The Metric Lab is a deep dive using the FantasyLabs Tools to analyze the predictiveness of different statistics and proprietary data. The series provides analysis by looking at the dynamics of value, ownership, and consistency by using our massive database of historical trends.
“What NHL Metrics Should I Prioritize in my Models and Player Selection?”
This will be a common question for people degenerates looking to dip their toe into DFS hockey over the coming weeks with NFL on the back burner. This time of year, most of you are probably crushing one of Jonathan Bales’ books before MLB starts. That alone is probably a positive expected value move but think of the Metric Lab as a far less entertaining and carefully researched version for NHL. Unlike Bales, I do promise to keep dick jokes to a minimum.
So there’s that.
I digress. If you’ve watched any of our NHL Inside the Lab shows, you know it makes sense to target peripheral stats. But what about hybrid stats like Corsi?
The bigger question for most of you: What the hell is Corsi?
Baseline Trend and Corsi For % (Month)
The argument for targeting shots over blocked shots is pretty simple: small events (shots) are exactly what lead to less predictable — but often more significant — events like goals and assists. So we are always looking for different versions to target the same thing to gain an edge. Corsi just so happens to be awesome.
If you google Corsi you get a detailed (and likely confusing) definition:
Corsi is an advanced statistic used in the National Hockey League to measure shot attempt differential while at even strength play. This includes shots on goal, missed shots on goal, and blocked shot attempts towards the opposition’s net minus the same shot attempts directed at your own team’s net.
Essentially this is all you need to know:
Corsi For = shots on goal + missed shots + shot attempts blocked
With that finally out of the way, using our Trends tool, we will continue to compare the Plus/Minus values to both Consistency and Ownership.
Plus/Minus: A player’s actual points minus his expected points in the context of their salary-based expectations. Note that DFS scoring is typically lower for NHL than other sports like NFL or NBA, so the NHL Plus/Minus values we see are likely to be relatively small.
Consistency: The percentage of games in which a player has produced within a standard deviation of his expected points based off of historical scoring and pricing. Can be used to identify high-floor players for cash games.
Today’s baseline trend features power play skaters in the 50th percentile or better in Corsi For over the past month:
It would be hard to consider someone in our player pool who doesn’t meet that type of floor.
Also, the value of this article is probably comparing Corsi to Shots, so here’s that chart again highlighting power play skaters using different percentile buckets for shots on goal over the past month. More importantly, the impact on Plus/Minus in comparison to both Consistency and Ownership:
Drumroll… now that same chart for Corsi For:
Intuitively, it makes sense that Consistency and Ownership are almost identical. However, two things stand out right away in regards to Plus/Minus.
- The truly elite Corsi For skaters are underpriced.
- The truly elite Corsi For skaters are simply much rarer than high-volume shooters.
I lean towards the latter because unlike shots, the counts (samples) are much lower for the top five percent Corsi For skaters. Let’s dive deeper into the different positions.
Positional Breakdown
Each position has unique intricacies as to which stats strongly affect value, but the true edge comes in identifying which of these metrics are not typically priced into their salaries. This chart looks at the same percentile buckets as above, but breaks things down by position and has removed samples (counts) fewer than 25:
It’s a small sample, but elite Corsi For centers like Connor McDavid (99th percentile over the past month), are extremely valuable based on their salary-based expectation and should almost always be in the conversation for cash games. The upside for centers in the 95th and 99th percentile is also much higher than targeting shots (10 and 12 respectively).
Note: Upside figures show the percentage of games in which a player has scored at least one-half standard deviation above his point expectation based on salary. Can be used to identify high-upside players for tournaments.
The 59.7 percent Consistency for top-one percent Corsi centers is the highest of any single metric — for any position — in our entire NHL database. Spoiler alert: this destroys the other positions where top-five percent shot attempt defensemen (52.7 percent) and top-one percent power play shot wingers (48.1 percent) lead the way.
We like to pay up for defensemen because they can be peripheral stat monsters, so it’s not surprising to see them lead the way in terms of overall sample size at the top end of Corsi For. Brent Burns (99th percentile Corsi for over the past month and year) sets the ceiling in DFS hockey yet again. Some would say he’s #good.
One last note. It’s interesting that top-one percent Corsi For wingers performed very poorly relative to the other percentile buckets. This could be noise — given how strong the top-five percentile bucket performs highlighted by players like Patrick Kane (98th percentile in Corsi For over the past month) — but it does provide caution as to the validity of the small centerman sample in this bucket as well.
Photo via Kim Klement-USA TODAY Sports
