NBA Parlay Correlation: The Math Behind Same-Game Pricing
Available in US
I once spent an entire weekend trying to figure out why one of my SGPs paid 4.20 when my back-of-the-envelope math said it should pay closer to 5.30. The answer was correlation, and the answer took me three flat afternoons of reading and one long conversation with a quantitative friend to actually understand. This article is the explanation I wish I had been handed at the start of that weekend, written for a UK punter who wants to know what the book is doing when it prices a same-game parlay.
Correlation is the reason same-game parlays do not behave like the regular accumulators most UK bettors learned on football coupons. The legs in an SGP are not independent events. They share underlying drivers, which means their outcomes move together in ways that change the math of the combined price. Understanding the correlation lets you read whether the SGP price is fair, where the book has under-adjusted, and when the product is worth playing.
Independent Events vs Correlated Events
The clean parlay you grew up on works because the legs are statistically independent. Arsenal winning their match has no causal relationship to Liverpool winning theirs. The probability of both happening is the product of the individual probabilities, full stop. If each is 60 percent likely, the parlay lands 36 percent of the time. The math is honest.
Same-game parlay legs do not have this independence. If Boston wins by twelve, the over on the total is more likely to have landed than if Boston wins by two. Tatum scoring 31 points is more likely on the night Boston wins by twelve than on the night they win by two. The probability of all three outcomes happening together is higher than naive multiplication suggests, because the same game-level conditions that drive one outcome drive the others.
The technical version of this statement is that the joint probability of correlated events exceeds the product of their marginal probabilities. A simpler way to feel it: when you build an SGP on a single game, you are essentially betting that the game unfolds in a specific shape. If the shape happens, multiple legs land together. If it doesn’t, multiple legs fail together. Either way, the variance of the combined bet is higher than the variance of equivalently priced independent legs would be – which is precisely why the book adjusts the price.
The mathematical adjustment is to subtract from the naively multiplied probability an amount that reflects the strength of the correlation. The book has built models that estimate this for every leg combination in their SGP builder. The model is not perfect, and the imperfection is where occasional value lives.
Positive Correlations Most Often Used
The strongest positive correlations in NBA same-game parlays are the ones that share a single dominant driver. Three combinations come up most often in my own SGP construction when I am hunting genuine value.
Team moneyline plus the team’s spread cover is the cleanest positive correlation in basketball betting. If a team wins by a margin large enough to cover the spread, both legs land. If they win by less than the spread or lose outright, the spread loses. The combined probability is essentially just the probability of covering the spread – which is lower than the moneyline alone, since the spread loss includes the narrow win. The correlation is mechanical and almost perfect.
Team moneyline plus team total over is positively correlated but more loosely. A team wins games it scores more in, but it can also win narrowly with both teams in the low 90s. The correlation strength depends on the specific matchup and the line. On a high-paced matchup with a tight total, the correlation is weaker. On a slow-paced matchup with the total set high, the correlation is stronger because the team must score above its season average just to clear the line.
Star player points over plus their team’s win is positively correlated and clinically reliable. Stars who put up 30+ generally play heavy minutes in competitive games their team wins. The exceptions are the games where the star scores big in a losing effort and the games where the star sits early in a comfortable win. Both happen, but neither happens as often as the median scenario where the star plays heavy minutes in a winning team performance.
The reason these combinations get the heaviest correlation adjustment from the book is that they are the ones casual bettors stack most often. The book’s pricing model has years of data to estimate the joint probability accurately, and the resulting compression of the headline price reflects that data.
The Hostile Correlations You Should Avoid
The flip side of positive correlation is negative correlation, where legs work against each other. These appear less often in the SGP builders – the book usually blocks the worst combinations – but they slip through more often than you might expect.
Player A’s points over plus Player B’s points over, when both are on the same team and play similar roles, is mildly negatively correlated. There are only so many possessions and shots to go around. A night when Player A shoots 25 times is often a night when Player B shoots 12 instead of his usual 17. The book may price the combination as if the legs were independent, but the joint probability is lower than independent multiplication suggests – which means you are paying a price as if the events were correlated favourably when in fact they cancel one another out.
Star player assists over plus the same star’s points over is structurally complicated. High-assist nights for ball-dominant stars usually involve them creating for teammates rather than scoring themselves. The two outcomes can both happen – a 30-and-10 game is iconic for a reason – but their joint probability is meaningfully lower than the product of their marginals. The book may not adjust for this fully, in which case you can spot value in the under-under combination instead.
Team total over plus opponent total under can be priced fairly if the spread expectation aligns, but it is a thinner edge than it looks. A team scoring well while their opponent scores poorly is a specific narrative – a blowout from one side – and the book has pricing tools that catch this scenario accurately most of the time.
The general rule is to ask whether the legs share an underlying driver or trade off against one another. Shared drivers favour you when stacked. Trade-offs work against you and the book may or may not have noticed.
