clock menu more-arrow no yes mobile

Filed under:

What a new quality of competition metric says about the Flyers

QoC: it has levels like 1-2-3.

Heather Barry - SB Nation ©

Traditionally, quantifying the quality of competition that a player faces has been done through one of two ways. The first being based on ice time, and the second being based on on-ice metrics like Corsi and Expected Goals. While neither are without flaws, each way of evaluation gives us useful tools to separate those who have been sheltered versus those who have been metaphorically thrown into the fire. However, what they can’t do is tell us just how well a player does against the best of the best — that’s where PuckIQ’s three distinct competition levels come into play.

The Method

Since this is a pretty new concept, we’ll be creating a bit of a SparkNotes version of how their method works while also providing some examples. I highly recommend reading their explainer in full to thoroughly understand what goes into their competition levels at this time.

Currently, only forwards are used to determine competition level. This is because a forward’s statistical profile will of course differ mightily from their defensive peers. So, for now only forwards are used to determine if a skater’s competition level at a given time is Elite, Middle, or Gritensity. But wait, what do those three categories mean exactly?

To be classified as an Elite forward you need to check off the following four boxes.

You must:

  1. Have a points per 60 minutes rate greater than 2.21 at all situations.
  2. Have an average time on ice per game greater than the 75th percentile.
  3. Have a relative Corsi For percent (CF%) greater than the 40th percentile.
  4. Have a relative Dangerous Fenwick For percent (DFF%) greater than the 40th percentile.

Dangerous Fenwick is a PuckIQ-original metric and for those well versed in their on-ice metrics, you can think of it as another Expected Goals model. For those of you who maybe aren’t as well versed, Dangerous Fenwick is shots, both on goal and missed shots, weighted by the type of shot that was taken and where it was taken from. Each shot is assigned a value based on the likelihood of it becoming a goal. Like Corsi, a DFF% greater than 50 suggests that the player’s team won the shot quality battle while they were on the ice, while a percent below 50 suggests the opposite. Relative metrics show us the difference in a team’s performance when a player is on the ice versus when they’re off the ice.

If a forward possesses all four of the above things over the past two seasons, then they are considered Elite. To show the process of finding the true elites, we’ll be taking a look at the Flyers’ 2018-19 roster.

Using Natural Stat Trick’s database to look at scoring rates, we find five Flyers that reach the points per 60 threshold; Claude Giroux, James van Riemsdyk, Jakub Voracek, Sean Couturier, and Travis Konecny. Easy enough. Moving on to the time on ice cutoff, they found that an average time on ice north of 17:05 would get a forward into the 75th percentile. Three seasons later it may not be exactly the same, but it’s safe to assume that the figure hasn’t moved much at all given that there’s still 60 minutes in a full regulation game, plus the possibility of five minutes of overtime. If we wanted to be precise we’d have to find the exact cutoff over the last two seasons, but lucky for us our five Flyers’ average ice times aren’t near the 17:05 mark. Konecny, averaging 15:08 per game, and van Riemsdyk, averaging 15:55, fall short of the mark while Giroux, Voracek, and Couturier all clear the bar with ease.

Now that we have three Flyers remaining, it’s time to apply the final two filters; the relative shot metrics. We hit a bit of a road block here as we’re unable to combine data from multiple seasons, but we do have the ability to eyeball it and come up with an answer. At the time, to be in the 40th percentile of forwards you’d need to have a CF% relative above -1.2%. Again, to be precise we’d need to find the exact cutoff, but for this exercise we’ll move forward with that figure. With a very nice +6.9% CF relative in 2018-19, and a lesser but still very good +5.5% in 2017-18, Couturier moves forward. As does Giroux with a +5% CF relative last season, and a +4.9 the season prior. While not as strong, Voracek, whose figures over the last two seasons were -0.3% and +1.2%, makes the cut as well and moves onto the final stage.

We’re onto the final stage and let’s just make it simple; we can very quickly see that all three of Couturier (+7.1%, +5.1%), Giroux (+1.6%, +5.6%), and Voracek (+1.5%, +2.6%) have a DFF% relative over the last two seasons well above the 40th percentile.

Ask any Flyers fan who the “elite” forwards on the team are and generally you’d receive the same answer that the filters spit out: Giroux, Couturier and Voracek. Although some would surely argue over Voracek being included. Still, if the glove fits, the glove fits.

