market first with the no-vig calculator), set a conservative correlation, and compare the correlation-adjusted fair odds to the book's quoted SGP price. If the quote is longer than your fair odds, you have found a spot where the book's blanket correlation haircut was too shallow. If the bet is only +EV at the optimistic end of your correlation range, it is not a bet — it is entertainment priced at a 20% hold.
SGP Correlation Formulas
Independent legs: P_ind = p₁ × p₂ × … × pₙ Fair odds (independent) = 1 / P_ind This tool's correlation blend (ρ = slider, 0–50%): P_corr = P_ind + ρ × (1 − P_ind) / 2 Pricing: Fair odds = 1 / P_corr EV per $1 = P_corr × quoted SGP odds − 1 +EV when quoted odds > 1 / P_corr
Worked Examples
Mahomes 250+ pass yds at p = 62%, Kelce 60+ rec yds at p = 58%. Independent product: 36.0% → fair 2.78. Multiplying the book's single prices (1.75 × 1.91) gives 3.34 — a payout no SGP will quote. With a 30% correlation blend: P = 36.0% + 0.30 × 64.0% / 2 = 45.6% → fair 2.19. If the book quotes 2.30, that is +4.8% EV if the correlation is real — and −17.3% if the legs were actually independent. The entire edge lives in the correlation estimate.
Same-team RB rushing over + QB passing over, each at p = 55%. Independent product: 30.3% → fair 3.31. But one game script suppresses the other: a run-heavy day caps passing volume and vice versa, so the true joint probability might be closer to 25%. At a quoted 3.10, naive math says −6.2% EV; the realistic number is 0.25 × 3.10 − 1 = −22.5%. With negative correlation, multiplication errs in the other direction — it flatters the parlay.
Frequently Asked Questions
Why are Same Game Parlay legs correlated?
SGP legs come from the same game, so they are not independent events — they share one game script. For example, 'Mahomes 300+ pass yards' and 'Kelce 70+ rec yards' are positively correlated: when Mahomes throws for 300, Kelce very likely had a big game too. Books partially account for this in their SGP pricing, but the adjustment is a blanket one, and mispricing happens. This analyzer lets you adjust the correlation percentage to see how it changes true probability and EV.
What is the correlation tax on an SGP?
It is the gap between the payout you would get by multiplying the single-leg odds and the shorter price the book actually quotes for the SGP. Part of that discount is mathematically justified: positively correlated legs win together more often than independence implies, so the fair combined odds really are shorter than the naive multiplication. The rest is extra margin the book keeps. Your job is to judge whether the quoted price sits above or below the fair correlated odds.
How much hold does a typical SGP have?
Same Game Parlays typically carry 15-25% hold — significantly more than single bets (4-5%) or cross-game parlays (10-15%). Books price correlation conservatively and layer extra margin on top, which is why SGPs are among the most profitable products for sportsbooks. Always check each leg's standalone odds first: if the SGP quote is far below the multiplied singles on legs that are only weakly correlated, the extra discount is pure hold.
Can a Same Game Parlay ever be positive EV?
Occasionally. Books apply broad correlation haircuts, and strongly correlated combinations — a QB alternate passing line stacked with his top receiver's alternate yardage line plus the team moneyline — are sometimes discounted less than their true joint probability justifies. The test is simple: estimate the correlation-adjusted probability, convert it to fair odds, and compare with the quoted price. Edges are rare and small, so be conservative with your correlation estimate.
How do I estimate the correlation percentage between legs?
Look at historical co-occurrence in game logs: how often did both legs hit in the same game versus what independence would predict? Same-team QB-to-WR1 yardage stacks are strongly positive, star scoring props correlate moderately with team totals, and rush-heavy versus pass-heavy props on the same offense are negatively correlated. When unsure, run the slider across a range — if the bet is only +EV at the top of your correlation range, it is not a bet.