Where does Trevor Zegras belong? Part 2: does the job change?
Before asking whether he's better at centre, ask whether centre is a different job. Ice time, teammates, competition, zone starts, score state — and the control that turns a 10-point finding into half a point.
Part 1 built the thing that didn't exist: a per-game record of whether Trevor Zegras played centre or wing across 2025-26. Thirty-nine games at centre, fifty-two on the wing, every row published.
The obvious next question is whether he's better at one. This post isn't that either. It's the question you have to answer first: when the position on the sheet changes, does the job actually change? If moving to centre also means more ice time, softer teammates, tougher opponents and different game situations, then any performance gap you eventually find is measuring four things at once.
So: ice time, teammates, competition, zone starts, score state — centre versus wing.
The trap, and the control that catches it
Here is a number that looks like a finding. On the wing, Zegras spent 40.5% of his even-strength minutes with his team trailing. At centre, only 30.1%. A ten-point swing. You could write a whole paragraph about being trusted with the lead.
It's an artifact. His centre games are bunched into the back third of the season, and the Flyers were simply a better team by then — they trailed less, so everyone on the roster played fewer minutes from behind. It has nothing to do with Zegras.
The fix runs through the whole post: every number is also measured against his own team's other forwards, in the same games. Against that baseline the ten-point gap collapses to half a point. Whatever changed, it wasn't his deployment.
This is the single most important thing here. The raw split says trailing time fell by 10.4 points. The controlled split says 0.5. One of those is a story about a hockey player and the other is a story about a calendar, and you cannot tell which by looking at the raw number.
What actually changed: nothing we can measure
Each row below is centre minus wing, after the teammate control. The bar is the estimate; the shaded band is the 95% interval from resampling games 4,000 times. The white line is zero — no difference.
Every single interval crosses zero. Not one difference survives contact with the sample size.
But "we found nothing" is doing two different jobs in that chart, and the distinction matters more than the headline:
- Ice time is a real null. 14.50 minutes at 5v5 per game at centre; 14.50 on the wing. The interval is ±1 minute — narrow enough to say his workload genuinely does not change. That's a finding, not a shrug.
- Teammates and competition are near-nulls. Both intervals are tight and sit almost on top of zero. If anything he faces marginally easier competition at centre (−0.010, interval −0.020 to +0.001) — the opposite of the "earned the hard matchups" story.
- Zone starts are genuinely unresolved. This is the one live candidate. He starts in the offensive zone 4.4 points more often at centre than his own teammates do — the largest effect on the board, and it survives the calendar control (the raw gap is 6.1 points, so roughly a third of it was era). But the interval runs from −4.3 to +12.7. Thirty games cannot settle it. Anyone telling you they know the answer here is reading a point estimate as if it were a measurement.
One thing that holds regardless: he is sheltered in both roles. Wherever he lines up, he starts in the offensive zone 7 to 11 points more often than the other Flyers forwards. Moving him to centre didn't change that; it may have deepened it.
The one difference that is real
Linemates aren't a noisy estimate — they're a fact about who stood next to him. And they change completely.
On the wing he was Christian Dvorak's winger — three quarters of his even-strength minutes. At centre, Dvorak drops to 12% and Owen Tippett becomes his winger at 63%, with no other centre above 18%, because he's the one in the middle now.
That's worth more than it first appears. Nothing about linemates was used to build the role labels in Part 1 — those came from the published lineup and from who took faceoffs. So this is a third, independent source agreeing with the other two. If the labels were wrong, this chart would be mush.
Why a pile of nulls is the good outcome
It would have been more fun to find that centre-Zegras plays twelve tougher minutes a night. It would also have wrecked the actual question. Every one of these variables is a confounder for a performance comparison: if the job changed, then any difference in his numbers could be the job rather than the player.
What the chart says is that the job barely changes. Same ice time, same calibre of teammate, same calibre of opponent, same game situations. Which means a performance comparison — Part 3 — isn't fighting a tangle of confounds. It has one thing to control for, zone starts, and that is a single variable with a well-understood adjustment.
The binding constraint was never confounding. It's that 30 games is not many, and it will still be 30 games when we get to the part everyone actually wants.
What this doesn't cover
- It's all 5v5. The teammate, competition, zone and score numbers come from even strength, because that's where the site's deployment machinery lives. His power-play time is about 3.4 minutes a night in both roles — so the special-teams half of the centre-versus-wing argument is untouched by anything above.
- "Quality" here means one specific thing — a teammate's or opponent's net xG-RAPM, ice-time weighted. It mixes forwards and defencemen, because that is the definition the player cards already use, and inventing a second one for this post would have been worse.
- Regular season only, so the ten playoff games sit this one out. They're all centre games, and including them would reintroduce exactly the era confound the control exists to remove.
- The eight disputed games from Part 1 are still in here, labelled by what he did rather than what was planned. If you watched one of them and we got it wrong, that's still worth telling us — the offer from Part 1 stands.
Part 3 is the performance split, and it will be pre-registered before anything is computed: what would count as an answer, decided in advance, because role changes are a response to how a player is going and it is very easy to measure a coach's judgement and call it a player's fit.
Expect it to be inconclusive. That's not pessimism — it's what 30 games buys, and this post is the reason we now know that's the actual limit rather than a mess of confounding.