What Happened To Thomas Harley?

A deep dive into what's been troubling the Star defenseman this season

Thomas Harley is having a bad season. There’s not really much debate on this point. Just looking at basic box score stats, he’s on pace for just 4 goals and 39 points after scoring at a 16 goal, 51 point pace over the past two seasons. On a team with a +28 goal differential, he sports a -3 plus/minus; worst among the team’s defensemen. By the eye test, he’s been just as bad, with numerous bad turnovers resulting in chances and goals against. And the advanced stats paint an even uglier picture, if possible.

All this comes in his last season playing on a team-friendly bridge deal with a massive $10.6m extension set to kick in next year, leaving Stars fans wondering where this came from and if this is what should be expected moving forward. How does one of the best defensemen in the league over the past two year, a Team Canada Olympian no less, suddenly look so bad at just 24 years old?

I hope to answer that question with a thorough look into Thomas Harley’s season so far. The bulk of this article will be a look at Harley’s microstats courtesy of Corey S.’s AllThreeZones project. Buckle in, there’s a lot (a LOT) to get into.

How Bad Has Harley’s Season Been?

Catch All Metrics

To start, let’s establish how bad Harley’s season has been. Across almost every single value stat available, Thomas Harley is having a massive down year compared to his 2023-24 and 2024-25 seasons. By Evolving-Hockey’s SPAR and xSPAR models adjusted for TOI, he’s had the largest drop-off of any defenseman with a similar workload from their age 22-23 seasons to their age 24 season.

Most other value models agree, whether it’s Hockey Stat’s WAR model, Net Rating, or Louis Boulet’s SPAR model, Harley has gone from a true 1D to at best an average defenseman and and worst a below replacement defenseman actively dragging the Stars down whenever he steps on the ice. The one model that has not significantly downgraded Harley is HockeyViz’s sG model.¹ According to sG, Harley still grades out as a true 1D. The rest of the single value metrics agree though, Harley is going through a hellishly bad season.

Single value stats are mostly based on impact estimates calculated from on/off splits, with adjustments for score, venue, teammates, competition, and zone starts and additional value added for finishing and penalties drawn. The exact methodology varies from model to model with each weighting components differently, but since on/off splits are a major component of every major model I’m going to break down that way for the rest of this article.

On/Off Splits

For most of the season, Harley has performed pretty badly. By relative to teammate xG and corsi measures, he’s been the worst of the Stars’ regular defensemen. Looking at a game log, we can see that the Stars have been better with him on the bench than on the ice for most of the season.

Harley went through an especially bad stretch from roughly game 9 at Nashville to game 18 before being shut down with an injury. Overall though, he’s hovered around the rest of the team’s performance, more often below than above.

Potential Mitigating Factors

Injury

One big story from earlier in the season was Harley being placed on Injured Reserve for 12 games with a lower body injury. Before that, however, it was clear his injury was affecting him, as can be seen in his on off splits. For the ten games before he was shut down, his play was substantially below the rest of the team’s play when he was on the bench. With only 38 games so far this season, that stretch makes up a pretty large percentage of his games played, and is certainly contributing to his poor overall numbers. However, he’s played enough that it shouldn’t be dragging his numbers down this much. Even before that stretch and after he was only roughly as good as the rest of the team and since returning the team has continued to perform slightly worse with him on the ice.

It’s definitely possible, maybe even likely, his injury is not fully healed, but it doesn’t seem like enough to explain the entirety of his drop off from the previous two seasons.

Teammates and Competition

Another major factor in performance, and one included in most models, is teammates and to a much lesser extent competition. In 2023-24 Harley’s most common defensive partner was Miro Heiskanen, himself one of the top defensemen in the league with the two of them sharing about 43% of Harley’s 5-on-5 ice time and the rest played mostly with Joel Hanley, Jani Hakanpaa, and Nils Lundkvist. In 2024-25 Harley’s time with Heiskanen was reduced to just 23% of his minutes and his most common partner was Ilya Lyubushkin at 36% with his third most common partner being Matt Dumba. This season, Harley has played mostly alongside Lyubushkin (30%), Lundkvist (26%) and Alex Petrovic (19%) spending just 13% of his time with Heiskanen.

Although the quality of his partners has clearly decreased, Harley played the majority of his time each of the previous season with fairly low quality linemates. He’s has certainly not been a Heiskanen merchant and was the Stars’ best performing defenseman when Heiskanen missed 32 games with a knee injury in 2024-25.

Using Evolving-Hockey’s Quality of Teammates too, an average of all teammates SPAR/xSPAR value weighted by ice time shared with Harley, his quality of teammates this season is down from 2023-24, but roughly the same as last season, which makes sense for a roster that saw very little turnover. Using the same tool for Quality of Competition, Harley has seen a slight uptick in competition each season, and has seen his relative rank rise in each season as well. As with QoT though, his usage is roughly on par with last season.

