Tom Fitzgerald wants it all in front of him at this time of year. There are reports from the New Jersey Devils’ pro scouts, opinions from his front-office and coaching staffs and reams of data from their deep analytics team.
Not too long ago, an NHL general manager heading into the free-agent frenzy of July 1 would have plenty to work with — but based only on the first two parts of the above formula. Now that the age of data is in full swing, the process of targeting free agents and negotiating and signing contracts is evolving. All 32 NHL teams — plus plenty of agents — have analytics teams producing detailed data on everything from shot locations to scoring chances to zone entries and beyond.
Advertisement
“You’re definitely over-prepared,” Fitzgerald said. “You want to be over-prepared. For me, I can’t get enough of (the data). You’re not making decisions based solely off of it. There has to be a combination of the data, the scouts and my own eyes. You go over and over your lists (ahead of July 1) and you use everything you can get your hands on to try and figure out how to best improve your team.”
Every GM, no matter how long they’ve been on the job, uses data. And the inexact science of finding a free-agent fit needs all the help it can get. The stakes and scrutiny this time of year are high — it won’t take more than a few minutes after a signing for media and fans to criticize deals as too expensive or too long.
In talking to a handful of general managers, team executives and agents, some of whom were granted anonymity in order to speak candidly on an intensely competitive and often secretive process, it becomes clear that the preparations for July 1 and beyond have to include every available detail.
“Our group has done a really good job of putting together every ounce of information that happens in a game,” Fitzgerald said of the Devils’ data team, which is headed by former The Athletic contributor Tyler Dellow. “’How’s this guy off the rush? How is he defending the rush? How is he when he plays with these two guys? What about with a different two guys?’ We have everything you could need or think of — it’s a lot. Sometimes, it’s too much.”
Still, data is only one part of the equation. Scouting reports and in-game data are filtered by the analytics department and are mixed with various eye tests, allowing the group to zero in on a list of potential fits. As July 1 nears, a mostly complete picture has formed.
“The data is largely secondary at this point in the season,” said Eric Tulsky, the longtime Hurricanes front-office analyst who was named GM last week. “Our starting point is the scouting opinions. The tweaks are the coaches’ opinions.”
Advertisement
“(Data is) not going to make the determination,” said Islanders GM Lou Lamoriello, who has taken a one-person analytics department and turned it into a six-person team. “But you see flags here and there. You’re trying to refine your opinions. And if you’re not looking at it, you’re making a mistake.”
That’s just the first part of the equation, though. An executive can analyze every aspect of a free agent’s fit with their team, only to run up against a player seeking more term and dollars than it has a budget for. Data comes into play in negotiations, too.
“Negotiating contracts is still half science, half art. There’s an art to it,” one longtime NHL agent said. “You can’t just pick numbers out of mid-air, but you also can’t just be a robot. So a team can come into free agency and their data and their charts say our player is worth x times x, we have a different idea and if you say, ‘No, our data is what we’re going with,’ you’re going to miss out. It’s still a give-and-take.”
The frenzy’s evolution
You don’t have to go back very far in NHL history to see the how the timeline of day-one, minute-one free agency began.
The New York Rangers are seen as the first organization to start wooing free agents the minute the market opened. In the summer of 2001, according to a person involved in the planning, they had assistant GM Don Maloney waiting outside Darius Kasparaitis’ front door at midnight to deliver a “Come to the Rangers” gift basket, replete with a Tiffany apple for Kasparaitis’ wife, Rangers jerseys for his whole family and a DVD of various New York celebrities wooing potential free agents to play at Madison Square Garden.
It’s been a short ride in terms of years, but a long way since Chris Rock, the cast of “The Sopranos” and Tim Robbins shot quick hits expounding on all that New York City and the Rangers had to offer, until now, when how a player fits into a team’s system (with plenty of video, scouting reports and data) is the top concern before anyone gets to the money.
Advertisement
The through line from the era of putting on a show to putting out a PowerPoint on how a player fits is the work of pro scouts. Teams may have as few as three scouts covering the league — general managers and assistant GMs travel plenty, too, and make their own notes — but the process of identifying potential targets for July 1 begins the previous September.
“I feel like the games our guys see in September and October are just as important for us as the ones in March and April,” said a Western Conference executive. “Our guys do reports on everyone they see in a game, but particularly focus on the upcoming free agents and restricted free agents — there’s profiles on every aspect of a player’s game, and you start in preseason to get that process going on whether you think they’d be a fit for us.”
