Jan Levine: Top off-season questions continue: Analytics versus the eye test A polarizing topic to say the least

Hi Jan ----

Your blog is referencing an Athletic article that we cannot see without a subscription. I have no idea what 10 players are being discussed...

The Athletic
team
league
NHL
NHL

Teams
Scores & Schedule
Standings
• • •
Ten worst contracts
Offseason grades
Ten NHL players analytics were wrong about and the lessons learned

LOS ANGELES, CALIFORNIA - MARCH 04: Drew Doughty #8 of the Los Angeles Kings turns with the puck during a 4-2 Kings win over the St. Louis Blues over the St. Lous Blues at Crypto.com Arena on March 04, 2023 in Los Angeles, California. (Photo by Harry How/Getty Images)
By Harman Dayal
Jul 27, 2023
190

Save Article
NHL analytics have exploded in popularity over the last decade.

Teams are investing in staff and building departments for the sole purpose of analyzing data and in the public sphere, it’s significantly enhanced the information and knowledge that fans and media have about players and teams.

The majority of the hockey world now agrees that data has some value as a tool that can offer objective insights into player performance. The old days of intense analytics versus eye test debates are mostly gone because people understand that it’s not an either-or proposition, you should be using both the eye test and data.

ADVERTISEMENT


Hockey fans have a more sophisticated understanding of players around the NHL — especially players from teams they don’t watch — because of how accessible public analytics are.

That, of course, should be celebrated.

But if we’re going to rely on these tools, we should also be aware of their limitations and the cases where stats can be misleading.

I’ve been diving into analytical tools like shot shares, expected goals and PDO; types of microstats like zone exits, zone entries and passing data that Corey Sznajder tracks; powerful tools like Dom Luszczyszyn’s excellent Net Rating model and more for years now. Today, I wanted to highlight examples where analytics were misleading about a player’s true value.

The purpose of this exercise isn’t to blame analytics. These are often situations where I was personally wrong in putting too much stock in the data and not considering other factors.

Experience over time has made me handle analytics with more scrutiny.

Instead of drawing conclusions at face value because a model feels strongly about a player, I try and find the holes, blind spots and possible ways I could be wrong. When a player I need to offer deep analysis on is acquired, I’ll watch hours of game film and dissect those observations to add a new perspective. I weigh a lot more qualitative factors now. And honestly, it all helps paint a clearer, more nuanced understanding of players.

Here are 10 players that analytics were misleading about, and some of the lessons I’ve learned as a result.

Chris Tanev
Chris Tanev has been one of the NHL’s best shutdown defensemen for a long time. He’s excellent positionally, has special defensive awareness and instincts, a disruptive stick, a long reach and a fearless attitude.

Tanev was an analytics darling for most of his Canucks tenure. At the end of the 2019-20 season, Vancouver had to decide whether to re-sign him to a long-term deal or let him walk as an unrestricted free agent. Tanev was on the cusp of turning 31, had durability issues (before 2019-20, his games played totals were 55, 42 and 53 going back to the 2016-17 season) and while he was still an effective top-four defenceman, his two-way metrics were on the decline.

Tanev’s isolated impact on helping the Canucks control five-on-five shots, for example, had diminished considerably from his peak. He’d gone from an elite driver in his prime to looking below average.



Evolving-Hockey’s RAPM tool

Colleague Dom Luszczyszyn’s Game Score Value Added (GSVA) model projected that Tanev was only two or three years away from being a replacement-level contributor.



Tanev was a great fit with Hughes, but committing four or five years of term to an oft-injured 30-year-old defender with declining numbers seemed very risky. I thought the Canucks were wise to let him walk to Calgary. And, of course, I was totally wrong.

Tanev immediately resurrected his game, posting dominant two-way numbers and re-establishing himself as one of the best defensive defencemen in the NHL. The Canucks, meanwhile, have desperately missed Tanev’s defensive reliability at even strength and the penalty kill.

There are two main lessons I gained from this.

No. 1: The environment a defender plays in can influence their numbers a lot. Tanev went from a club that generally struggled to control play and had far less two-way help from forwards in Vancouver to an excellent possession team in Calgary. Even the stats that supposedly isolate or adjust for those teammate- and environment-related factors don’t do a sufficient job of accounting for those differences.

No. 2: Defencemen — especially stay-at-home types — are harder to properly evaluate using analytics than forwards because it’s much harder to statistically measure defensive impact.

Neal Pionk and Jacob Trouba
In 2019, the Jets traded Jacob Trouba to the Rangers for Neal Pionk and a first-round pick. Trouba was coming off a 50-point season, played shutdown minutes alongside Josh Morrissey and profiled like a bona fide top-pair defender. Pionk, on the other hand, was a young player still trying to establish himself.

I thought it was a sweet deal for the Rangers and a disappointing haul for Winnipeg. The opposite turned out to be true.

Trouba hasn’t lived up to that potential and is now overpaid, owning an $8 million AAV that ties him for the 14th-highest cap hit among all NHL defensemen. Pionk, meanwhile, immediately blossomed in Winnipeg.

This is another reminder of how significantly a change of scenery can influence a defender’s value.

As for Pionk, he had promising offensive tools but was flawed defensively and his underlying numbers were ugly. In hindsight, there was a plausible explanation for his poor analytical profile.

Pionk was coming off his first full NHL season and yet he was thrust into a top-pairing role alongside Marc Staal, who was rapidly declining for a large chunk of it. Almost every rookie offensive defenseman would struggle if he had to match up against the opposition’s best players, with a subpar partner, on a bad, rebuilding team. Pionk had also gotten hurt in the middle of the season and lost his confidence.

With all of these factors at play, Pionk’s first NHL season in New York shouldn’t have been graded nearly as harshly. When a defender has bad numbers, we should dive deeper into who their partner was, what kind of club they were on, the difficulty of the role they were playing and their experience.

In Trouba’s case, it’s not that the analytics were wrong, but more that the monster final year he had in Winnipeg was an outlier.

Trouba ran the Jets’ first-unit power play in 2018-19 and ranked 13th among all NHL defensemen in points, ahead of players like Erik Karlsson, Drew Doughty and Zach Werenski. That created an expectation that he could be a high-end offensive piece for the Rangers, despite the fact 33 points was his career high before that final year with the Jets.

Tage Thompson
An abnormally high shooting percentage during a single season is one of the most common signs that a player’s goal-scoring rate could slow down moving forward. Over multi-season stretches, even many of the top snipers struggle to convert on 16-17 percent or more of their shots. That means when a player like Chris Kreider, for example, scores 52 goals in 2021-22 on the back of a 20.2 percent shooting clip, we usually hypothesize that he’ll slow down the following year, as he did last season with 36 goals.

There are countless examples of teams that got burned by making bets on players with inflated shooting percentages.

When Tage Thompson broke out for 38 goals in 2021-22, it seemingly came out of nowhere. I still believed in Thompson as a player, and didn’t mind the big extension the Sabres signed him to, but heading into last season, I figured he’d be closer to a consistent 30-goal scorer rather than flirting with 40.

Why? Well, Thompson had nearly tripled his previous five-on-five shooting clip in 2021-22 and hadn’t flashed elite goal-scoring potential at lower levels, especially at even strength. As The Athletic’s Jonathan Willis had pointed out, Thompson had a pedestrian 10.6 career shooting percentage in the AHL against minor-league goalies and in his draft year, he scored just one even-strength goal in 36 games for the University of Connecticut.

Thompson made that regression talk look silly by exploding for 47 goals last season. Of course, there are factors such as Thompson’s positional shift to center, improved linemate quality, confidence, skill and rare 6-foot-6 frame that probably weren’t weighed heavily enough.

But there’s also the point that Thompson’s 15 percent shooting clip from 2021-22 was high but not crazy. This wasn’t a case of a player scoring at a ridiculous 25 percent clip; 15 percent is actually somewhat reasonable for an elite sniper to maintain year in, year out. We just weren’t sure he could maintain it because Thompson’s previous track record didn’t hint at star potential.