Specific teammate and competition chemistry may be playing a role, but Harley’s usage is not that different from his breakout 2024-25 season.

Zone Starts

Finally, perhaps Harley is being given more difficult zone starts this season. While his offensive zone starts have declined each year since 2023-24, per HockeyViz, his defensive zone starts are also at a career low. Relative to the rest of the team, Harley is being used in a more offense-friendly manner than last season, with the Heiskanen-Lindell pairing taking the more difficult defensive assignments. In short, Harley’s zone deployment has been easier this season than it’s ever been.

System of a Down Year

Injury is likely a fairly major factor in Harley’s lackluster play this season, though not enough to explain the magnitude. Another factor that’s harder to quantify is system changes. The Stars fired head coach Pete DeBoer over the summer and replaced him with Glen Gulutzan. While specifics of the system aren’t apparent at the macro level, AllThreeZones’s microstats can provide some clues as to what the Stars are doing differently under Gulutzan.

The downside of microstats is that since they have to be tracked by hand, the sample size is much smaller. There are only nine tracked games for 2025-26, so far all coming from earlier in the season. Of those nine, three came before Harley tried to play through injury (at Colorado, at St Louis, vs Columbus), three came while he was playing through his injury (at Tampa Bay, at Nashville, at Ottawa), and three came while he was on IR (at Calgary, vs San Jose, vs Pittsburgh), so although tracking stats stabilize relatively quickly, keep in mind that the confidence intervals on the 2025-26 season should be fairly large.

Skating

Before getting into the microstats, however, we can look at something that has a much broader sample size and reveals something about how the defensemen are being asked to play.

Across the board, the Stars defense skates is slower than they’ve been in previous seasons. While 2023-24 to 2024-25 saw declines for all returning defensemen, that trend has accelerated this season. While injuries (Harley, Lundkvist, Heiskanen) and age (Lindell, Lyubushkin) are possible explanations for individual players, the overall pattern is quite striking and suggests that defensemen are asked not to skate as much as they did in previous seasons. For a player like Harley, who’s skating is one of his best assets, this seems like a poor fit.

Zone Exits

Defensemen play an outsized role in zone exits. As the furthest players back, they are often tasked with puck retrievals and moving the puck from the point of retrieval out of the zone, either directly or by passing it off to another player to exit the zone.

Retrievals

There is a slight uptick in the share of retrievals recovered by defensemen overall, though not enough to suggest a system change in this regard. However, what’s really notable is that Harley has been the primary puck retriever on his pairing this season.

This chart is a little harder to read, but to sum it up, Dallas is facing more pressure on Puck retrievals than in the past few seasons. As it pertains to Harley, not only is he taking on a larger share of retrievals, but he’s also being pressured more on his retrievals than his partner and escaping that pressure less often than in the past. That is, he’s taking both a higher volume and a harder workload and succeeding less. And despite the easier workload he’s leaving for his partners, they are performing even worse than him on those retrievals than they have in prior seasons. Whether this is a feature of a system change or simply opponents playing differently, is difficult to say. But the combination of higher pressure and Harley’s partners being unable to escape it seems to have resulted in Harley taking on a heavier burden and seeing his own efficiency drop as a result.

Exits

As with retrievals, there isn’t much of a change in terms of which positional player is responsible for exiting the zone. However, when Harley is on the ice, he exits at a comparable rate to a forward. This has seen an even uptick this year, with Harley accounting for almost a full quarter of zone exits when he’s on the ice.

As for the types of zone exits Dallas attempts, here we can see a change. over the past two seasons, but especially this season, passing exits have decreased while carried exits have increased. This is particularly noticeable among forwards.

With Harley on the ice though, that trend is reversed among forwards. Harley himself is passing less and carrying the puck more, which is counter to what I would have expected based on the skating data. Still, the data is fairly clear here that there is a stronger preference for skating the puck out of the defensive zone than passing it. Is that a winning strategy though?

In the aggregate, this seems like a solid strategy. Forward carried exits have increased in efficiency even with their proportion increasing. However for Harley specifically, his carried exits fail at a greater rate as their frequency increases and his defensive partner’s success at passing the puck out of the zone has cratered. As with retrievals, it’s not hard to see why Harley is taking on a larger workload this season with his partner’s inability to help, but he hasn’t been able to make up for it by himself.

Zone Entries

Entry Defense

The last type of microstat defense to mean. Typically Harley has left entry defense to his partner, and together they defend entries at roughly the same rate.

This season their effectiveness has plumeted though. Harely’s on ice numbers are an pretty small sample (only 98 entries defended), but the numbers are really striking. When Harley’s on the ice, the defensive blue line is essentially an open door. This may partly explain why so many of his retrievals are pressured.

The whole team is worse at defending entries than before, but not to nearly the same degree. Harley in particular has gone from being well above the team average at entry defense to well below.

