Correlations and DFS Stacks for Super Bowl LVIII

February 10, 2024

Thunder Dan Palyo analyzes the correlations between all relevant players for Kansas City and San Francisco and gives you his favorite ways to stack both sides for Super Bowl LVI in your DraftKings and FanDuel DFS showdown lineups.

The Super Bowl is finally just days away. The rematch of two heavyweights should make for a fantastic game and I'm going to attempt to give you some ideas for how to build your single-game DFS lineups for the big game.

I first attempted this type of research two years ago for the Bengals-Rams matchup. My goal is to try to find an edge for GPPs by identifying which players on teams positively correlate with each other and which of those don't. We have a full season of data to process with the main offensive players on both teams having played the majority of their teams' games. I went ahead and created a chart for each team that maps out the correlations and I attempted to identify some of the best two and three-player stacks.

I'll be focusing mainly on the DraftKings Showdown slate since we can't play defenses on FanDuel and they don't have the dynamic captain pricing either. I'm also going to include two correlated lineups to give you an idea of how to use the data to make some GPP builds. Let's have some fun with these builds and get our deep dive down the rabbit hole of correlation started!

DFS Correlations: Super Bowl LVIII

Kansas City Chiefs

(click to enlarge)

One quick note about these numbers before we get too deep into the weeds. If you are not familiar with correlational coefficients (the very last column on the right of the charts), they can range from between -1 and 1, and the higher the number (or closer to 1.0) the more evidence for a strong correlation (as one number in the set goes up, the other is likely to increase as well). The closer the number is to -1, the more likely that the two data sets are negatively correlated (as one number goes up, the other number decreases).

In layman's terms, we are looking for statistical evidence that one player's production (or lack thereof) may be linked with another player's production.

The first thing that jumps out at me here is that Mahomes is positively correlated to all of his receiving options but not with his running back Isiah Pacheco. The trio of Mahomes-Pacheco-Rice (a traditional QB-RB1-WR1 stack) actually had the lowest average of the three combos that I calculated at just 45.9 total points per game. So if you are thinking of using Pacheco in lineups, it might make sense to fade Mahomes at his lofty salary and try to capture some of the Chiefs' passing production elsewhere.

The best option might be pairing Pacheco with Rice as those two had a fairly strong correlation of .38. And while Rice likely has a tougher matchup than TE Travis Kelce, he will most likely be the less-rostered option of the two. The Pacheco-Kelce does not correlate at all with this season's data, likely due to both players being somewhat touchdown-dependent for their scoring.

Mahomes definitely correlated more strongly with MVS than Justin Watson if you are looking to take a shot on a deep-ball connection. Both receivers are projected to be around 13-15% rostered with Watson just a few hundred bucks cheaper on DraftKings.

The defense-kicker correlation is there but is incredibly weak at just .07. Butker and the defense/special teams combined for 31 points twice this season but finished with just an average of 17 DK points per game. I don't think playing them both together and taking up two roster slots is a very compelling play.

All of the KC receivers are negatively correlated, so if I am building any 3-man KC stacks, I am not likely to have two wide receivers paired with Mahomes. The top 3-player combination this season was Mahomes-Rice-Kelce with 93.8 points in Week 7 with Mahomes-Pacheco-Kelce only one point behind in the same week.

It's an obvious play, but stacking Mahomes with just Kelce makes sense with their 0.57 correlation and incredible chemistry. They are two of the chalkiest plays on the board, but rostering them together will likely make your lineup more unique than playing either of them alone as their salaries together take up 40% of your salary cap even with neither player in the captain slot.

The last thing I will mention is that the RB1-defense and QB1-OPP QB1 correlations are both pretty strong negative correlations. The logic that the defense having a good game will lead to Pacheco following suit has not been true this season and neither has the "shootout" logic that suggests that when one QB is going off, the other one is likely putting up points, too. So something I will be doing a lot with my builds is avoiding playing both QBs together.

San Francisco 49ers

(click to enlarge)

The Niners were an offensive juggernaut this season other than Christian McCaffrey, you didn't know where the production was coming from on a week-to-week basis. I'm intrigued by the fact that when these two teams met four years ago it was the Niners who were coming into the game as the defensive stalwarts and KC as the offense that could not be stopped and this year, with all due respect to the Niners defense, it's really quite the opposite this time around.

Let's start with the best positive correlations. Brock Purdy's best pairing just might be Deebo Samuel who leads the way at 0.46. Not far behind is George Kittle with a 0.34 mark. Brandon Aiyuk is a distant third of the pass catchers at 0.15 and one look at his game log is a reminder of the type of rollercoaster of a season he had with wildly good results sporadically mixed in with average or below-average production.

The good news here is if you want to stack up Purdy with Deebo and Kittle, there's a neutral 0.03 relationship between Deebo and Kittle while the rest of Purdy's weapons all have negative correlations with each other. Purdy and McCaffrey have a neutral mark of 0.02 so they aren't an ideal solo stack, but we also have to consider grabbing as many raw points as possible, too. McCaffrey will carry the highest rostership of any player in showdown contests, but pairing him with anyone other than Purdy may not be wise as CMC's ceiling games have often led to relatively mediocre production from the rest of the Niners' studs.

The kicker-defense pairing for San Francisco shows a much stronger correlation at 0.28 than Kansas City's and the duo combined for 20+ DK points six different times this season. It was a boom/bust year, however, as we can see there were five weeks where they combined for fewer than 10 DK points and their overall average of 15.7 DK points was lower than Butker and the KC defense.

I'll end with a few last observations. Brock Purdy had very slight positive correlations with the opposing QB and WR1, but not enough to move me off my stance that I am avoiding "shootout" builds with both QBs. McCaffrey, like Pacheco, also had a negative correlation with his defense so that traditional pairing doesn't interest me much. One last pairing I included here was the Niners' defense with the opposing QB which carried the strongest negative correlation I observed at 0.73. That simply means you should not be playing Mahomes and the KC defense in the same lineup (not that you wanted to) as that combination is one of the least likely to both score well within the same game.

Conclusion

Sometimes you have to throw conventional wisdom and correlations out the window, I get it. We have a tight salary cap that is going to limit what we can build and only so many roster spots. Every lineup should tell a story of how you think the game is going to go, we shouldn't be randomly smashing plays in based on what fits. Give the optimizer a rest and build a few of these lineups by hand with a very specific outcome in mind.

I have been on the record for two weeks now that the Chiefs are going to win this game in a relatively low-scoring affair. So I will construct some of my builds towards that outcome, but if you are playing a lot of different lineups then you should be building with multiple outcomes in mind. Playing every angle is impossible and definitely not the best approach, but spreading out your exposures to different players is going to give you a better chance at one of those lineups ending up at the top of the standings, too.

Sample DFS lineups

Lineup 1: Chiefs win behind a big day from Kelce, McCaffrey still gets his, MVS catches a deep bomb

Lineup 2: Niners win behind a big passing day for Purdy who hits Kittle and Deebo for TDs, McCaffrey fade, Pacheco + Rice finish as top KC performers with Mahomes/Kelce fade.

More Fantasy Football Analysis