The lesson isn’t to ignore shooting percentage as a signal. That would be very foolish and lead to tons of regrettable mistakes — Thompson is a very special case and the overwhelming majority of players won’t follow his career arc. Rather, the lesson is that we shouldn’t go too far on the other side of the pendulum either in blindly assuming a player with a spiked shooting percentage is bound to regress. Occasionally, there’s more than meets the eye.

Vladislav Gavrikov
Vladislav Gavrikov caught my eye two or three years ago.

It was cool seeing this unheralded, hulking defender use his long reach and solid skating to disrupt plays, win battles down low and excel as a defensive-minded presence. But by the time the trade deadline rolled around and Gavrikov had emerged as a trade candidate, it seemed like somebody was at risk of overpaying based on all the hype around him.

Many stats-based analysts were dumping on Gavrikov, pointing to his putrid underlying numbers in Columbus. From experience, I wasn’t too worried about that because he was being thrown to the wolves as a No. 1 defenseman on a bad team as a result of Zach Werenski’s injury, defending top players, playing with a subpar partner and starting a ton of his shifts in the defensive zone.

It was easy to see him bouncing back once slotted into a more appropriate role on a better roster. I’d been wrong enough times and seen Gavrikov play enough to be less scathing, but a fair chunk of the public discourse characterized him as the next Ben Chiarot because of his analytical profile.

Gavrikov turned out to be the perfect fit for the Kings. He was a stud on the second pair, posted dominant play-driving results and drove a 14-6 five-on-five goal differential in L.A.’s favor. Once again, context and team situation play a huge role in determining an individual player’s numbers.

I should caution that not every high-profile defenseman ends up following this arc. I don’t want people to read about all these positive examples and assume every high-profile defender with bad numbers on a terrible club will end up with an enormously positive ending.

Oliver Ekman-Larsson is an example of a player with poor advanced stats on a terrible Coyotes roster — with tons of legitimate reasons to believe in a bounce back with a fresh start — who turned out to be a disastrous acquisition for the Canucks. Jeff Petry is another case — he never rebounded to his peak form when he moved from Montreal to Pittsburgh last season.

So yes, stats aren’t close to the be-all, end-all for evaluating defenders on bad teams, but they shouldn’t be totally ignored either.

Jason Richardinson
The Vancouver Canucks were searching for a defensive-minded third-line center heading into the 2021-22 season. The Seattle expansion draft was also looming that offseason and some contenders were looking to trade away valuable players they couldn’t protect in the draft and didn’t want to lose for nothing. Dallas made Jason Richardinson available because of that, and the Canucks swooped in to acquire him for a third-round pick.

Richardinson was limited offensively but he had a stellar defensive profile in his final season with the Stars. The rangy checking center helped Dallas seize a 59 percent control of five-on-five scoring chances. He was a top-10 forward in the NHL that season at suppressing expected goals against and the Stars surrendered just 1.96 goals against per hour with him on the ice. He was also one of the Stars’ most-used forwards on the penalty kill.

It wasn’t just the analytics that pointed to Richardinson’s value either — the coaching staff had immense trust. He ranked fourth among all Stars forwards in five-on-five ice time per game, averaging more five-on-five minutes than the likes of Jamie Benn and Roope Hintz.

Richardinson turned out to be a disastrous fit in Vancouver. His sturdy defensive game was nowhere to be found and his offensive performance was completely anemic. It went so poorly that last fall, the Canucks paid a second-round pick just to dump his contract to Chicago and pick up Riley Stillman.

What I (and the Canucks) didn’t account for was that Richardinson’s game was a unique fit in Dallas. The Stars played an insufferably boring, low-event style that promoted defense above all under Rick Bowness at the time. That was the perfect fit for Richardinson, who offers no offense but has some redeemable checking qualities. The Canucks, like most NHL organizations, played a very different structure and tempo than the Stars.

The mistake was assuming he’d be the same player in a totally different system, on a significantly weaker defensive team.


Jason Richardinson. (Sergei Belski / USA Today)
Colin Miller
Colin Miller’s name hit the rumor mill in the summer of 2019 when he was a member of the Vegas Golden Knights. Vegas needed to clear money and Miller, with three years remaining on a $3.875 million cap hit at the time, was the odd man out.

At the time, it looked like a golden opportunity to affordably acquire a top-four right-shot defender. Miller averaged just shy of 20 minutes per game the last two seasons, he was a year removed from scoring 41 points and he owned strong play-driving numbers.

Why has it never worked out for him as a top-four defenseman?

The missing context is that his matchups were heavily sheltered. Miller wasn’t trusted defensively, so coaches mostly deployed him against the opposition’s third and fourth lines.

When you acquire a bottom pairing who’s delivered excellent results in a sheltered capacity, you’re rolling the dice if you expect them to succeed further up the lineup. Some like Vince Dunn, Nate Schmidt and Mackenzie Weegar turn out to be the real deal. But others like Miller can’t take the next step.

Drew Doughty
Drew Doughty is the second-highest-paid defenseman in the NHL. It’s pretty widely accepted he isn’t quite worth the full freight of his $11 million cap hit. But two or three years ago, we reached a point where many in the analytics community weren’t just arguing Doughty was overpaid. Many were arguing he was washed up.

I remember coming across Jack Han’s fantastic article in the middle of the Doughty wars. Han broke down some game film and had a fascinating theory that changed my perspective at the time. He showed many revealing video clips and concluded that Doughty still has “elite tools and ability to take over a game, but seems to lack the motivation or inspiration to do hard things every shift in order to help his team control play.”

The takeaway was that Doughty still had the skill set to be a strong top-pair player but perhaps lacked intensity because he was on a losing, rebuilding team. Sure enough, the Kings have made the playoffs the last two years, and in that meaningful environment, Doughty is playing excellent hockey again. He wasn’t washed up after all.

Jesse Puljujarvi
Jesse Puljujarvi was one of the Oilers’ most polarizing players in recent memory.

The big Finnish winger was hot out of the gate in 2021-22 with 10 goals and 23 points in his first 28 games, playing alongside Connor McDavid. But then he missed action due to COVID-19, lost his rhythm, suffered a lower-body injury in February and found himself in the bottom six, notching just four goals in his final 37 games.

Puljujarvi was all over the rumor mill last summer and that’s around the time the debates in Edmonton reached a fever pitch.

Many went to bat for Puljujarvi, pointing out that his play-driving numbers were really strong, with his puck retrievals and forechecking helping any line he was on control possession and generate a ton of scoring chances. They pointed out he scored at a 45-point-per-82-game pace in 2021-22 and made the case he’s a solid middle-six player. Others fought back and claimed his scoring chance generation doesn’t really matter if he’s a terrible finisher and that he doesn’t have the IQ to make plays with the puck.

Puljujarvi was re-signed to a one-year, $3 million last summer but it never worked out. He continued whiffing on chances last season, completely lost his confidence and his lack of penalty-killing utility meant he didn’t fit the prototypical profile of a defensive bottom-six winger.

Travis Dermott
When Travis Dermott first arrived on the scene in Toronto, there was a lot of hope for him as a future top-four defender.

He was mobile, athletic, a slick puck-mover and immediately crushed his bottom-pair assignments with strong two-way metrics at just 21 years of age. Some even argued Dermott should be fed more minutes in the 2018 playoffs and that Ron Hainsey, who was overmatched on the top pair, should have his workload reduced.

When you make such an impressive first impression in the NHL at 21, it’s only natural fans expect there’s another level to reach.

Dermott, unfortunately, couldn’t hit that potential. Like Colin Miller, this is the case of a solid third-pair defender with excellent analytics who simply couldn’t translate his game further up the lineup in non-sheltered minutes. He was too mistake-prone and couldn’t be trusted defensively.

I’ve learned to put even more stock into the difficulty of a player’s matchups. The head coach is often right when he decides that the promising young No. 6 defender with amazing analytics in a sheltered role doesn’t deserve a larger role.