Tulsky’s Hurricanes stand out for their style of play under Rod Brind’Amour and how detailed Carolina is about finding players, either through trades or in free agency, that fit the aggressive, full-ice, man-on-man style they’ve employed to relatively good success over the last six years. “They know what they want and they want to get it cheap,” the agent said. “They’re always on the lookout for market inefficiencies.”
And even though Carolina’s data team has sway with owner Tom Dundon, Tulsky said the Canes’ process hasn’t changed much since he’s been there. “There’s elements of the way the scouting process has changed, elements of the data have changed, but I don’t think we’ve changed that much, honestly,” he said. “The scouts generally give us the names of players we should focus on, more often than the other way around.”
The Western Conference executive said his team’s postseason meetings are a collaborative effort between the scouting staff, the data team and the front office. He’s heard plenty of stories about opposite sides within an organization, with the “eye test vs. data” battle getting contentious, but hasn’t seen it with his team. “The scouting reports are side by side with the analytics group,” he said. “We rank them, and at the end-of-season meetings, we go through the details to have a list ready to go.
“Both staffs, scouting and analytics, are great for us. Nobody’s pushing. Nobody’s saying, this is my guy. You hear about situations where maybe someone higher up doesn’t have much faith in the data, or someone from the data side speaking out too strongly, but our groups are great. It’s always a dialogue. The goal is to make the team better, and we’re all part of that.”
Nearly all of the GMs we spoke to said that internal data rarely drives the scouting process. “I don’t want our pro scouts to be distracted,” Lamoriello said. “There’s still a feel to it, or you wouldn’t need them. The worst thing you can do, in my opinion, is discourage a pro scout from a player they like. You never want to have a statistic or a number make them doubt their assessment. It’s the same with our amateur scouts and the draft — look at all the projections everyone has for the draft. Don’t get distracted by the wrong things.”
Advertisement
One Eastern Conference executive said his team does guide some of its pro scouting process based on data but doesn’t want to overdo it. “You can give them too much intel, for sure,” he said. “Just be in the building and watch.”
‘We don’t get much of the jargon’
From the agents’ side, the buildup to July 1 hasn’t changed a lot. The Eastern Conference executive said he’s received emails and texts from agents over the years extolling the advanced metrics of certain free agents, but that’s just part of the business of agents selling their clients to teams any way they can. “Sportlogiq, Natural Stat Trick — if there’s a way they can plead their case better, we will hear about it,” he said with a laugh.
But the agents we spoke to tend to see the data-driven arguments more from the team side. “We don’t get a lot of the jargon,” said Matt Keator, who represents Florida Panthers defenseman Brandon Montour, potentially one of the biggest-ticket free agents this year. “You’re not really hearing a lot of, ‘Here’s his zone entry and exit numbers, here’s this data point.’ It’s usually more, ‘We like how he breaks the puck out, we like what he does on the power play,’ things like that. We don’t get much of the (data) jargon.”
“Anyone who tells you, from the agent side, that analytics has changed their approach is full of s—,” the agent who requested anonymity said. “Most teams start from the same place. This year, it’s probably: We want to add a scoring winger. There’s (Steven) Stamkos, there’s (Jonathan) Marchessault, there’s (Tyler) Toffoli, there’s (Jake) Guentzel. Some teams prefer Guentzel; others might prefer Toffoli. Are they making their judgments based on metrics? Of course not. They want a guy they think can score 40 goals. That’s where it starts and ends this time of year, at the top of the lists.”
Mistakes will be made
The availability of data, and a willingness to incorporate it into a team’s free-agency search, should mean fewer missteps. Right? Well, perhaps not.
Look at some of the goalie swings and misses in recent years. The Oilers took a huge swing on Jack Campbell in the summer of 2022, signing him to a five-year deal worth $5 million a year. Campbell spent most of this past season in the AHL. Just a cursory glance at Clear Sight Hockey, Steve Valiquette’s data site that counts 10 NHL teams as clients and has very accurate goaltending metrics, showed that Campbell was barely above league average in goals saved above expectation in his breakout 2021-22 season with the Leafs. A long-term deal for a goalie is a risky bet, even if it’s an extension; signing a goalie away from another organization, where systems are different and how shots funnel to the net are different, is a massive risk at a large number and term.
