Fansided

NFL DFS: GPP strategy for FanDuel and Draftkings

By Dan Palyo

SANTA CLARA, CA - DECEMBER 23: George Kittle #85 of the San Francisco 49ers runs with the ball after catching a pass against the Chicago Bears during an NFL football game at Levi's Stadium on December 23, 2018 in Santa Clara, California. (Photo by Thearon W. Henderson/Getty Images)
SANTA CLARA, CA - DECEMBER 23: George Kittle #85 of the San Francisco 49ers runs with the ball after catching a pass against the Chicago Bears during an NFL football game at Levi's Stadium on December 23, 2018 in Santa Clara, California. (Photo by Thearon W. Henderson/Getty Images) /
facebooktwitterreddit
Prev
4 of 7
Next
TAMPA, FL – AUG 23: Baker Mayfield (6) of the Browns throws a pass during the preseason game between the Cleveland Browns and the Tampa Bay Buccaneers on August 23, 2019 at Raymond James Stadium in Tampa, Florida. (Photo by Cliff Welch/Icon Sportswire via Getty Images)
TAMPA, FL – AUG 23: Baker Mayfield (6) of the Browns throws a pass during the preseason game between the Cleveland Browns and the Tampa Bay Buccaneers on August 23, 2019 at Raymond James Stadium in Tampa, Florida. (Photo by Cliff Welch/Icon Sportswire via Getty Images) /

NFL DFS: Correlations

Over the last few years, I’ve become a little obsessed with correlation coefficients in DFS, particularly for stacking purposes in NFL DFS GPPs. A correlation coefficient is a number between 0 and 1 (for positive correlations) and 0 and -1 (for negative correlations) often expressed as a two-digit decimal. The closer the coefficient is to 1, the stronger the positive correlation between the positions (or players). The closer the coefficient is to 0, the weaker the correlation and less statistically relevant it is.

FantasyLabs has a very cool correlation matrix that I can stare at for hours. You can clearly see that the strongest correlation over the last five years has been the QB-WR1 pairing at .53, followed by QB-WR at .51, and QB-TE1 at .47. It’s pretty intuitive that a quarterback and his top receivers would have the strongest correlations of any positions. QB-RB1 comes in at .42 and with the way that many lead backs are now involved in the passing game, it does make a lot of sense to stack a QB with their running back in an attempt to get exposure to every touchdown that an offense scores. Or to add the RB1 to an existing QB-WR or QB-TE stack to make a three-player stack.

However, there is actually one correlation that is stronger than any of those we’ve looked at so far and it’s QB-Opposing QB at .58. This makes a lot of sense when you think about games that turn into “shootouts” and while you can’t roster two QBs on FanDuel or Draftkings, you can add opposing players to your existing stacks to create “game stacks” of the games that you think have the best potential to be extremely high scoring.

In fact, the QB-opposing WR1 correlation is a strong one at .37, which makes a QB-WR1-opposing WR1 three-man stack a really solid three-man stack in GPPs. If you find a game with a high total that you really like to hit the over, it would make sense to stack that game multiple different ways, utilizing both QBs and several combinations of players from both teams. Four or five pieces from a game are really not even out of the question if you think the game has the potential to be a scoring bonanza.

There are definitely some combinations that you want to avoid, too. QB-Defense has a poor correlation and RB1-RB2 is another that you want to steer clear of using because it’s rare that both backs on a team can make value if they are splitting the rushing production. If you’re going to stack a player with defense, it should be the RB1 as that position has the strongest correlation of any of the offensive positions. Defenses that create turnovers also give their offense short fields and more chances to score. Defenses that can hold opponents to low point totals also can create game scripts that favor a run-heavy approach in the fourth quarter of games as teams attempt to run down the clock by grinding out yards on the ground.