Beyond that, however, Harley and his partner are performing significantly worse than previously in terms of allowing controlled entries. The rest of the team is performing at the same level as previous season, so this appears to be a personnel issue rather than a system issue.

Once across the blue line, Harley and his partners are fairly strong at limiting chances off carried entries, but the volume they allow outweighs the efficiency at defending them.

What’s really poor, however, is the number of chances Harley’s pairing allows off dump-ins. They are noticeably worse than the rest of the team at allowing the opposition to regain possession off dumped puck and generate chances off of dumped in entries.

Entry Offense

Now to the other side of the zone entry coin. Harley’s entry attempt rate is roughly the same as each previous season, as are the entire team’s. Once again however, his partner is less involved than ever.

Unlike with retrievals and exits, the slack has mostly been picked up by forwards instead of Harley. There’s a good reason for the lack of involvement from Harley’s partner. As with other metrics, Harley’s partners have been much less effective than in previous seasons.

The lack of effectiveness is driven mainly by a drop in carried entries, though the share of successful dump-ins for Harley’s partners has declined too. For his part, Harley is carrying the puck into the zone at a higher rate than ever.

One apparent systemic change is a greater emphasis on dumping the puck in versus carrying it in. As with zone exits though, Harley bucks the trend. The effectiveness once in the zone

In a mirror of their zone entry defense, Harley’s pair is much less effective off dump-ins than in previous seasons, both in recovering dump-ins and in generating chances off of them. Harley’s chances per carried entry are similar to previous seasons, but seriously lag the rest of the team.

In-Zone Offense

In the offensive zone, we can see that Harley is less involved than ever, while his partner is more involved. Using weighted shot contributions (1 for the shot attempt, 0.75 for the primary assist, 0.375 for the secondary assist, and 0.125 for the tertiary assist divided by a total value of 2.25), we see that Harley is less involved in offense than he has been in each previous season.

This might be a reflection of a systemic change, though a change from 2024-25. If we focus just on assists (using the same weighting as before, minus the weight for taking a shot attempt), Harley’s involvement is roughly the same as his partner’s while the forwards he share the ice handle more of the puck movement. This means his lack of involvement comes mainly from shooting the puck less while his partner is shooting the puck more.

As far what kinds of shot assists, here’s the story gets uglier.

While the team has put more emphasis on stretch passes, below the goal passes, and low to high passes, when Harley is on the ice, those types of passes have all seen drop offs. That means the shot assists are likely more of the pass around the edges variety than dangerous puck movement.

Turning to the shots Harley takes, individual shot attempts are down, as shown above. In 2023-24 and 2024-25 Harley was a threat to shoot across the entire offensive zone. This season, he’s taken his shots almost exclusively from above the faceoff circles.

This, along with the lack of dangerous passes suggests to me that Harley is not moving as much in the offensive zone as he did before. The reason for that could be injury, lack of trust in a partner he’s covered for in other phases of the game, or something else.

In-Zone Defense

Lastly, a brief look at in-zone defense. It’s harder to pull firm defensive conclusions from the data, but I think shot assists are a good proxy for how Dallas is able to break up puck movement and cycles.

It seems like there has been a strong team focus on reducing royal road passes. In Harley’s minutes however, the number of royal road passes has skyrocketed this season. How much of that is his responsibility can’t be sussed out without looking at film, but that goes a long way towards explaining his poor expected goal and actual goal numbers relative to the rest of the team.

What Does This Mean for Harley and the Stars?

To sum things up: Harley is skating less, but so is the entire Stars defense. Harley is handling a larger share of puck retrievals and zone exits than last season and seeing less success than previously but his partners have been especially bad in this regard. When defending zone entries, both Harley and his partner have been abysmal at preventing entries, and particularly poor at defending dump-in pucks. On entries, Harley’s partner is similarly ineffective and entries are less dangerous when he’s on the ice than on the bench. In zone, Harley has been less involved in shooting the puck and appears to be more static on offense than he was from 2023-25, along with being ice cold as a shooter. Finally, on defense, Harley and his partner have been awful at preventing royal road passes.

All things taken together, I actually would place most of the blame for Harley’s poor season at the foot of his defensive partners. It’s clear that they have performed extremely poorly and Harley has compensated by attempting to do more. While this may demonstrate a lack of trust in his partner, a trust he should lean into more to make the system work (Do Your Job), it is understandable and indicates the problem may not be with Harley. Going back to the total value stats cited at the beginning, it’s interesting that sG has continued to credit Harley as a 1D and a net positive on shot rate. Getting back to playing with an effective partner could go a long way to improving Harley’s impact number.

Let’s end on an optimistic note.

Of the defensemen comparable to Harley from the very first chart of this article, most saw a bounce back to their previous level in their age 25 season. While Harley is having a bad year, there’s a lot more information in the 157 games he played as a 22 and 23 year old than in the 38 games he’s played as a 24 year old.

1. In my opinion the gold standard for public hockey analytics.