Conclusion
Analytics are useful for telling you what results are happening when a player is on the ice (e.g. the team dominates opponents in terms of possession and scoring chances when Player X is on the ice, or the club gets caved in and bleeds a lot defensively when Player Y is on the ice). It provides objective data and helps account for biases. The eye test — and other contextual and qualitative factors — tell you why those results happened and what it actually reveals about the player’s ability, skill set and value.

Analytical models have often been right when they’ve suggested this player is overrated or that player is underrated. I truly hope people don’t take this article as an opportunity to bash analytics or the work of a smart, talented colleague like Luszczyszyn because the hockey public’s overall knowledge and conversations around players are sharper and more sophisticated due to these tools.

The point is that the numbers alone sometimes point you in the wrong direction.

Anytime we evaluate a player, we should accept their statistical profile as valuable information but then dive deeper into the possible ways it might not be telling the entire story. That’d be much better than drawing rash, immediate conclusions when a player’s analytical profile card goes viral on Twitter.

(Photo of Drew Doughty: Harry How / Getty Images)

What did you think of this story?

Rate as MEH
MEH

Rate as SOLID
SOLID

Rate as AWESOME
AWESOME


Harman Dayal is a staff writer for The Athletic covering the Vancouver Canucks. He combines NHL video and data analysis and tracks microstats as part of his coverage. Follow Harman on Twitter @harmandayal2

OPEN IN APP
National
Boxing
Bundesliga
Champions League
Championship
College Football
Copa del Rey
Culture
Europa League
European Championship
FA Cup
Fantasy Baseball
Fantasy Basketball
Fantasy Football
Fantasy Hockey
Fantasy Premier League
Formula 1
Gaming
Golf
International Soccer
La Liga
League Cup
League One
League Two
LNH
Men's College Basketball
Men's World Cup
Mixed Martial Arts
MLB
MLS
Motorsports
NBA
NFL
NHL
NWSL
Olympics
Opinion
Premier League
Scottish Premiership
Serie A
Soccer
Sports Betting
Sports Business
Tennis
UK Women's Football
WNBA
Women's College Basketball
Women's Euros
Women's Hockey
Women's World Cup
NCAA Women's Basketball
The Athletic Ink
Podcasts
Headlines
Real Time
US
Arizona
Atlanta
Baltimore
Bay Area
Boston
Buffalo
Carolina
Chicago
Cincinnati
Cleveland
Columbus
Dallas
Denver
Detroit
Houston
Indiana
Jacksonville
Kansas City
Las Vegas
Los Angeles
Memphis
Miami
Minnesota
Nashville
New Orleans
New York
Oklahoma
Oregon
Orlando
Philadelphia
Pittsburgh
Sacramento
San Antonio
San Diego
Seattle
St. Louis
Tampa Bay
Utah
Washington DC
Wisconsin
Canada
Calgary
Edmonton
Montreal
Montréal (français)
Ottawa
Toronto
Vancouver
Winnipeg
Partners
Tickets by StubHub
Share
Buy a Gift
HQ
About Us
Careers
Code of Conduct
Editorial Guidelines
Business Inquiries
Press Inquiries
Support
FAQ
Forgot Password?
Redeem Gift
Contact Us
Terms of Service
Group Subscriptions
Newsletters
The Pulse
The Bounce
The Windup
Prime Tire
Full Time
©2023 The Athletic Media Company. All rights reserved.
Privacy Policy
Support
Sitemap
Twitter
Facebook
Instagram

Download on the App Store

Get it on Google Play

The Athletic
team
league
NHL
NHL

Teams
Scores & Schedule
Standings
• • •
Ten worst contracts
Offseason grades
Ten NHL players analytics were wrong about and the lessons learned

LOS ANGELES, CALIFORNIA - MARCH 04: Drew Doughty #8 of the Los Angeles Kings turns with the puck during a 4-2 Kings win over the St. Louis Blues over the St. Lous Blues at Crypto.com Arena on March 04, 2023 in Los Angeles, California. (Photo by Harry How/Getty Images)
By Harman Dayal
Jul 27, 2023
190

Save Article
NHL analytics have exploded in popularity over the last decade.

Teams are investing in staff and building departments for the sole purpose of analyzing data and in the public sphere, it’s significantly enhanced the information and knowledge that fans and media have about players and teams.

The majority of the hockey world now agrees that data has some value as a tool that can offer objective insights into player performance. The old days of intense analytics versus eye test debates are mostly gone because people understand that it’s not an either-or proposition, you should be using both the eye test and data.

ADVERTISEMENT


Hockey fans have a more sophisticated understanding of players around the NHL — especially players from teams they don’t watch — because of how accessible public analytics are.

That, of course, should be celebrated.

But if we’re going to rely on these tools, we should also be aware of their limitations and the cases where stats can be misleading.

I’ve been diving into analytical tools like shot shares, expected goals and PDO; types of microstats like zone exits, zone entries and passing data that Corey Sznajder tracks; powerful tools like Dom Luszczyszyn’s excellent Net Rating model and more for years now. Today, I wanted to highlight examples where analytics were misleading about a player’s true value.

The purpose of this exercise isn’t to blame analytics. These are often situations where I was personally wrong in putting too much stock in the data and not considering other factors.

Experience over time has made me handle analytics with more scrutiny.

Instead of drawing conclusions at face value because a model feels strongly about a player, I try and find the holes, blind spots and possible ways I could be wrong. When a player I need to offer deep analysis on is acquired, I’ll watch hours of game film and dissect those observations to add a new perspective. I weigh a lot more qualitative factors now. And honestly, it all helps paint a clearer, more nuanced understanding of players.

Here are 10 players that analytics were misleading about, and some of the lessons I’ve learned as a result.

Chris Tanev
Chris Tanev has been one of the NHL’s best shutdown defensemen for a long time. He’s excellent positionally, has special defensive awareness and instincts, a disruptive stick, a long reach and a fearless attitude.

Tanev was an analytics darling for most of his Canucks tenure. At the end of the 2019-20 season, Vancouver had to decide whether to re-sign him to a long-term deal or let him walk as an unrestricted free agent. Tanev was on the cusp of turning 31, had durability issues (before 2019-20, his games played totals were 55, 42 and 53 going back to the 2016-17 season) and while he was still an effective top-four defenceman, his two-way metrics were on the decline.

Tanev’s isolated impact on helping the Canucks control five-on-five shots, for example, had diminished considerably from his peak. He’d gone from an elite driver in his prime to looking below average.



Evolving-Hockey’s RAPM tool

Colleague Dom Luszczyszyn’s Game Score Value Added (GSVA) model projected that Tanev was only two or three years away from being a replacement-level contributor.



Tanev was a great fit with Hughes, but committing four or five years of term to an oft-injured 30-year-old defender with declining numbers seemed very risky. I thought the Canucks were wise to let him walk to Calgary. And, of course, I was totally wrong.

Tanev immediately resurrected his game, posting dominant two-way numbers and re-establishing himself as one of the best defensive defencemen in the NHL. The Canucks, meanwhile, have desperately missed Tanev’s defensive reliability at even strength and the penalty kill.

There are two main lessons I gained from this.

No. 1: The environment a defender plays in can influence their numbers a lot. Tanev went from a club that generally struggled to control play and had far less two-way help from forwards in Vancouver to an excellent possession team in Calgary. Even the stats that supposedly isolate or adjust for those teammate- and environment-related factors don’t do a sufficient job of accounting for those differences.

No. 2: Defencemen — especially stay-at-home types — are harder to properly evaluate using analytics than forwards because it’s much harder to statistically measure defensive impact.

Neal Pionk and Jacob Trouba
In 2019, the Jets traded Jacob Trouba to the Rangers for Neal Pionk and a first-round pick. Trouba was coming off a 50-point season, played shutdown minutes alongside Josh Morrissey and profiled like a bona fide top-pair defender. Pionk, on the other hand, was a young player still trying to establish himself.

I thought it was a sweet deal for the Rangers and a disappointing haul for Winnipeg. The opposite turned out to be true.

Trouba hasn’t lived up to that potential and is now overpaid, owning an $8 million AAV that ties him for the 14th-highest cap hit among all NHL defensemen. Pionk, meanwhile, immediately blossomed in Winnipeg.

