The Possession Paradox

Why the Team That Out-Shoots Everyone Can't Win a Playoff Series 

The LA Kings have made the playoffs five consecutive seasons. Over that stretch, their CF% — Corsi For percentage, the share of all 5-on-5 shot attempts a team generates out of the total in the game — has been the second-most consistent in the league, trailing only Carolina. By the metric that best predicts regular-season success, the Kings have built one of the most reliable teams in hockey.

They have not been past the second round once in that span.

The data predicted their regular season with real accuracy. It predicted their playoff outcomes not at all. That gap is not a flaw in the model — it's the finding.

What Corsi actually predicts

Across five seasons and all 32 franchises, CF% correlates with final regular-season points at r = 0.50 — a real, usable relationship, but far from deterministic. (An r of 0.50 means possession quality and points move together meaningfully; an r of 1.0 would mean possession quality determines points with no exceptions. We're closer to the middle of that range than the top.) Corsi tells you a great deal about whether a team will be good over 82 games. It tells you much less about whether that team will win in June.

Our Cup-winner model makes the size of that gap explicit. Using five seasons of playoff outcomes, the single strongest predictor of who wins the Stanley Cup isn't possession quality or points — it's prior playoff depth, the accumulated experience of an organization having been in high-stakes playoff situations before. In our model, playoff depth predicts Cup winners at roughly 11 times the weight of CF%. Possession quality builds a team's regular-season floor. Something else entirely builds its playoff ceiling.

Three factors that bridge the gap

Our analysis points to three things that separate a team like Carolina — elite possession and playoff success — from a team like the Kings, who have the possession without the results:

1. High-danger CF%, not aggregate Corsi. Shot attempts from the slot and other high-danger areas correlate with final points at r = 0.70 — meaningfully stronger than raw Corsi. Playoff hockey compresses ice and cuts down on low-quality perimeter shots, so a team whose possession advantage comes mostly from low-danger volume loses more of its edge than a team whose possession is built on quality chances.

2. Goaltending that holds under playoff-specific pressure, rather than a regular-season average. A goalie's 82-game save percentage says little about how they perform in a tight, high-leverage series against a team that has scouted them for two weeks.

3. Organizational playoff depth — not individual player playoff experience, but the accumulated experience of the organization itself having been in these exact situations before.

The Vegas case

The clearest illustration is Vegas in 2022-23. Their playoff CF% that postseason was median — nothing special relative to the rest of the field. Their playoff save percentage was 94.0%, the highest of any team in our five-season dataset. By every possession-based measure, Florida should have won that series. The model, built primarily on possession and playoff depth, missed it — because what actually decided the outcome was a level of goaltending performance under pressure that no regular-season number fully captures.

That miss is worth sitting with. It doesn't invalidate the model; it shows precisely where the model's inputs run out. Possession and organizational playoff depth explain most playoff outcomes. They don't explain all of them, and the residual is often exactly this: a goaltender who gets hot at the right moment.

The design implication

If you're building a roster with the Cup as the goal, CF%, points, and coaching system are necessary but not sufficient. The data suggests a fourth dimension has to be built deliberately: accumulated organizational playoff experience. That isn't something you can buy with one trade-deadline rental. It's something an organization accumulates by making the playoffs repeatedly and absorbing what each exit teaches, the way Carolina did across eight playoff trips before their Cup.

So here's the question worth sitting with if you're the one making these decisions: would you rather build a team with 58% CF% that's never been past the second round, or a team with 51% CF% that's been to three conference finals?

The data has an answer. It might not be the one your possession metrics are telling you.