Now, at the opposite end of the spectrum, to be classified as a Gritensity forward you need to check off just one of the following three boxes.

If you:

  1. Played in at least 20 games and were among the bottom 16.7th percentile of players in relative Corsi.
  2. Scored at a per 60 rate that has you in the bottom 16.7th percentile of players.
  3. Are among the bottom 10% of players in average time on ice.

At a quick glance, average ice time alone has Tyrell Goulbourne, Nicolas Aube-Kubel, Phil Varone, and Mikhail Vorobyev qualifying as gritensity forwards. Others, like Jori Lehtera and Corban Knight, qualify through either their scoring rates or their relative Corsi. (Note: this is not a complete list.)

Yup, that checks out. Keep in mind that a gritensity skater is not automatically a “bad” NHLer. Many can be considered that, but some are just rookies who have yet to earn their coach’s trust yet, and thus haven’t been able to show their true talent level.

If a player does not meet the four requirements to be considered elite and nor do they qualify as a gritensity player, then they, like the majority of the NHL, are in the middle.

Since there are three forwards on the ice during 5-on-5 play, there are a couple different combinations of players that can be on the ice at a given time. To decide the quality of a player’s competition, the rule is that if there is at least one elite forward on the ice and no gritensity forward, the competition level is elite. With the inverse of that being true as well. If both or neither type of forward are on the ice, the competition level is middle.

While this data does tell us a lot, it’s important to remember that the quality of a player’s teammates has a massive impact on their on-ice results as well.

What it says about the Flyers’ defensemen

Now that we know what Elite, Middle, and Gritensity competition looks like, we can see how the Flyers performed with each of their defensemen on the ice against exact competition levels.

Unsurprisingly, Ivan Provorov leads the way in time spent against elite competition. In fact, among league-wide defensemen he sits 4th in total ice time and 16th in time on ice percentage against the best of the best. Last season 151 defensemen played at least 250 minutes versus elite competition and among them Provorov’s 48.48 CF% ranks 62nd and his 50.17 DFF% ranks 49th. Within the top half in Corsi and the top third in weighted shots — pretty good! And things look even better when you take into account how the Flyers did with him on the ice compared to when he was off of the ice.

His +1.6 CF% relative is good for 53rd among eligible defenders and his +4.4 DFF% relative takes him all the way to 19th in the league. Now, when comparing players on multiple teams with relative metrics it’s important to keep in mind how well their teams performed via each stat. For example, Troy Stecher’s +7.4 DFF% relative has him third in the league and is just over two and half percentage points higher than Zdeno Chara’s +4.8%. While it’s obviously a positive sign that Stecher was able to improve the Canucks’ weighted shot share by such a large margin, when you compare his 47.34 DFF% to Chara’s 54.05%, the added context better illustrates who had the better performance. The key here is to make sure we don’t undervalue a player’s impacts because they played on a good team, or overvalue a player’s impacts because they played on a lesser team. There’s a balance, and the Flyers weren’t exactly a powerhouse last season so the relative metrics for their top skaters will look even better because of it.

However, at the team level we don’t have to worry about doing that.

As previously stated, Provorov sits 19th in the league in DFF% relative against elite competition. But as you can probably tell, he was not the Flyers’ lead defender here — that would be Travis Sanheim. With a +6.6 DFF% relative, Sanheim is bested by only five other eligible defenders in the league to go along with his 29th-best 52.1% DFF%. More on his impressive numbers shortly. Seeing that they were paired together for the majority of the second half of the season, it’s fair to say that the two had a beneficial partnership, and that Provorov’s results were likely hurt by his previous partners. Even then he still had the second-highest DFF% relative of the six defensemen.

The interesting part of his statistical profile to me is that Flyers owned less of the dangerous shot share when he was on the ice against middling competition than any Flyers defenseman not named Robert Hagg. To figure out if this was an outlier or a trend let’s go through his first three seasons in the league:

Provorov’s DFF% Relative Year to Year

Competition 16-17 17-18 18-19
Competition 16-17 17-18 18-19
Elite 0.1 3.5 4.4
Middle -3.3 -0.2 -2.5
Gritensity -5.0 -2.3 2.5

Huh, it is in fact not an outlier! Interestingly enough Provorov has historically made the biggest, most positive impacts while up against the elites of the hockey world while falling behind his teammates versus middling and below competition. You’d certainly rather have it that way than the opposite given his role. But whether you chalk it up to a statistical oddity, partner effects, or there being some kind of tangible reason like play style which leads to the good-not-great forwards being able to have more success against him, it’s a weird one. Up until this past season even the fourth liners were able to find more success against him than others, just as they did when Shayne Gostisbehere was on the ice this year.