This is another reminder of how significantly a change of scenery can influence a defender’s value.

As for Pionk, he had promising offensive tools but was flawed defensively and his underlying numbers were ugly. In hindsight, there was a plausible explanation for his poor analytical profile.

Pionk was coming off his first full NHL season and yet he was thrust into a top-pairing role alongside Marc Staal, who was rapidly declining for a large chunk of it. Almost every rookie offensive defenseman would struggle if he had to match up against the opposition’s best players, with a subpar partner, on a bad, rebuilding team. Pionk had also gotten hurt in the middle of the season and lost his confidence.

With all of these factors at play, Pionk’s first NHL season in New York shouldn’t have been graded nearly as harshly. When a defender has bad numbers, we should dive deeper into who their partner was, what kind of club they were on, the difficulty of the role they were playing and their experience.

In Trouba’s case, it’s not that the analytics were wrong, but more that the monster final year he had in Winnipeg was an outlier.

Trouba ran the Jets’ first-unit power play in 2018-19 and ranked 13th among all NHL defensemen in points, ahead of players like Erik Karlsson, Drew Doughty and Zach Werenski. That created an expectation that he could be a high-end offensive piece for the Rangers, despite the fact 33 points was his career high before that final year with the Jets.

Tage Thompson
An abnormally high shooting percentage during a single season is one of the most common signs that a player’s goal-scoring rate could slow down moving forward. Over multi-season stretches, even many of the top snipers struggle to convert on 16-17 percent or more of their shots. That means when a player like Chris Kreider, for example, scores 52 goals in 2021-22 on the back of a 20.2 percent shooting clip, we usually hypothesize that he’ll slow down the following year, as he did last season with 36 goals.

There are countless examples of teams that got burned by making bets on players with inflated shooting percentages.

When Tage Thompson broke out for 38 goals in 2021-22, it seemingly came out of nowhere. I still believed in Thompson as a player, and didn’t mind the big extension the Sabres signed him to, but heading into last season, I figured he’d be closer to a consistent 30-goal scorer rather than flirting with 40.

Why? Well, Thompson had nearly tripled his previous five-on-five shooting clip in 2021-22 and hadn’t flashed elite goal-scoring potential at lower levels, especially at even strength. As The Athletic’s Jonathan Willis had pointed out, Thompson had a pedestrian 10.6 career shooting percentage in the AHL against minor-league goalies and in his draft year, he scored just one even-strength goal in 36 games for the University of Connecticut.

Thompson made that regression talk look silly by exploding for 47 goals last season. Of course, there are factors such as Thompson’s positional shift to center, improved linemate quality, confidence, skill and rare 6-foot-6 frame that probably weren’t weighed heavily enough.

But there’s also the point that Thompson’s 15 percent shooting clip from 2021-22 was high but not crazy. This wasn’t a case of a player scoring at a ridiculous 25 percent clip; 15 percent is actually somewhat reasonable for an elite sniper to maintain year in, year out. We just weren’t sure he could maintain it because Thompson’s previous track record didn’t hint at star potential.

The lesson isn’t to ignore shooting percentage as a signal. That would be very foolish and lead to tons of regrettable mistakes — Thompson is a very special case and the overwhelming majority of players won’t follow his career arc. Rather, the lesson is that we shouldn’t go too far on the other side of the pendulum either in blindly assuming a player with a spiked shooting percentage is bound to regress. Occasionally, there’s more than meets the eye.

Vladislav Gavrikov
Vladislav Gavrikov caught my eye two or three years ago.

It was cool seeing this unheralded, hulking defender use his long reach and solid skating to disrupt plays, win battles down low and excel as a defensive-minded presence. But by the time the trade deadline rolled around and Gavrikov had emerged as a trade candidate, it seemed like somebody was at risk of overpaying based on all the hype around him.

Many stats-based analysts were dumping on Gavrikov, pointing to his putrid underlying numbers in Columbus. From experience, I wasn’t too worried about that because he was being thrown to the wolves as a No. 1 defenseman on a bad team as a result of Zach Werenski’s injury, defending top players, playing with a subpar partner and starting a ton of his shifts in the defensive zone.

It was easy to see him bouncing back once slotted into a more appropriate role on a better roster. I’d been wrong enough times and seen Gavrikov play enough to be less scathing, but a fair chunk of the public discourse characterized him as the next Ben Chiarot because of his analytical profile.

Gavrikov turned out to be the perfect fit for the Kings. He was a stud on the second pair, posted dominant play-driving results and drove a 14-6 five-on-five goal differential in L.A.’s favor. Once again, context and team situation play a huge role in determining an individual player’s numbers.

I should caution that not every high-profile defenseman ends up following this arc. I don’t want people to read about all these positive examples and assume every high-profile defender with bad numbers on a terrible club will end up with an enormously positive ending.

Oliver Ekman-Larsson is an example of a player with poor advanced stats on a terrible Coyotes roster — with tons of legitimate reasons to believe in a bounce back with a fresh start — who turned out to be a disastrous acquisition for the Canucks. Jeff Petry is another case — he never rebounded to his peak form when he moved from Montreal to Pittsburgh last season.

So yes, stats aren’t close to the be-all, end-all for evaluating defenders on bad teams, but they shouldn’t be totally ignored either.

Jason Richardinson
The Vancouver Canucks were searching for a defensive-minded third-line center heading into the 2021-22 season. The Seattle expansion draft was also looming that offseason and some contenders were looking to trade away valuable players they couldn’t protect in the draft and didn’t want to lose for nothing. Dallas made Jason Richardinson available because of that, and the Canucks swooped in to acquire him for a third-round pick.

Richardinson was limited offensively but he had a stellar defensive profile in his final season with the Stars. The rangy checking center helped Dallas seize a 59 percent control of five-on-five scoring chances. He was a top-10 forward in the NHL that season at suppressing expected goals against and the Stars surrendered just 1.96 goals against per hour with him on the ice. He was also one of the Stars’ most-used forwards on the penalty kill.

It wasn’t just the analytics that pointed to Richardinson’s value either — the coaching staff had immense trust. He ranked fourth among all Stars forwards in five-on-five ice time per game, averaging more five-on-five minutes than the likes of Jamie Benn and Roope Hintz.

Richardinson turned out to be a disastrous fit in Vancouver. His sturdy defensive game was nowhere to be found and his offensive performance was completely anemic. It went so poorly that last fall, the Canucks paid a second-round pick just to dump his contract to Chicago and pick up Riley Stillman.

What I (and the Canucks) didn’t account for was that Richardinson’s game was a unique fit in Dallas. The Stars played an insufferably boring, low-event style that promoted defense above all under Rick Bowness at the time. That was the perfect fit for Richardinson, who offers no offense but has some redeemable checking qualities. The Canucks, like most NHL organizations, played a very different structure and tempo than the Stars.

The mistake was assuming he’d be the same player in a totally different system, on a significantly weaker defensive team.


Jason Richardinson. (Sergei Belski / USA Today)
Colin Miller
Colin Miller’s name hit the rumor mill in the summer of 2019 when he was a member of the Vegas Golden Knights. Vegas needed to clear money and Miller, with three years remaining on a $3.875 million cap hit at the time, was the odd man out.

At the time, it looked like a golden opportunity to affordably acquire a top-four right-shot defender. Miller averaged just shy of 20 minutes per game the last two seasons, he was a year removed from scoring 41 points and he owned strong play-driving numbers.

Why has it never worked out for him as a top-four defenseman?

The missing context is that his matchups were heavily sheltered. Miller wasn’t trusted defensively, so coaches mostly deployed him against the opposition’s third and fourth lines.

When you acquire a bottom pairing who’s delivered excellent results in a sheltered capacity, you’re rolling the dice if you expect them to succeed further up the lineup. Some like Vince Dunn, Nate Schmidt and Mackenzie Weegar turn out to be the real deal. But others like Miller can’t take the next step.