You can defy the data and do well, though. The Rangers went for Jonathan Quick last summer, coming off an abysmal season for the surefire Hall-of-Famer — he had ranked last, 106th out of 106, in GSAx with the Kings and Golden Knights. The Rangers were also looking for an inexpensive backup to work behind Igor Shesterkin, not a No. 1 like the Oilers tried to get with Campbell. There’s also what one executive calls “human intel,” the constant internal and external discussions between teams on how a player fits into a room and what his personality is like.
Quick has excelled as a Ranger, not only on the ice but off it. He was worth the risk.
Plenty of other mistakes have been, and will continue to be, made in the age of data. As the agent outlined above, teams will be jumping over one another to sign those four top-six scoring wingers, possibly giving out an extra $1 million to $2 million per year and one or two years more term to win the chase. It takes discipline to stick to your list and your projections, or risk having to explain to your owner why you had to unload a young, promising player because you gave a 30-something five years.
Advertisement
“That’s where you can get into trouble,” Fitzgerald said. “You’ve got your lists, your scouts and your analytics teams have done the work. You try not to blow it up in the middle of the chase.”
Data helps more at the secondary free-agent level
Once you get beyond the first-hour signings on July 1, data can help teams identify depth targets and separate players at the lower end of the pay scale. Carolina has done this well — sometimes too well, as effective bottom-sixers such as Jordan Martinook (acquired in a midseason trade with the Coyotes in 2018-19) and Stefan Noesen (2021 free-agent signing) likely are hitting the market this summer after successful runs in the Hurricanes’ system.
The Devils, who have been armed with loads of cap space in recent years after making the playoffs just twice since 2012, have taken some swings at building up their depth to complement their young, highly skilled core. They added Dougie Hamilton and Ondrej Palat in higher-number deals, and also some lesser signings that haven’t worked out.
“We’ve had opportunities to hit balls in the gaps,” Fitzgerald said, “and we’ve ended up grounding out. We didn’t strike out, but we didn’t get the low-risk, high-reward guy we hoped we were getting.”
Teams in a tight cap squeeze have to get creative to add, making the data a bit more important when they have only limited dollars to spend. “Our analytics people will bring up guys they like — maybe it’s a player who looks good on paper that’s been spending a lot of time in the AHL or as a scratch,” the Western Conference executive says. “And there’s usually warts there — our scouts will know why that player’s in the minors. So (the data) can certainly get a conversation started and, if the price is right, maybe you bring that player in and he surprises you.”
The holy grail of free-agent data
The place that every GM wants to get to with data and free agency is finding a way to calculate fit based on a player who plays a different system, with different teammates, different quality of competition. “Fit” is an all-encompassing term for teams, players and agents, and the work to refine it continues, though not with much precision as of yet.
“We are working toward that. I don’t think we’re terribly close just based on data at this point,” Tulsky said. “It can act as a preliminary screen, it can rule in and out at the extremes. But most players lie in the middle where you need eyeballs on them.”
Advertisement
“It’s hard to compare a player from Chicago to a player from Dallas. It just is,” Fitzgerald said. “We know the systems, we know the differences and now it’s, how do you see the player working in our system? We have to make sure our data team is gathering the right information, that our scouts are looking at players playing different systems than ours and not downgrading a player because he’s chasing to the point (in a dice-style defensive structure) because that’s not how we play.
“You would love to have numbers that clear those things up. We like what we have now, though.”
Valiquette’s Clear Sight does workups ahead of free agency for his team clients all the time, but he cautions them — and his site directs users — to use data in conjunction with video.
“The data are the breadcrumbs that lead to the video and the context,” he said. “If you’re a team looking to add a scorer, you have to dig deeper than goals and assists. Data can help but you have to look at the video — who’s he scoring against? What time in the game do those points come? You’d hate to commit to someone without knowing all that.”
The analytics age in hockey is not new, but it’s also not yet having an outsized impact on teams when they look to acquire players. That doesn’t mean you can dismiss what the data says on a given player. But, for now, it is part of the equation when trying to figure out who can improve a team.
Maybe that’s a big-enough step.
“You’ll always have teams taking big swings in free agency, and you may wonder if they looked at any data at all on the guy they just signed,” Valiquette said. “Some of them really make you wonder. We’re all fishing for the same thing. I don’t think there’s any shortcuts around doing the homework. The analytics is a shortcut to get to the places you want to be. You go to the wrong place and it’s hard to get back.
“Analytics has to be a piece of your process. It’s a starting point; it’s not the final answer.”
(Top photo of Tom Fitzgerald: Bruce Bennett / Getty Images)
