NHL DFS: Early Season Advanced Stats plus Thursday GPP Stacks
Expected Goals is a metric that attempts to quantify how many goals a team should have scored based on the shot location and a variety of other factors. Our projection model leans heavily on longer-term trends and thus we are mostly making in-season adjustments based on vegas odds at this point to our macro ratings. This is further confounded by the fact that the NHL has been changing shot location information which has made these models slightly less reliable than in previous years.
The fancy stat boys typically say that these trends stabilize around 20-25 games as for as predictive value for the rest of the 2019-20 season.
Still, it is helpful to try to understand which teams may be improving relative to their 2018 baselines at a team level both for xGF (team offense) and xGA (team defense). The chart below shows the 2018 and YTD team performance at 5v5 in adjusted xG. All information from Corsica.
The table is sorted by each team’s xGF% which is effectively a measure of how well a team has carried the play at 5v5. A few interesting notes:
The Flyers look stronger than the standings suggest. We had theorized in our season preview that Philadelphia could be an improved defensive club and that seems true early on. In addition to under-performing last year’s baselines, they have improved both in generating offense as well as preventing high-quality scoring chances. While their metrics aren’t likely sustainable (See last night’s game vs Edmonton) Philadelphia is no longer a team to pick on.
Buffalo at Los Angeles is an interesting matchup with two teams trending up. The Kings have improved offensively through a short spurt of games and Drew Doughty looks revitalized. They have even got production from Ilya Kovalchuk. While I’m hesitant to buy in LA’s resurgence fully I think it is an interesting contrarian GPP spot to target, particularly the fully correlated LAK1 line. They are priced at $15.7K tonight and represent an upside pivot off of the popular New Jersey Devils “Hail Mary” line (Hallmieri, get it? I’ll be here all week!). Buffalo is on a B2B which helps, though I won’t get carried away with exposure since Buffalo themselves has shown tremendous improvement in xGA/60 at 5v5, at least partially attributed to Dahlin’s continued growth.
The New York Rangers are still #BAD defensively. They look quite bad early and maybe the tank is the move. NJD1 is our top value stack on the entire slate. They are fully correlated, Hall and Palmieri have strong individual floors from their shooting and if they factor in on the scoresheet they have double bonus potential. Jack Hughes is a high upside prospect that has the offensive talent to produce a ceiling game alongside them and played 17 minutes last game with Hischier down. I would have been excited to play them at $18.5K much less $15.9K on DraftKings. The pricing generally is tough on tonight’s slate so the $15.9K price is extremely attractive and it is possible that Hall could hit 40 percent ownership in GPP, similar to what happened with Ovechkin last night. The spot for Hall is much better, but Ovechkin’s performance at least shows the viability of fading chalk in the NHL. A full fade could leave you in the poor house, but be sure to differentiate your NJD1 lineups from the field somehow.
What to do with San Jose and St. Louis? Legitimately, I’d be surprised if anybody lost more on St. Louis than me last year. I started out the season playing the Tarasenko line heavily to no avail, and then stacked against a below-average Blues defensive club right before their epic cup run and goaltending from Binnington. And then they beat the Bruins in the Stanley Cup to put the icing on the cake. The personnel is the same for St. Louis and I’m still hesitant to attack them with stacks until the sample size builds as I can’t quite create a narrative around the change. If you do buy into the early season trends Vancouver 1 could be a contrarian GPP target. The Sharks is easier to create a bit of a narrative as they lost Pavelski, recently signed an aging Marleau, and generally have limited depth in their bottom six or defense. I used to avoid stacking vs the Sharks unless it was to take advantage of Martin Jones meltdowns but its possible if these trends continue there could be more chances given up + Martin Jones meltdowns.
Macro Roster Construction:
Is 3-3 stacking dead? Not so fast, but it at least bares a lot more thought than just dummying 3-3 stacks into an optimizer. Simply put, forwards on the same line are less correlated in the new DraftKings score format than they were last year, even if they are fully correlated on the Power Play. Seguin and Benn go from 0.48 to 0.43. Gaudreau and Monahan from 0.65 to 0.59. Oshie and Backstrom from 0.40 to 0.34. The correlations are still there but they are weaker than they were previously. Savvy GPP players have already become more open minded about running 3-3, 3-2, or even 3-man stacks with one-offs than they were previously.
Full Correlation Spots for tonight include Arizona 1, Boston 1, Buffalo 1, Calgary 1, Detroit 1, LA Kings 1, New Jersey 1, New York Rangers 1, Ottawa 1, St. Louis 1, Tampa Bay 1, Vancouver 1. Consider mixing these types of stacks as either 3-man stacks or 2-man stacks, then rotate in two man combos from value lines.
Some more thoughts for tonight’s slate from The Numbers Guy coming soon…