Drew Doughty
Drew Doughty is the second-highest-paid defenseman in the NHL. It’s pretty widely accepted he isn’t quite worth the full freight of his $11 million cap hit. But two or three years ago, we reached a point where many in the analytics community weren’t just arguing Doughty was overpaid. Many were arguing he was washed up.

I remember coming across Jack Han’s fantastic article in the middle of the Doughty wars. Han broke down some game film and had a fascinating theory that changed my perspective at the time. He showed many revealing video clips and concluded that Doughty still has “elite tools and ability to take over a game, but seems to lack the motivation or inspiration to do hard things every shift in order to help his team control play.”

The takeaway was that Doughty still had the skill set to be a strong top-pair player but perhaps lacked intensity because he was on a losing, rebuilding team. Sure enough, the Kings have made the playoffs the last two years, and in that meaningful environment, Doughty is playing excellent hockey again. He wasn’t washed up after all.

Jesse Puljujarvi
Jesse Puljujarvi was one of the Oilers’ most polarizing players in recent memory.

The big Finnish winger was hot out of the gate in 2021-22 with 10 goals and 23 points in his first 28 games, playing alongside Connor McDavid. But then he missed action due to COVID-19, lost his rhythm, suffered a lower-body injury in February and found himself in the bottom six, notching just four goals in his final 37 games.

Puljujarvi was all over the rumor mill last summer and that’s around the time the debates in Edmonton reached a fever pitch.

Many went to bat for Puljujarvi, pointing out that his play-driving numbers were really strong, with his puck retrievals and forechecking helping any line he was on control possession and generate a ton of scoring chances. They pointed out he scored at a 45-point-per-82-game pace in 2021-22 and made the case he’s a solid middle-six player. Others fought back and claimed his scoring chance generation doesn’t really matter if he’s a terrible finisher and that he doesn’t have the IQ to make plays with the puck.

Puljujarvi was re-signed to a one-year, $3 million last summer but it never worked out. He continued whiffing on chances last season, completely lost his confidence and his lack of penalty-killing utility meant he didn’t fit the prototypical profile of a defensive bottom-six winger.

Travis Dermott
When Travis Dermott first arrived on the scene in Toronto, there was a lot of hope for him as a future top-four defender.

He was mobile, athletic, a slick puck-mover and immediately crushed his bottom-pair assignments with strong two-way metrics at just 21 years of age. Some even argued Dermott should be fed more minutes in the 2018 playoffs and that Ron Hainsey, who was overmatched on the top pair, should have his workload reduced.

When you make such an impressive first impression in the NHL at 21, it’s only natural fans expect there’s another level to reach.

Dermott, unfortunately, couldn’t hit that potential. Like Colin Miller, this is the case of a solid third-pair defender with excellent analytics who simply couldn’t translate his game further up the lineup in non-sheltered minutes. He was too mistake-prone and couldn’t be trusted defensively.

I’ve learned to put even more stock into the difficulty of a player’s matchups. The head coach is often right when he decides that the promising young No. 6 defender with amazing analytics in a sheltered role doesn’t deserve a larger role.

Conclusion
Analytics are useful for telling you what results are happening when a player is on the ice (e.g. the team dominates opponents in terms of possession and scoring chances when Player X is on the ice, or the club gets caved in and bleeds a lot defensively when Player Y is on the ice). It provides objective data and helps account for biases. The eye test — and other contextual and qualitative factors — tell you why those results happened and what it actually reveals about the player’s ability, skill set and value.

Analytical models have often been right when they’ve suggested this player is overrated or that player is underrated. I truly hope people don’t take this article as an opportunity to bash analytics or the work of a smart, talented colleague like Luszczyszyn because the hockey public’s overall knowledge and conversations around players are sharper and more sophisticated due to these tools.

The point is that the numbers alone sometimes point you in the wrong direction.

Anytime we evaluate a player, we should accept their statistical profile as valuable information but then dive deeper into the possible ways it might not be telling the entire story. That’d be much better than drawing rash, immediate conclusions when a player’s analytical profile card goes viral on Twitter.

(Photo of Drew Doughty: Harry How / Getty Images)

What did you think of this story?

Rate as MEH
MEH

Rate as SOLID
SOLID

Rate as AWESOME
AWESOME


Harman Dayal is a staff writer for The Athletic covering the Vancouver Canucks. He combines NHL video and data analysis and tracks microstats as part of his coverage. Follow Harman on Twitter @harmandayal2

OPEN IN APP
National
Boxing
Bundesliga
Champions League
Championship
College Football
Copa del Rey
Culture
Europa League
European Championship
FA Cup
Fantasy Baseball
Fantasy Basketball
Fantasy Football
Fantasy Hockey
Fantasy Premier League
Formula 1
Gaming
Golf
International Soccer
La Liga
League Cup
League One
League Two
LNH
Men's College Basketball
Men's World Cup
Mixed Martial Arts
MLB
MLS
Motorsports
NBA
NFL
NHL
NWSL
Olympics
Opinion
Premier League
Scottish Premiership
Serie A
Soccer
Sports Betting
Sports Business
Tennis
UK Women's Football
WNBA
Women's College Basketball
Women's Euros
Women's Hockey
Women's World Cup
NCAA Women's Basketball
The Athletic Ink
Podcasts
Headlines
Real Time
US
Arizona
Atlanta
Baltimore
Bay Area
Boston
Buffalo
Carolina
Chicago
Cincinnati
Cleveland
Columbus
Dallas
Denver
Detroit
Houston
Indiana
Jacksonville
Kansas City
Las Vegas
Los Angeles
Memphis
Miami
Minnesota
Nashville
New Orleans
New York
Oklahoma
Oregon
Orlando
Philadelphia
Pittsburgh
Sacramento
San Antonio
San Diego
Seattle
St. Louis
Tampa Bay
Utah
Washington DC
Wisconsin
Canada
Calgary
Edmonton
Montreal
Montréal (français)
Ottawa
Toronto
Vancouver
Winnipeg
Partners
Tickets by StubHub
Share
Buy a Gift
HQ
About Us
Careers
Code of Conduct
Editorial Guidelines
Business Inquiries
Press Inquiries
Support
FAQ
Forgot Password?
Redeem Gift
Contact Us
Terms of Service
Group Subscriptions
Newsletters
The Pulse
The Bounce
The Windup
Prime Tire
Full Time
©2023 The Athletic Media Company. All rights reserved.
Privacy Policy
Support
Sitemap
Twitter
Facebook
Instagram

Download on the App Store

Get it on Google Play
- MeltingPlastic

Thanks. To me, the 10 players were not the key as to why I posted the column. More so the analytics versus eye test debate. The players are somewhat ancillary to the discussion but support the view

Hi Jan ----

Your blog is referencing an Athletic article that we cannot see without a subscription. I have no idea what 10 players are being discussed...
- TommyGTrain

Fun fact if you go to the article via twitter on an iPhone you can see it for free if you click “show reader” in the top right

Fun fact if you go to the article via twitter on an iPhone you can see it for free if you click “show reader” in the top right
- picklerick

Thanks, didn’t know that

The Athletic
team
league
NHL
NHL

Teams
Scores & Schedule
Standings
• • •
Ten worst contracts
Offseason grades
Ten NHL players analytics were wrong about and the lessons learned

LOS ANGELES, CALIFORNIA - MARCH 04: Drew Doughty #8 of the Los Angeles Kings turns with the puck during a 4-2 Kings win over the St. Louis Blues over the St. Lous Blues at Crypto.com Arena on March 04, 2023 in Los Angeles, California. (Photo by Harry How/Getty Images)
By Harman Dayal
Jul 27, 2023
190

Save Article
NHL analytics have exploded in popularity over the last decade.

Teams are investing in staff and building departments for the sole purpose of analyzing data and in the public sphere, it’s significantly enhanced the information and knowledge that fans and media have about players and teams.

The majority of the hockey world now agrees that data has some value as a tool that can offer objective insights into player performance. The old days of intense analytics versus eye test debates are mostly gone because people understand that it’s not an either-or proposition, you should be using both the eye test and data.

ADVERTISEMENT


Hockey fans have a more sophisticated understanding of players around the NHL — especially players from teams they don’t watch — because of how accessible public analytics are.

