Signature stats
Pressure Index and Control %: a picture of game flow, not a prediction
Broadcasts talk about momentum constantly. We built a clean measure of it, tested whether it predicts the next goal, and found that it barely does. We still chart it, because it describes a game well, and we say plainly what it can't do.
Updated · Method version sequences v1
How it's calculated
P_team(t) = Σ impulse_i × 0.5^((t − t_i) / 60 s)
- impulse
- unblocked attempt: its xG; blocked attempt: 0.02; offensive-zone faceoff win: 0.02; offensive-zone takeaway or forced giveaway: 0.03; offensive-zone hit: 0.01
- 60 s
- half-life: an event counts half as much a minute later
PI(t) = (P_home − P_away) / (P_home + P_away + 0.05)
- 0.05
- damping, so a single early shot in a quiet game can't swing PI to ±1
- Control
- |PI| > 0.25; Control % = share of game time a team spends above that line
Pressure Index over one period
Example data: invented for illustration
Invented period: the home team controls the first five minutes, the away team takes over during a power play around the 10-minute mark, then play evens out.
- HOME pressure
- AWAY pressure
- Goal
- Control (|PI| > 0.25)
Worked example
Thirty seconds of pressure
Includes example dataAt 0:00 the home team takes an unblocked shot worth 0.10 xG. At 0:20 it wins an offensive-zone draw (0.02). At 0:25 the away team lands an offensive-zone hit (0.01). What is PI at 0:30?
- Home: 0.10 × 0.5^(30/60) + 0.02 × 0.5^(10/60) = 0.0707 + 0.0178 = 0.0885.
- Away: 0.01 × 0.5^(5/60) = 0.0094.
- PI = (0.0885 − 0.0094) / (0.0885 + 0.0094 + 0.05) = 0.0791 / 0.1479 = 0.53.
- 0.53 is above 0.25, so the home team is in control at 0:30.
PI 0.53: home control. The events are illustrative; the weights and formula are the real ones.
The test: does pressure predict the next goal?
For every goal in our 200-game 2025-26 sample (1,188 goals), we checked which team had the higher Pressure Index just before it. The team with more pressure scored 51–52% of the time. A coin flip scores 50%.
A related myth: the team that just conceded "responds". It doesn't. In the two minutes after a goal, the conceding team generated 0.087 xG and the scoring team 0.084, no meaningful difference.
Reliability
As a description the index is exact: anyone with the same events gets the same curve. As a predictor it fails, and we publish that result.
| Measure | Split-half r | Sample | Verdict |
|---|---|---|---|
| Next-goal prediction (higher PI team scores next)51–52% vs 50% for a coin flip. | — | 1,188 goals, 2025-26 | Noise |
| Post-goal response (xG in next 2 min)0.087 conceding vs 0.084 scoring. | — | 200 games, 2025-26 | Noise |
Split-half r: we split each sample in two (for example odd and even games) and correlate the two halves across players or teams. Near 0 means the stat is mostly noise; near 1 means it is a stable trait.
Limitations
- The weights (0.02 for a blocked attempt, 0.03 for a takeaway and so on) are judgement calls, chosen to be sensible, not fitted. Different weights draw a slightly different curve.
- Hits were recorded unevenly across arenas before 2024-25, which can distort the index in older seasons.
- Power plays dominate the index while they last. Read the ribbon alongside the penalty markers.
FAQ
Is momentum real in hockey?
Teams do go through stretches of sustained pressure, and the Pressure Index shows them. But those stretches barely predict who scores next: in our test on 1,188 goals, the team with more pressure scored next 51–52% of the time.
What does "control" mean on the game chart?
A team is in control when the Pressure Index is beyond ±0.25 in its favour. Control % is the share of the game a team spent there.
Why chart it if it doesn't predict goals?
Because it describes the game well: where each team pushed, how a power play tilted the ice, and whether a lead was protected or earned. We just don't use it as a forecast.
Do teams push back after conceding?
Not measurably. In the two minutes after a goal the conceding team generated 0.087 xG and the scoring team 0.084.