Editor’s Note: FanDuel is running a $1,500,000 fantasy football league in Week 16. The top 66,000 teams win cash with $100,000 paid to first place on Sunday. Join now!
FanDuel advice: Using Vegas to pick a roster
Do high over/unders really correlate to better FanDuel performances?
Jason Getz-USA TODAY Sports
When I first started playing FanDuel fantasy football, one of the guiding tenets I was taught was to follow the points. Find the games where points will be scored and use players in those games.
It stands to reason, of course. If you can identify the 42-30 game, any given player in that game is more likely to tally fantasy points than that guy’s analogue in a 17-10 affair. And handily enough, every week we get a general projection of the amount of points that might be scored in a game in the form of betting-line over/unders.
In fact, that was the advice I was alluding to at the top: fill your lineup with guys in games that have the highest over/unders and maximize your chances of success. What this presupposes, of course, is that the over/unders correlate heavily with the outcomes of games. So I charted every game (through Week 14) this year by over/under (as cataloged by footballlocks.com) against the actual totals of the games, looking for some kind of trend line:
The identities of the axes are pretty irrelevant, but that R-squared value? A whopping 2 percent? Yeah, that isn’t exactly the correlation you look for. We remember when the Vegas folk nail it (“They said it would be +6, and the team scored a touchdown with five seconds left to lose 34-28? Amazing!”), but don’t go back to the lines when they are way off.
Now, this is woefully incomplete. A handful of big misses, even in a season of 14 weeks, can skew that R-squared by a lot. And their over/unders are grouped to 40-something points, give or take, because regression always wins, while actual scores skew much more wildly. The Vegas guys do know a lot of things. So I wondered, how did the biggest performances (the ones I noted for Saturday's column) shake out as it related to the over/unders? You know, was the advice I received accurate? Did games with higher over/unders yield more huge performances?
Well, that's a little more complicated to figure. But I'll start here: the highest over/under of the season was in the Eagles-Falcons Week 1 game at 56 points (which, with a final score of 26-24, worked out fairly well, actually). The lowest was the Browns-Jets game, also in Week 1, set at 38.5 points (and again, the final score was 31-10, so Vegas did well there). That gives us a range of 17.5 points. I can divide that into three categories:
- Over/unders from 38.5 to 44.5 (98 games)
- Over/unders from 45 to 49.5 (86 games)
- Over/unders from 50 to 56 (25 games)
From there, it’s simple to see where the biggest performances (30-plus points for quarterbacks; 25-plus for other positions; there were 71 such performances through Week 14) fall on that scale:
- There were 28 big games from over/unders 38.5-44.5 (28.6 percent)
- There were 30 big games from over/unders 45-49.5 (34.9 percent)
- There were 13 big games from over/unders 50-56 (52 percent)
To a point, this ignores double-ups -- games where multiple performers put up huge games (of which, as I noted Saturday, there are several) -- but in general, it works. In 25 games with the highest over/unders, we got 13 huge performances. In games in the other groups (which were roughly three to four times more plentiful), the big games were only a little more than twice as common.
The takeaway? Well, that aphorism from the top does apply, for the most part. You're more likely to find a huge performance (a FanDuel-winning performance) in the games Vegas expects to be high-scoring. That doesn't mean you can't go elsewhere (Thomas Rawls' 36 fantasy points in Week 11 were the second-highest running back score in the data, and came from a game with a 40-point over/under), but your odds are better at the top of the list.
Vegas always knows.
See More:
