42-60%: O-zone-start spread. 0.41: Usage vs results (corr). 0.02: xG a zone start is worth (12s). 1st: Public PWHL usage map Not all ice time is equal. A coach can start one player in the offensive zone shift after shift, fresh off a whistle 180 feet from her own net, and bury another in the defensive zone against the other team's top line. Over a season that shapes a player's raw numbers as much as her ability does. Men's-hockey analysts have mapped this for years (the Vollman player-usage charts, Behind the Net's zone-start ratios). For the PWHL it has not been done, for a dull reason: the public feed records faceoff locations but never which way a team is attacking, so you cannot tell whose offensive zone a draw is in. We got it back the hard way. A team's attacking direction in a period is written in where it shoots: shots cluster toward the net you are attacking. Take the average shot location for each team in each period and the direction is unambiguous in more than 99% of cases. From there, every non-neutral faceoff becomes an offensive- or defensive-zone start for the skaters on the ice. That reconstruction is the new thing here. What follows is the first public 5-on-5 usage map of the league. The usage map Each dot is a skater. Left to right is her offensive-zone-start share (how sheltered her deployment is); bottom to top is her on-ice expected-goals share at 5-on-5. The dashed line is the league trend, and it slopes up: sheltered skaters post better on-ice results (correlation 0.41). Hold that thought, because the slope is not what it looks like. What a zone start is actually worth It is tempting to read the upward slope as proof that sheltering drives results. It mostly does not. We measured the draw itself directly: the net expected goals a team generates in the 12 seconds after an offensive-zone draw, minus the same after a defensive-zone draw, is about 0.02 expected goals per faceoff . That is tiny, and it lines up with the men's-hockey finding (Eric Tulsky's work) that a zone start's effect washes out within about ten seconds. A single draw barely moves a game. So why do sheltered players have better numbers? Because the arrow mostly points the other way. You earn offensive starts by being good : a line that hems the other team in draws the next faceoff in the attacking zone. Zone-start share is partly a reward for driving play, not just a gift from the coach. The correlation is real, but it is mostly good players collecting offensive draws, not the draws making them good. The honest reading of this map: zone starts tell you how a skater is used, not how good she is. Reading the map descriptively The right edge is the sheltered group: Abby Newhook (60%), Julia Gosling (58%), Anna Segedi (58%) take the most offensive-zone draws in the league. The left edge is the buried group: Abby Hustler (42%), Johanna Fällman (43%), Kendall Cooper (43%) live in their own end. And the left edge is harder than it looks, because zone start and quality of competition travel together: the defender who starts in her own zone is usually the one matched against the other team's top line. Two penalties for the price of one. That makes the lower-left of this chart the toughest job in the league, and a skater who sits above the trend out there is doing real work. Above and below the trend (an annotation, not a rating) Distance above or below the league line is worth a look, but we are deliberately careful with it. It is a raw, descriptive gap, not a competition-adjusted rating: it does not account for the tougher opponents the buried skaters face, and because zone starts are partly earned, the line over-corrects genuine play-drivers. Its split-half reliability across the season is only r = 0.33, which is too soft to rank anyone. So treat the names below as flags to watch, not a leaderboard. The competition-adjusted rating job belongs to a model that controls for opponents and teammates, which on this site is RAPM, not this chart. Above the trend. These skaters out-perform what their usage predicts, and for the buried ones the gap is conservative: the competition we did not adjust for is working against them, so the real edge is probably larger. Kelly Pannek (F, 47% o-zone starts, 57% on-ice xGF) : buried in her own end yet still comes out ahead of the usage trend.. Kendall Coyne Schofield (F, 53% o-zone starts, 60% on-ice xGF) : comes out above the usage trend.. Lee Stecklein (D, 48% o-zone starts, 55% on-ice xGF) : comes out above the usage trend. Below the trend. Decent-to-soft starts, results that trail the league pattern. Worth a second look at role and linemates before any conclusion. Taylor Baker (D, 49% o-zone starts, 38% on-ice xGF) : lands well below the usage trend.. Chloé Aurard-Bushee (F, 51% o-zone starts, 40% on-ice xGF) : lands well below the usage trend.. Taylor House (F, 48% o-zone starts, 29% on-ice xGF) : lands well below the usage trend. What this read isn't Usage, not ability. The map shows how skaters are deployed. The zone start itself is worth almost nothing per draw, so do not read position on this chart as a talent ranking.. Zone starts are partly earned. Good lines draw offensive faceoffs by hemming teams in, so o-zone-start share is part cause and part consequence of strong play. The trend line is descriptive, not a shelter curve.. Competition is not separated here. Buried deployment and tough matchups are tangled together, and this chart adjusts for neither. The over/under-trend gap is an annotation; the competition-adjusted rating is RAPM's job.. 5-on-5, on-ice, our xG. Special teams are excluded, on-ice xGF% reflects the whole unit and the opponent, and the expected-goals values are our model's. This is shot-quality-weighted, not a possession measure (we have no zone-entry or possession-time data). Methodology Attacking direction (the new piece). For each game, period, and team we take the mean x-coordinate of that team's shots; a mean past center means it attacks the high-x end that period. This was unambiguous in over 99% of period-team splits, so the zone classification is reliable.. Zone starts. Each non-neutral 5-on-5 faceoff is an offensive- or defensive-zone start for every on-ice skater, by her team's attacking end. Neutral-zone draws are excluded (standard). OZS% = offensive starts / (offensive + defensive). On-the-fly changes are excluded by construction (only faceoffs count).. The zone-start effect. Net expected goals for the attacking team in the 12 seconds after a draw, offensive minus defensive zone, pooled across the league.. The trend and the annotation. A least-squares line of on-ice xGF% on OZS% gives the descriptive trend; the over/under gap is the residual, shrunk for sample and split-half tested (r = 0.33). Skaters qualify with 150+ non-neutral 5-on-5 starts and a 5-on-5 on-ice-event floor.. Prior art. Zone-start usage mapping follows Gabe Desjardins (Behind the Net), Eric Tulsky (the small, fast-decaying start effect), and Rob Vollman's player-usage charts; deployment visualization echoes Micah Blake McCurdy (HockeyViz). The PWHL-first piece is recovering attacking direction from shot coordinates.. Numbers regenerate on every build. A play-by-play re-sync or an xG recalibration reshapes the map. This pairs with our shift-age Fatigue Efficiency read (defenders fade late partly because they are buried in their own end, which you can now see here). The clean next step is to fold a zone-start term into RAPM so the rating itself is deployment-adjusted.