That, of course, should be celebrated.

But if we’re going to rely on these tools, we should also be aware of their limitations and the cases where stats can be misleading.

I’ve been diving into analytical tools like shot shares, expected goals and PDO; types of microstats like zone exits, zone entries and passing data that Corey Sznajder tracks; powerful tools like Dom Luszczyszyn’s excellent Net Rating model and more for years now. Today, I wanted to highlight examples where analytics were misleading about a player’s true value.

The purpose of this exercise isn’t to blame analytics. These are often situations where I was personally wrong in putting too much stock in the data and not considering other factors.

Experience over time has made me handle analytics with more scrutiny.

Instead of drawing conclusions at face value because a model feels strongly about a player, I try and find the holes, blind spots and possible ways I could be wrong. When a player I need to offer deep analysis on is acquired, I’ll watch hours of game film and dissect those observations to add a new perspective. I weigh a lot more qualitative factors now. And honestly, it all helps paint a clearer, more nuanced understanding of players.

Here are 10 players that analytics were misleading about, and some of the lessons I’ve learned as a result.

Chris Tanev
Chris Tanev has been one of the NHL’s best shutdown defensemen for a long time. He’s excellent positionally, has special defensive awareness and instincts, a disruptive stick, a long reach and a fearless attitude.

Tanev was an analytics darling for most of his Canucks tenure. At the end of the 2019-20 season, Vancouver had to decide whether to re-sign him to a long-term deal or let him walk as an unrestricted free agent. Tanev was on the cusp of turning 31, had durability issues (before 2019-20, his games played totals were 55, 42 and 53 going back to the 2016-17 season) and while he was still an effective top-four defenceman, his two-way metrics were on the decline.

Tanev’s isolated impact on helping the Canucks control five-on-five shots, for example, had diminished considerably from his peak. He’d gone from an elite driver in his prime to looking below average.



Evolving-Hockey’s RAPM tool

Colleague Dom Luszczyszyn’s Game Score Value Added (GSVA) model projected that Tanev was only two or three years away from being a replacement-level contributor.



Tanev was a great fit with Hughes, but committing four or five years of term to an oft-injured 30-year-old defender with declining numbers seemed very risky. I thought the Canucks were wise to let him walk to Calgary. And, of course, I was totally wrong.

Tanev immediately resurrected his game, posting dominant two-way numbers and re-establishing himself as one of the best defensive defencemen in the NHL. The Canucks, meanwhile, have desperately missed Tanev’s defensive reliability at even strength and the penalty kill.

There are two main lessons I gained from this.

No. 1: The environment a defender plays in can influence their numbers a lot. Tanev went from a club that generally struggled to control play and had far less two-way help from forwards in Vancouver to an excellent possession team in Calgary. Even the stats that supposedly isolate or adjust for those teammate- and environment-related factors don’t do a sufficient job of accounting for those differences.

No. 2: Defencemen — especially stay-at-home types — are harder to properly evaluate using analytics than forwards because it’s much harder to statistically measure defensive impact.

Neal Pionk and Jacob Trouba
In 2019, the Jets traded Jacob Trouba to the Rangers for Neal Pionk and a first-round pick. Trouba was coming off a 50-point season, played shutdown minutes alongside Josh Morrissey and profiled like a bona fide top-pair defender. Pionk, on the other hand, was a young player still trying to establish himself.

I thought it was a sweet deal for the Rangers and a disappointing haul for Winnipeg. The opposite turned out to be true.

Trouba hasn’t lived up to that potential and is now overpaid, owning an $8 million AAV that ties him for the 14th-highest cap hit among all NHL defensemen. Pionk, meanwhile, immediately blossomed in Winnipeg.

This is another reminder of how significantly a change of scenery can influence a defender’s value.

As for Pionk, he had promising offensive tools but was flawed defensively and his underlying numbers were ugly. In hindsight, there was a plausible explanation for his poor analytical profile.

Pionk was coming off his first full NHL season and yet he was thrust into a top-pairing role alongside Marc Staal, who was rapidly declining for a large chunk of it. Almost every rookie offensive defenseman would struggle if he had to match up against the opposition’s best players, with a subpar partner, on a bad, rebuilding team. Pionk had also gotten hurt in the middle of the season and lost his confidence.

With all of these factors at play, Pionk’s first NHL season in New York shouldn’t have been graded nearly as harshly. When a defender has bad numbers, we should dive deeper into who their partner was, what kind of club they were on, the difficulty of the role they were playing and their experience.

In Trouba’s case, it’s not that the analytics were wrong, but more that the monster final year he had in Winnipeg was an outlier.

Trouba ran the Jets’ first-unit power play in 2018-19 and ranked 13th among all NHL defensemen in points, ahead of players like Erik Karlsson, Drew Doughty and Zach Werenski. That created an expectation that he could be a high-end offensive piece for the Rangers, despite the fact 33 points was his career high before that final year with the Jets.

Tage Thompson
An abnormally high shooting percentage during a single season is one of the most common signs that a player’s goal-scoring rate could slow down moving forward. Over multi-season stretches, even many of the top snipers struggle to convert on 16-17 percent or more of their shots. That means when a player like Chris Kreider, for example, scores 52 goals in 2021-22 on the back of a 20.2 percent shooting clip, we usually hypothesize that he’ll slow down the following year, as he did last season with 36 goals.

There are countless examples of teams that got burned by making bets on players with inflated shooting percentages.

When Tage Thompson broke out for 38 goals in 2021-22, it seemingly came out of nowhere. I still believed in Thompson as a player, and didn’t mind the big extension the Sabres signed him to, but heading into last season, I figured he’d be closer to a consistent 30-goal scorer rather than flirting with 40.

Why? Well, Thompson had nearly tripled his previous five-on-five shooting clip in 2021-22 and hadn’t flashed elite goal-scoring potential at lower levels, especially at even strength. As The Athletic’s Jonathan Willis had pointed out, Thompson had a pedestrian 10.6 career shooting percentage in the AHL against minor-league goalies and in his draft year, he scored just one even-strength goal in 36 games for the University of Connecticut.

Thompson made that regression talk look silly by exploding for 47 goals last season. Of course, there are factors such as Thompson’s positional shift to center, improved linemate quality, confidence, skill and rare 6-foot-6 frame that probably weren’t weighed heavily enough.

But there’s also the point that Thompson’s 15 percent shooting clip from 2021-22 was high but not crazy. This wasn’t a case of a player scoring at a ridiculous 25 percent clip; 15 percent is actually somewhat reasonable for an elite sniper to maintain year in, year out. We just weren’t sure he could maintain it because Thompson’s previous track record didn’t hint at star potential.

The lesson isn’t to ignore shooting percentage as a signal. That would be very foolish and lead to tons of regrettable mistakes — Thompson is a very special case and the overwhelming majority of players won’t follow his career arc. Rather, the lesson is that we shouldn’t go too far on the other side of the pendulum either in blindly assuming a player with a spiked shooting percentage is bound to regress. Occasionally, there’s more than meets the eye.

Vladislav Gavrikov
Vladislav Gavrikov caught my eye two or three years ago.

It was cool seeing this unheralded, hulking defender use his long reach and solid skating to disrupt plays, win battles down low and excel as a defensive-minded presence. But by the time the trade deadline rolled around and Gavrikov had emerged as a trade candidate, it seemed like somebody was at risk of overpaying based on all the hype around him.

Many stats-based analysts were dumping on Gavrikov, pointing to his putrid underlying numbers in Columbus. From experience, I wasn’t too worried about that because he was being thrown to the wolves as a No. 1 defenseman on a bad team as a result of Zach Werenski’s injury, defending top players, playing with a subpar partner and starting a ton of his shifts in the defensive zone.

It was easy to see him bouncing back once slotted into a more appropriate role on a better roster. I’d been wrong enough times and seen Gavrikov play enough to be less scathing, but a fair chunk of the public discourse characterized him as the next Ben Chiarot because of his analytical profile.

