Win more chances, win more games.
You already believe it. We checked anyway. Over seven NHL seasons, the teams in the top fifth on 5-on-5 chance share played .600 hockey. The bottom fifth played .372.
On deals signed 2021–2026 by players with 40+ games the season before, teams signed nearly three players who did neither for every one on HQ's list. HQ's list: top third on V/60 who defend.
Does winning the chances mean winning more games?
Yes. We checked every NHL team for seven seasons, 2019-20 to 2025-26: 222 team-seasons, 8,509 games.
| Teams sorted by their share of the 5-on-5 chances, on HQ's chance model | Share of the chances | Win % | Goal difference per game |
|---|---|---|---|
| Bottom fifth | 44.5% | .372 | −0.75 |
| Middle fifth | 50.1% | .515 | +0.06 |
| Top fifth | 55.2% | .600 | +0.60 |
The fair test is the games you haven't seen yet. We split every season into two halves, odd games and even. If you knew a team's chance share in one half, how well could you guess its win percentage in the other? On a scale where 1.0 is a perfect guess, chance share scores 0.62. Shot share scores 0.55. Goal share scores 0.50.
One point of chance share is worth about 2.7 standings points.
A chance is an unblocked shot, counted at its odds of going in. In win %, a shootout counts as half a win. Split by date instead (first half against second) and chances still lead at 0.59, with goals at 0.55 and shots at 0.52. Chances and goals are within luck of each other there.
You made the move. How do you know it worked?
Three common answers, and the question each one leaves open.
“Games won and lost.”
They tell you who won. They can't tell you whether you won because of the player or in spite of them.
“We go by points.”
Goals are volatile, and points follow the goals. They move around more from one season to the next than chances made do.
How closely a number matches itself a season later (1.0 would be perfect), NHL regulars in How to Read HQ's season-to-season check, 2019-20 to 2025-26: chances made 0.84 for forwards and defenders; points per 60, 0.74 and 0.68.
“We go by our model.”
Models are often tested against themselves. To know what a model predicts, you need a benchmark set before the test.
The model doesn't matter if you can't measure its success.
If you can't define success, start there. Here's our bar.
A move worked only if all four happen on the new team.
No partial credit, and a player who gets sent down counts as a miss. Who won the chances is graded with a simpler chance model that only sees where each shot came from, so HQ's model never grades itself.
HQ's list did it more than a third of the time, about three times as often as the average move.
Every NHL move from 2021-22 to 2025-26 by a player with 40+ games and 250+ 5-on-5 minutes the season before: 668 moves.
| On the new team | HQ's list (60 moves) | The average move (668) |
|---|---|---|
| Their team won the chances | 65% | 35% |
| Their team made more chances | 73% | 32% |
| Their team outscored at 5-on-5 | 60% | 37% |
| Their own game held | 62% | 33% |
| All four: it worked | 37% (22 of 60) | 12% (78 of 668) |
HQ's list: players in the top third at their position on V/60 (how much they drive possession and move the puck up the ice) who also defend, the season before the move. Price doesn't come into it.
HQ's list as the app shows it: 60 of 668 moves by players with 40+ games and 250+ 5-on-5 minutes the season before. The research paper grades its registered list, built from the frozen research files, on its own frame: 27 of 55.
It isn't a sure thing. 38 of the 60 moves missed at least one part. Every one of them is in HQ, next to the ones that worked.
Now we hold ourselves to it. These results come from tests against historical data. We wrote down what success would look like before we ran the test. In April 2027 and again in April 2028, we grade HQ's list, as the app showed it that season, against that season's movers, and share the result no matter what it says. April 2027 is the first read on a season that hadn't been played when we wrote the bar down.
Winning takes players who create chances and players who take them away.
Finding one who does either helps. Finding one who does both is where the money is.
The market doesn't pay for that, so they come for less than they're worth.
Tested on the money teams already spent
If you go by points, here's what points cost.
The same chances for about a third less of what the market paid.
On deals signed 2021–2026 by players with 40+ games the season before, the players on HQ's list at signing delivered a goal's worth of chances for every $4.6M. HQ's list: top third on V/60 who defend. The players the market paid for points: one for every $6.8M.
117 contracts
284 contracts
335 contracts
On the same deals, teams signed nearly three players who did neither for every one on HQ's list.
Three reads on every player, and what they're worth.
How much do they drive the play?
V/60: value per 60 minutes. Every pass, carry, entry, battle and recovery gets a price, and the chance itself pays the most. It shows up before the chances do: among all NHL skaters with the same chances made, the ones higher on V/60 made more chances the next season.
Do they create chances?
Chances made: the chances they shoot, or set up with the pass right before the shot, each worth its odds of going in, every 60 minutes. Goal or no goal, it counts.
Do they take them away?
Two lines. The chances they stop themselves: blocked shots, blocked passes, stopped entries. And the chances against with them on the ice, net of partners, opponents and where their shifts start. A player who creates chances and gives them all back isn't helping.
What are they worth?
Worth sets their V/60 against what the market pays for it: under, at or over their cap hit for a pro, against others their age for a junior. The market pays about nothing for defence today, so HQ puts the defence read beside the price, in words. Finding the bargains is one use. Knowing who's worth paying full price for is another.
Every number has receipts
Open any V/60 and see the possessions behind it: who started it, who moved it, who finished it, and what each touch was worth, good or bad. Your scouts can check our work against the tape.
Not another dashboard. Outcomes.
Find the players who drive the play and defend, before the market does.
Know what the market pays for players like them before you sign.
Avoid the multi-million dollar mistake.
Get more from the roster you already have.
Built for the people who make the calls
Three steps to your first decision
HQ can be online in your organization within days of data access.
-
STEP 1database
Bring your data
Event data and the other feeds your organization already licenses.
-
STEP 2tune
Make HQ yours
Scoring profiles price the plays the way your system values them. Change a profile, and every number re-prices.
-
STEP 3task_alt
Make the decisions
Who to sign, who to trade for, and what the market pays, with every number traced to the plays behind it.