Most sportsbooks will only allow you to put on parlays on uncorrelated events, but Fanduel has a separate section which will allow you to put on parlays within the same game. With the odds of each individual bet and the odds of the parlay, we can back out the implied correlation with some assumptions.
Assumptions are that each prop is a Bernoulli random variable (akin to a flip of a biased coin with a given probability). We then use the correlation coefficient formula ρxy = Cov(x,y) / σxσy which is straightforward for a Bernoulli random variable.
FD line
FD probability
True probability
Mahomes o310.5 pass yds
-112
0.528
0.500
Kelce o83.5 rec yds
-112
0.528
0.500
Parlay
182
0.355
?
Implied correlation of parlay
0.303
0.418
FD line
FD probability
True probability
Titans +7
-105
0.512
0.488
Henry o117.5 rush yds
-112
0.528
0.500
Parlay
165
0.377
?
Implied correlation of parlay
0.428
0.534
FD line
FD probability
True probability
Over 52
-105
0.512
0.488
Henry o117.5 rush yds
-112
0.528
0.500
Parlay
223
0.310
?
Implied correlation of parlay
0.156
0.263
FD line
FD probability
True probability
Henry o117.5 rush yds
-112
0.528
0.500
Kelce o83.5 rec yds
-112
0.528
0.500
Parlay
245
0.290
?
Implied correlation of parlay
0.043
0.159
FD line
FD probability
True probability
KC -7
-115
0.535
0.512
Henry o117.5 rush yds
-112
0.528
0.500
Parlay
427
0.190
?
Implied correlation of parlay
-0.373
-0.265
My takeaways:
Fanduel has variable correlations set for different groups of events, ranging from positive to negative, so while some are prohibitively expensive, some are not. Though the implied correlations of the parlays are sure to contain a vig, there are likely to be some +EV spots where FD is underestimating the correlation. These may be more beatable than individual props considering that same game parlays are a smaller overall market within sportsbooks.
Plus EV Thoughts 