Gavrikov turned out to be the perfect fit for the Kings. He was a stud on the second pair, posted dominant play-driving results and drove a 14-6 five-on-five goal differential in L.A.’s favor. Once again, context and team situation play a huge role in determining an individual player’s numbers.

I should caution that not every high-profile defenseman ends up following this arc. I don’t want people to read about all these positive examples and assume every high-profile defender with bad numbers on a terrible club will end up with an enormously positive ending.

Oliver Ekman-Larsson is an example of a player with poor advanced stats on a terrible Coyotes roster — with tons of legitimate reasons to believe in a bounce back with a fresh start — who turned out to be a disastrous acquisition for the Canucks. Jeff Petry is another case — he never rebounded to his peak form when he moved from Montreal to Pittsburgh last season.

So yes, stats aren’t close to the be-all, end-all for evaluating defenders on bad teams, but they shouldn’t be totally ignored either.

Jason Richardinson
The Vancouver Canucks were searching for a defensive-minded third-line center heading into the 2021-22 season. The Seattle expansion draft was also looming that offseason and some contenders were looking to trade away valuable players they couldn’t protect in the draft and didn’t want to lose for nothing. Dallas made Jason Richardinson available because of that, and the Canucks swooped in to acquire him for a third-round pick.

Richardinson was limited offensively but he had a stellar defensive profile in his final season with the Stars. The rangy checking center helped Dallas seize a 59 percent control of five-on-five scoring chances. He was a top-10 forward in the NHL that season at suppressing expected goals against and the Stars surrendered just 1.96 goals against per hour with him on the ice. He was also one of the Stars’ most-used forwards on the penalty kill.

It wasn’t just the analytics that pointed to Richardinson’s value either — the coaching staff had immense trust. He ranked fourth among all Stars forwards in five-on-five ice time per game, averaging more five-on-five minutes than the likes of Jamie Benn and Roope Hintz.

Richardinson turned out to be a disastrous fit in Vancouver. His sturdy defensive game was nowhere to be found and his offensive performance was completely anemic. It went so poorly that last fall, the Canucks paid a second-round pick just to dump his contract to Chicago and pick up Riley Stillman.

What I (and the Canucks) didn’t account for was that Richardinson’s game was a unique fit in Dallas. The Stars played an insufferably boring, low-event style that promoted defense above all under Rick Bowness at the time. That was the perfect fit for Richardinson, who offers no offense but has some redeemable checking qualities. The Canucks, like most NHL organizations, played a very different structure and tempo than the Stars.

The mistake was assuming he’d be the same player in a totally different system, on a significantly weaker defensive team.


Jason Richardinson. (Sergei Belski / USA Today)
Colin Miller
Colin Miller’s name hit the rumor mill in the summer of 2019 when he was a member of the Vegas Golden Knights. Vegas needed to clear money and Miller, with three years remaining on a $3.875 million cap hit at the time, was the odd man out.

At the time, it looked like a golden opportunity to affordably acquire a top-four right-shot defender. Miller averaged just shy of 20 minutes per game the last two seasons, he was a year removed from scoring 41 points and he owned strong play-driving numbers.

Why has it never worked out for him as a top-four defenseman?

The missing context is that his matchups were heavily sheltered. Miller wasn’t trusted defensively, so coaches mostly deployed him against the opposition’s third and fourth lines.

When you acquire a bottom pairing who’s delivered excellent results in a sheltered capacity, you’re rolling the dice if you expect them to succeed further up the lineup. Some like Vince Dunn, Nate Schmidt and Mackenzie Weegar turn out to be the real deal. But others like Miller can’t take the next step.

Drew Doughty
Drew Doughty is the second-highest-paid defenseman in the NHL. It’s pretty widely accepted he isn’t quite worth the full freight of his $11 million cap hit. But two or three years ago, we reached a point where many in the analytics community weren’t just arguing Doughty was overpaid. Many were arguing he was washed up.

I remember coming across Jack Han’s fantastic article in the middle of the Doughty wars. Han broke down some game film and had a fascinating theory that changed my perspective at the time. He showed many revealing video clips and concluded that Doughty still has “elite tools and ability to take over a game, but seems to lack the motivation or inspiration to do hard things every shift in order to help his team control play.”

The takeaway was that Doughty still had the skill set to be a strong top-pair player but perhaps lacked intensity because he was on a losing, rebuilding team. Sure enough, the Kings have made the playoffs the last two years, and in that meaningful environment, Doughty is playing excellent hockey again. He wasn’t washed up after all.

Jesse Puljujarvi
Jesse Puljujarvi was one of the Oilers’ most polarizing players in recent memory.

The big Finnish winger was hot out of the gate in 2021-22 with 10 goals and 23 points in his first 28 games, playing alongside Connor McDavid. But then he missed action due to COVID-19, lost his rhythm, suffered a lower-body injury in February and found himself in the bottom six, notching just four goals in his final 37 games.

Puljujarvi was all over the rumor mill last summer and that’s around the time the debates in Edmonton reached a fever pitch.

Many went to bat for Puljujarvi, pointing out that his play-driving numbers were really strong, with his puck retrievals and forechecking helping any line he was on control possession and generate a ton of scoring chances. They pointed out he scored at a 45-point-per-82-game pace in 2021-22 and made the case he’s a solid middle-six player. Others fought back and claimed his scoring chance generation doesn’t really matter if he’s a terrible finisher and that he doesn’t have the IQ to make plays with the puck.

Puljujarvi was re-signed to a one-year, $3 million last summer but it never worked out. He continued whiffing on chances last season, completely lost his confidence and his lack of penalty-killing utility meant he didn’t fit the prototypical profile of a defensive bottom-six winger.

Travis Dermott
When Travis Dermott first arrived on the scene in Toronto, there was a lot of hope for him as a future top-four defender.

He was mobile, athletic, a slick puck-mover and immediately crushed his bottom-pair assignments with strong two-way metrics at just 21 years of age. Some even argued Dermott should be fed more minutes in the 2018 playoffs and that Ron Hainsey, who was overmatched on the top pair, should have his workload reduced.

When you make such an impressive first impression in the NHL at 21, it’s only natural fans expect there’s another level to reach.

Dermott, unfortunately, couldn’t hit that potential. Like Colin Miller, this is the case of a solid third-pair defender with excellent analytics who simply couldn’t translate his game further up the lineup in non-sheltered minutes. He was too mistake-prone and couldn’t be trusted defensively.

I’ve learned to put even more stock into the difficulty of a player’s matchups. The head coach is often right when he decides that the promising young No. 6 defender with amazing analytics in a sheltered role doesn’t deserve a larger role.

Conclusion
Analytics are useful for telling you what results are happening when a player is on the ice (e.g. the team dominates opponents in terms of possession and scoring chances when Player X is on the ice, or the club gets caved in and bleeds a lot defensively when Player Y is on the ice). It provides objective data and helps account for biases. The eye test — and other contextual and qualitative factors — tell you why those results happened and what it actually reveals about the player’s ability, skill set and value.

Analytical models have often been right when they’ve suggested this player is overrated or that player is underrated. I truly hope people don’t take this article as an opportunity to bash analytics or the work of a smart, talented colleague like Luszczyszyn because the hockey public’s overall knowledge and conversations around players are sharper and more sophisticated due to these tools.

The point is that the numbers alone sometimes point you in the wrong direction.

Anytime we evaluate a player, we should accept their statistical profile as valuable information but then dive deeper into the possible ways it might not be telling the entire story. That’d be much better than drawing rash, immediate conclusions when a player’s analytical profile card goes viral on Twitter.

(Photo of Drew Doughty: Harry How / Getty Images)

What did you think of this story?