The Quantitative Back-Solve
The way I evaluate an SGP price is to back-solve for the implied correlation adjustment and compare it to my own estimate. The math is straightforward once you have done it a few times.
Take the three legs from my Boston-Tatum example. Boston moneyline 1.50 (66.67 percent), over the total 1.91 (52.36 percent), Tatum over 26.5 at 1.85 (54.05 percent). Naive multiplication gives a joint probability of 0.6667 * 0.5236 * 0.5405 = 0.1886, or 18.86 percent. Fair price before margin is 1 / 0.1886 = 5.30.
If the book offers the SGP at 4.20, that implies 1 / 4.20 = 23.81 percent joint probability. The compression is 23.81 – 18.86 = 4.95 percentage points. That compression includes both the book’s correlation adjustment and their margin. If I estimate the genuine correlation adjustment for these three legs is around 2.5 percentage points – based on my own read of how the three legs interact – then the remaining 2.45 percentage points is pure margin. As a percentage of the fair price, that margin is roughly 11 percent. Compare to a single moneyline at 5 percent margin, and the SGP is more than twice as expensive in margin terms.
The other direction of this calculation is more interesting. If the book offers the SGP at 5.10 instead of 4.20, the compression is only 0.86 percentage points. That implies the book has under-adjusted for correlation, since my own estimate said the true joint probability deserves a 2.5-point lift. The remaining 1.64 points of compression is the book’s margin – still positive for them, but small, and small enough that if my read is accurate the SGP is genuine value.
This calculation takes about ninety seconds once you have done it ten times. The reason most UK punters skip it is that the headline price is already on the screen and the calculation feels like work for a small reward. The reason serious bettors run it on every SGP they consider is that the SGP margin is structurally fat enough that even one or two value plays per month meaningfully improves a season’s returns.
What the Book Knows That You Don’t
The honest disclosure here is that the book’s correlation model is informed by far more data than yours. They have run the joint probability of every leg combination across years of NBA games. They know which combinations the casual bettor reaches for most often and which combinations have produced unexpected payouts in the past. The compression they apply is not arbitrary – it is calibrated.
The book’s advantage is most pronounced on the obvious correlations. If you are stacking moneyline plus spread plus team total over plus the star’s points over, you are betting one of the most data-saturated combinations in NBA SGP construction. The book has priced this exhaustively and the margin is at its widest. Your edge claim has to overcome a deep, well-calibrated model.
The book’s disadvantage is on the obscure combinations. A bench player’s three-point attempts over plus the team’s quarter-by-quarter scoring pattern is a combination that the model may not have priced as carefully. The lower the volume of historical data on a leg combination, the more likely the book has applied a generic adjustment rather than a specific one. These edges exist, but they require deeper analytical work to spot, and the stake limits on niche SGPs are typically lower than on standard ones.
The general principle is to bet SGPs in the segments of the market the book has paid least attention to. The headline combinations they market on their homepage are the ones priced most aggressively. The custom combinations a punter builds from less-trafficked legs are the ones where the model is weaker.
UK remote casino, betting and bingo together generated £7.8 billion in gross gambling yield in the most recent reporting year, and the SGP product is one of the highest-growth chunks of that revenue precisely because the casual bettor finds the headline prices irresistible. The infrastructure that funds operator profit funds the integrity monitoring that protects the market, but it also funds the marketing campaigns that make the headline prices visible enough to drive volume.
Bringing the Correlation Calculation Into Your Routine
The discipline that separates a profitable SGP bettor from a recreational one is making the back-solve calculation part of the routine rather than an occasional check. Every SGP I consider gets the same treatment: multiply the legs as independent, compare to the offered price, estimate the correlation adjustment, judge the residual margin. The whole exercise takes a minute and a half. Most SGPs fail one of the tests and I bet the strongest leg individually instead.
The math is the same for two-leg parlays and twelve-leg parlays – just more legs to multiply, more correlations to estimate. The difficulty scales linearly with the number of legs, but so does the cumulative margin. A twelve-leg SGP at a major UK operator may carry effective margins north of 20 percent, which is the kind of headwind no edge claim survives over a season’s worth of bets.
Once the back-solve is a habit, the next step is to apply the same correlation thinking to alternate lines, which is where the book’s pricing logic intersects with your own probability estimate most directly. I have written through the architecture of alternate lines strategy in the next piece.
Between 2 and 6 percentage points of compression beyond the independent multiplication is typical, depending on how strongly correlated the legs are. The book’s margin sits on top of that adjustment. A three-leg SGP with effective combined margin under 8 percent is competitively priced; above 12 percent is heavy. There is variation across operators, but no consistent ranking that holds across all NBA SGP categories. The same operator may price one combination tightly and another loosely. The reliable way to find the better price is line shopping across three or four books for each SGP you build, rather than committing to a single operator.How much correlation adjustment is typical on a three-leg NBA SGP?
Is there a UK operator that consistently prices SGPs more honestly?
Escrito por los editores de «nba Betting Expert».