My first thought when I saw his -5.5 DFF% relative against gritensity forwards was that maybe he struggled against heavy forechecking pressure — something that most fourth lines will apply. But a closer look at his year to year results shows that he had positive impacts over the first three seasons of his career and puts that question to rest. If it happens to continue moving forward, maybe there’s something there, but it seems likely that this was simply a one-off.

Outside of his rookie season, Gostisbehere has fared well against elite competition, even during his two down-years. This past season he was buried by a team-low 95.5 PDO versus elite forwards, leading to a -9 goal differential — another team-low. The Flyers owned just over 48% of the dangerous shot share while he was on the ice, yet only owned 36% of the goals that were scored. In 2017-18 Ghost’s DFF% against elite forwards was actually just over half a percentage point lower than it was last season, but he had a 52.63 GF%. Regression comes for us all.

Overall, Gostisbehere’s underlying numbers have been relatively consistent outside of a few blips here and there, and if his on-ice shooting percentage happens to swing the opposite way next season, the goal results will follow suit.

Then there’s Radko Gudas who, outside of an extremely strong 2015-16 campaign, hasn’t had a big impact positively or negatively against elite forwards over the last three seasons. There’s really not much here; Gudas beat up on the gritensity forwards and treaded water relative to his team versus middle and up competition this season. When it comes to Flyers defensemen, he’s generally been somewhere in the middle throughout his time with the organization. And that’s not bad!

Using Gudas as the divider between good results and bad results, these next two clearly fall below him. Through his first two full seasons in the league Hagg has had negative results versus all three competition levels by both Corsi and Dangerous Fenwick. That’s, well, not ideal. Andrew MacDonald on the other hand has some positives, including a +1.4 DFF% relative versus middling competition just this past season. Plus, three seasons ago — Provorov’s rookie season — MacDonald graded out quite well against elite competition with a +4.3 DFF% relative. But overall, and especially more recently, the positives are few and far between. With a lot less ice time against tough competition, MacDonald was able to post a positive relative Corsi for the first time in his Flyers career in 2018-19.

Last but certainly not least, we finally arrive at Sanheim. And boy are his numbers impressive. Even in his rookie season where he drew the ire of then-head coach Dave Hakstol enough to warrant an extended stay in the press box, the Flyers routinely owned a higher percentage of the shot share against the elite forwards with Sanheim on the ice versus when he was off the ice. But that’s old, unsurprising news and what’s really intriguing about Sanheim’s numbers this past season is that he excelled at both shot generation and shot suppression at the highest level — something no other Flyers blueliner achieved.

Sanheim’s 49.89 CF% and 52.1 DFF% against elite competition ranks first on the team, and when you break it down by offense and defense one thing becomes clear; Sanheim posted some of the best shot rates versus elite forwards across the entire league. Of the 151 eligible defensemen he was one of just 19 to have an above average impact on both shot generation and shot suppression rates by both Corsi and Dangerous Fenwick against elite forwards. Furthermore, only five defensemen had better on-ice Dangerous Fenwick generation and Dangerous Fenwick suppression rates than Sanheim. Now, it’ll take more than one full season to show us that he can actually replicate such a high level of play, but that is incredibly encouraging.

Just how good Sanheim was last season becomes even clearer by putting his name next to other teams’ defensemen.

Provorov looks to have had a better defensive impact, and Gostisbehere looks to have had a better offensive impact, but neither had a stronger combination of the two than Sanheim. His numbers truly have him in elite company with the likes of P.K. Subban, Seth Jones, and Ryan Suter. The only negative relative stat over the last two seasons for Sanheim is his DFF% against gritensity forwards this season — in every other situation the Flyers were a better team when Sanheim was on the ice. We’re just 131 games into his NHL career and he’s already having this kind of impact not only overall, but against the best forwards in the league.

Make sure to check out PuckIQ and their competition levels; I really believe that this is a large step in the right direction when it comes to analyzing which skaters not only play against the toughest competition, but also perform the best against them. The information we’ll have to work with will only improve from here, and it’s always exciting to have something new to look at.