Will Smith, away from Celebrini: the one stat that decides the Star Forward slot
Every roster plan we have written quietly leans on one unproven assumption: that Will Smith is a Star Forward, the contention clock's second-biggest swing after the goalie. The scorecard bar that tests it is the most important young-core line in the whole document, and it reduces to a single number. Across the 121 five-on-five minutes Smith played away from Macklin Celebrini in 2025-26, his high-danger chance share was 44.4 percent. Below water. That is the entire Star Forward question in one stat, and it deserves its own autopsy.
1. The gap between the two numbers is the whole story
| Measure | Number | What it says |
|---|---|---|
| On-ice xGF%, all situations | 56.6% | A top-line possession number, mostly logged next to Celebrini |
| High-danger chance share AWAY from Celebrini | 44.4% (121 min) | Below water the moment the anchor leaves the ice |
| Shooting talent vs league | +17.2% | A real, portable finisher's signal |
| Defensive impact (percentile) | 19th | Bottom-quintile; the line does not tilt because of his own work |
A driver carries his number with him when the elite center sits. Smith's collapses from 56.6 to 44.4 the instant Celebrini leaves. That is not a sample artifact; it is the signature of a player who finishes what a driver creates rather than creating it himself.
2. The Sherwood experiment ran the test for us
The cleanest evidence is a natural one. When the Sharks added Kiefer Sherwood at the deadline and the top line started humming, the line's expected-goals rate spiked, and Celebrini's counting numbers went up while Smith's went down by about 20 percent. Same center, same role, a better finisher inserted on the other wing, and the chances simply stopped routing back to Smith's stick. Read that carefully: when a superior shooter took the high-value ice, Smith's production left with it. A driver's production survives a linemate upgrade. A passenger's redistributes to whoever shoots better. The Sherwood months redistributed it.
3. The two readings, and the dollar figure between them
Reading A (role shifted): his number dropped because his job inside the line changed, not because he got worse. The talent is intact; give him his own line and the creation comes.
Reading B (Celebrini-inflated): the 59 points were partly a Celebrini subsidy, and the subsidy came out of the ledger the moment a better finisher claimed the space.
The difference between those two readings is the difference between an $8M-plus long-term deal and a $5M bridge in the 2027 summer, the same summer Celebrini's extension lands. The away-from-Celebrini high-danger share is the only number that separates them, and right now it reads 44.4, which points at Reading B.
The verdict
On the current evidence, Smith profiles as an elite finisher who needs a driver, not a driver himself, which means the contention clock's Star Forward slot is not yet filled. It is pending. That is not a downgrade of the player; a +17.2 percent shooting talent on a first-line right wing is a genuinely valuable, genuinely portable thing, and the Chernyshov-Celebrini-Smith line should let him keep feasting next to the anchor. But "Star Forward" in the Cup Checklist sense means a player who drives a line of his own, and the 44.4 percent says he has not shown that yet. The good news is that the test is already written: the scorecard asks him to hold a positive line across 200-plus minutes away from Celebrini in 2026-27. Clear that bar and the slot fills and the contract climbs toward $8M. Miss it and he is a magnificent passenger on a $5M bridge, and the Sharks go shopping for the Star Forward elsewhere. Watch the shifts when Celebrini is on the bench. That is the whole evaluation.
Built on docs/players/Will_Smith_2025-26_Review_Brisbee.md (the 56.6% on-ice xGF%, the 19th-percentile defense, the +17.2% shooting talent, the pre/post-Sherwood production split, the contract bands) and the season scorecard's away-from-Celebrini bar (44.4 HDCF% in 121 min). The 2027 RFA timing per the cap model. Reading B is the bear case, named so it is priced rather than ignored.