Rate as MEH
MEH

Rate as SOLID
SOLID

Rate as AWESOME
AWESOME


Harman Dayal is a staff writer for The Athletic covering the Vancouver Canucks. He combines NHL video and data analysis and tracks microstats as part of his coverage. Follow Harman on Twitter @harmandayal2

OPEN IN APP
National
Boxing
Bundesliga
Champions League
Championship
College Football
Copa del Rey
Culture
Europa League
European Championship
FA Cup
Fantasy Baseball
Fantasy Basketball
Fantasy Football
Fantasy Hockey
Fantasy Premier League
Formula 1
Gaming
Golf
International Soccer
La Liga
League Cup
League One
League Two
LNH
Men's College Basketball
Men's World Cup
Mixed Martial Arts
MLB
MLS
Motorsports
NBA
NFL
NHL
NWSL
Olympics
Opinion
Premier League
Scottish Premiership
Serie A
Soccer
Sports Betting
Sports Business
Tennis
UK Women's Football
WNBA
Women's College Basketball
Women's Euros
Women's Hockey
Women's World Cup
NCAA Women's Basketball
The Athletic Ink
Podcasts
Headlines
Real Time
US
Arizona
Atlanta
Baltimore
Bay Area
Boston
Buffalo
Carolina
Chicago
Cincinnati
Cleveland
Columbus
Dallas
Denver
Detroit
Houston
Indiana
Jacksonville
Kansas City
Las Vegas
Los Angeles
Memphis
Miami
Minnesota
Nashville
New Orleans
New York
Oklahoma
Oregon
Orlando
Philadelphia
Pittsburgh
Sacramento
San Antonio
San Diego
Seattle
St. Louis
Tampa Bay
Utah
Washington DC
Wisconsin
Canada
Calgary
Edmonton
Montreal
Montréal (français)
Ottawa
Toronto
Vancouver
Winnipeg
Partners
Tickets by StubHub
Share
Buy a Gift
HQ
About Us
Careers
Code of Conduct
Editorial Guidelines
Business Inquiries
Press Inquiries
Support
FAQ
Forgot Password?
Redeem Gift
Contact Us
Terms of Service
Group Subscriptions
Newsletters
The Pulse
The Bounce
The Windup
Prime Tire
Full Time
©2023 The Athletic Media Company. All rights reserved.
Privacy Policy
Support
Sitemap
Twitter
Facebook
Instagram

Download on the App Store

Get it on Google Play
- MeltingPlastic

RangerSaver is somewhere seeing this post thinking he needs to step his game up.

The real message is that which is called analytics is just number crunching, and analytics implies analysis of the crunched numbers.

RangerSaver is somewhere seeing this post thinking he needs to step his game up.
- aecliptic

Saver uses a lot of words to say noting. This article was informative.

Melting Plastic ---- Thanks for posting!

Fun fact if you go to the article via twitter on an iPhone you can see it for free if you click “show reader” in the top right
- picklerick

And if you're on a computer, you can usually open an article and hold the ESC key for 10+ seconds and the paywall doesn't load. Those two things are the reason I didn't renew my subscription

Saver uses a lot of words to say noting. This article was informative.
- blueshirts_fan

What's most intriguingly curious about the internet user and patron of this Hockeybuzz website in the year of our lord 2023 who goes by the assumed name of "RangerSaver" is if in fact his posts are serious in nature or rather they are merely a rouse of the comedic variety intended to invoke vitriolic emotions such as rage and disdain among the other consumers of Mr. Jan Levine's internet content posted on Hockeybuzz.com on a near-daily basis. One can only posit which route the Hockeybuzz.com user whose pen-name is "RangerSaver" chooses to take when he makes his excessively, exceedingly, and extravagantly-long internet message board posts via entries through his keyboard device, but where there lies no question is in that he undoubtedly suffers from a severe case of the malady commonly referred to and known as "carpal tunnel syndrome".

And if you're on a computer, you can usually open an article and hold the ESC key for 10+ seconds and the paywall doesn't load. Those two things are the reason I didn't renew my subscription
- eichiefs9

Does this work for other paywalls?

Does this work for other paywalls?
- Fenrir

Just use free porn sites like an adult

And if you're on a computer, you can usually open an article and hold the ESC key for 10+ seconds and the paywall doesn't load. Those two things are the reason I didn't renew my subscription
- eichiefs9

How did they mess their main source of revenue up so badly

Does this work for other paywalls?
- Fenrir

Some, yeah. Not all though. I find the "bigger" the company/organization/etc... the less that trick works. I've tried it on sites like NYT, Washington Post, etc... and it doesn't do anything.

But you have to do it like the second the page fully loads. It might take a few tries to get the timing down but it works for me.

How did they mess their main source of revenue up so badly
- picklerick

I don't know but I had pretty much stopped reading it all that much by the time I was up for renewal. Their "new" Islanders beat writer is kind of a dud too, so that didn't really help either. And the fact that I don't care at all about 50% of the big four sports and don't really read a ton about baseball either.

If they allowed you to subscribe by sport (or team) I'd probably do it for hockey alone if they offered that, since you don't get to utilize all the features by using those two tricks, but otherwise I'm fine weaseling my way around the paywall until they inevitably fix those issues.

What's most intriguingly curious about the internet user and patron of this Hockeybuzz website in the year of our lord 2023 who goes by the assumed name of "RangerSaver" is if in fact his posts are serious in nature or rather they are merely a rouse of the comedic variety intended to invoke vitriolic emotions such as rage and disdain among the other consumers of Mr. Jan Levine's internet content posted on Hockeybuzz.com on a near-daily basis. One can only posit which route the Hockeybuzz.com user whose pen-name is "RangerSaver" chooses to take when he makes his excessively, exceedingly, and extravagantly-long internet message board posts via entries through his keyboard device, but where there lies no question is in that he undoubtedly suffers from a severe case of the malady commonly referred to and known as "carpal tunnel syndrome".
- eichiefs9

right on brand that this is only 2 sentences

In 3 seasons McDavid is a UFA…the same offseason Trouba and Panarin are off the books…so is that when we give McDavid $20 mill a year??? I figure he’ll be dying to get out of Edmonton by then

Today @ 11:32 AM ET
In 3 seasons McDavid is a UFA…the same offseason Trouba and Panarin are off the books…so is that when we give McDavid $20 mill a year??? [] I figure he’ll be dying to get out of Edmonton by then

In 3 seasons we will again be starting a rebuild.

In 3 seasons we will again be starting a rebuild.
- Brukie

McDavid wouldnt be a bad brick

In 3 seasons McDavid is a UFA…the same offseason Trouba and Panarin are off the books…so is that when we give McDavid $20 mill a year??? I figure he’ll be dying to get out of Edmonton by then
- Slimtj100

He'll be going to toronto

What's most intriguingly curious about the internet user and patron of this Hockeybuzz website in the year of our lord 2023 who goes by the assumed name of "RangerSaver" is if in fact his posts are serious in nature or rather they are merely a rouse of the comedic variety intended to invoke vitriolic emotions such as rage and disdain among the other consumers of Mr. Jan Levine's internet content posted on Hockeybuzz.com on a near-daily basis. One can only posit which route the Hockeybuzz.com user whose pen-name is "RangerSaver" chooses to take when he makes his excessively, exceedingly, and extravagantly-long internet message board posts via entries through his keyboard device, but where there lies no question is in that he undoubtedly suffers from a severe case of the malady commonly referred to and known as "carpal tunnel syndrome".
- eichiefs9

Honestly, I really do wish he wrote in a much more concise manner because he writes very relevant stuff. I just wish he took less of a Pierre McGuire route. I dont need to know peoples favorite colors and meals, what HS they went to and what their favorite designer is. Its like Im reading someones dissertation every time.

Honestly, I really do wish he wrote in a much more concise manner because he writes very relevant stuff. I just wish he took less of a Pierre McGuire route. I dont need to know peoples favorite colors and meals, what HS they went to and what their favorite designer is. Its like Im reading someones dissertation every time.
- aecliptic

There's nothing wrong with the substance of what he writes, just the fact that he uses 30 words when 5 would suffice. It reminds me of being assigned essays in high school that needed to be X amount of words and trying to fluff up the paragraphs with extra words

There's nothing wrong with the substance of what he writes, just the fact that he uses 30 words when 5 would suffice. It reminds me of being assigned essays in high school that needed to be X amount of words and trying to fluff up the paragraphs with extra words
- eichiefs9

There should be a "translate" button on each one of his posts