NHL Player Correlations and DraftKings New Scoring Format
When DraftKings changed their NHL scoring, it became immediately clear to savvy NHL DFS players that they would need to change the way that they built their lineups. For years, running stacks with 3F from the same line was the defacto tournament approach, often adding in a correlated power-play defenseman and a correlated goalie to the mix. With the scoring environment set up to favor a high volume of shots over assists and shot volume against more than wins, the correlations naturally changed.
Firing 3-3 stacks had become so commonplace that some of the sharpest DFS players were moving away for it for leverage, and the change in the scoring system requires us to continuously revisit that approach.
Yes, the 3-man stack can still come to fruition, but we’ve seen more of a mix of 3-2 stacks, PP stacks or even 3-1-1-1 become successful. My lineup that won a seat to the Fantasy Hockey World Championship featured a 3-man stack (TB3) and a PP stack for Edmonton.
Despite the Live Final seat, I am still struggling to adapt to the proper roster construction if building multiple lineups and particularly struggle with MME on the largest slates, knowing that 3-3 is viable but possibly not optimal and that paying up for whichever 40+ point forward scores will be a requirement. This has lead me to bad GPP results and infrequent Top 1 percent finishes in large-field tournaments.
We know that even strength correlation (same line) matters. We know that power-play usage matters. We know that peripherals like shot volume matter.
But how can we determine which lines are best to stack? The logical place to start is by running correlation data for each team to determine which lines offer the most correlation. Below are some views of important lines on tonight’s slate based on correlation in DraftKings scoring between players year to date. I specifically left off the correlations for Nashville given their forward usage.
Florida has one of the higher team totals on the slate tonight and the top line is a popular line to stack on any given night. The Panthers were one of the biggest surprises when I ran the correlation data though, given that they feature full even strength and PP ice time together. Alexander Barkov is positively correlated with Jonathan Huberdeau (0.33) so no surprise there but he featured a surprising negative (albeit on 0.02) correlation factor with Evgenii Dadonov who has played with the top trio consistently for the majority of the year, and instead had strong correlation (0.25) with PP QB Keith Yandle and even strength minute monster Aaron Ekblad (0.18). While I don’t expect the Dadonov correlation to stay negative leaving him off of 3-man FLA1 stacks would be a contrarian approach that allows you to gain leverage on the field when Dadonov can’t come through alongside his linemates.
The Sabres top trio skates together in all situations and Jack Eichel has strong correlations to both Sam Reinhart (0.19) and Victor Olofsson (0.34). This keeps this line in play for 3-man stacks but it is also worth noting a very strong correlation for Eichel with Jeff Skinner (0.28) and Rasmus Dahlin (0.20) both which are stronger than Reinhart correlations. Eichel himself may carry modest ownership but contrarian DFS players should consider targeting stacks with Skinner and Dahlin over Reinhart (or in addition to Reinhart) for strong correlation and contrarian elements.
The Larkin line for Detroit features strong correlations and another fully correlated line as Mantha (0.32) and Bertuzzi (0.28) both have strong correlations to their pivot. One totally off the board player that surprisingly is correlated to this trio is Danny Dekeyser (0.55 correlation) and while I don’t think he’s interesting on a slate this large that correlation is one to keep in the back of your mind for a shorter slate. Because they feature a stronger correlation than Buffalo 1 I’d give Detroit a tiebreaker as my preferred stack in a single entry tournament.
Las Vegas 2
Cody Eakin slides between two talented shooters in Mark Stone and Jonathan Marchessault. While the correlation data is missing for this particular trio, we do know that Paul Stastny had strong correlations with Pacioretty (0.53) and Stone (0.42) while Pacioretty-Stone featured correlations of 0.25 between each other. The majority of Eakin’s performance will be linked directly to the success of Marchessault and Stone but given the lack of PP time, we would expect Eakin to offer less correlation. Eakin is not rosterable without his two wingers but you could consider playing the full trio or playing either Stone or Marchessault as a one-off play since they play on different power-play units.
Finally getting the “TORONTO 1” label from DailyFaceoff its been clear that the AM34 line for Toronto has been the superior fantasy performer. Another line skating together in all situations, AM34 features strong correlations with both William Nylander (0.42) and Andreas Johnsson (0.30) making them a contrarian value target tonight on the road in Las Vegas. Morgan Reilly has a neutral correlation with Toronto’s top trio and a much stronger correlation Tavares and Marner so I’ll pass on the 4-man stack here. As an added value, top lines in Las Vegas have been averaging 30 DK points on the season.
Correlations in Calgary
I would have expected a ton of correlation from Elias Lindholm but he is just 0.12 correlated with both Gaudreau, Monahan and Mark Giordano and actually features a stronger correlation (0.15) with Matthew Tkachuk. Gaudreau and Monahan continue to have elite correlation (0.60) between each other and 0.32 correlation with Mark Giordano. Considering the cost of Lindholm is higher than Giordano, our projection on Lindholm is lower than Giordano, and our correlation is weaker, I think opting for Giordano over Lindholm or in addition to Lindholm would be a really strong tournament option. The 0.10 correlation is still positive, so I am not writing off the 3-man stack but presenting interesting contrarian alternatives to consider when stacking Calgary moving forward.
(EDIT: The Flames now have trash lines.)
St Louis Value
Some strong value candidates tonight include the Blues lines due to salary savings. Jayden Schwartz and Brayden Schenn have skated together the majority of 5v5 and PP minutes on the year but still feature just a 0.15 positive correlation. With low volume shooter Tyler Bozak now skating alongside it is hard to be overly excited about rostering them as a 3-man stack despite the good value ratings. Instead, I find myself drawn to Ryan O’Reilly and David Perron who feature a girthy 0.70 correlation. Sammy Blais offers a neutral correlation so the STL2 line is the type of line I’d be looking to consider as part of 2-man stacks versus a full 3-3 lineup.
My biggest takeaways from this analysis are that the correlations are now more team and player-specific than ever before, and just dummying in 3-3 stacks into tournaments is akin to lighting money on fire, especially given it is so popular with the field. In the short-term, this will lead me towards building a few more thoughtful lineups versus taking more of a shotgun approach, but if you are set on running MME with 3-3 or 3-2 stacks I would be increasingly thoughtful about which lines you include in your mix.
If you’re interested in other player or line-specific correlations feel free to @ me in our subscriber slack.