Signature stats
Transition attempts: a rush definition that works on modern data
Rush chances matter: they come against a defence that is not set. But the public definition was designed when the feed was denser, and on today's data it barely fires. We rebuilt it around possession chains.
Updated · Method version sequences v1
How it's calculated
We first split every game into possession chains. The team with the puck changes on faceoffs (the winner), shots (the shooter), giveaways (the other team), takeaways (the taker) and hits (the team that was hit). A new chain starts on a change of team, a whistle, or a gap of more than 20 s with no events.
is_transition = origin ∈ {TK, opp GV, recovery, hit-forced recovery, FO win} AND origin zone ∈ {NZ, DZ} AND (t_shot − t_origin) ≤ 10 s AND same play
- origin
- the event that started the attacking team's chain
- origin zone
- zone of that event from the attacking team's point of view, computed from coordinates
- same play
- no whistle between the origin and the shot
We store the time from origin to shot and the origin type on every attempt, and we keep NST's rule as a separate compatibility flag so you can compare the two.
Share of 5v5 attempts and xG flagged
Research sample: 200 games from 2025-26 (2025020500–699)
NST's rush rule flags 1.7% of 5v5 attempts and 0.7% of xG; the transition definition flags 8.1% and 8.6%.
- Share of attempts
- Share of xG
xG per 5v5 chain, by how the chain started
Research sample: 200 games from 2025-26 (2025020500–699)
Opponent giveaways in their own zone start the most dangerous chains (0.050 xG); neutral- and defensive-zone faceoff wins the least (0.009).
Worked example
A rush the old rule misses
Includes example dataAt 12:04 a defenceman takes the puck off an opponent in his own zone. The feed records nothing else until his team's shot from the slot at 12:11.
- NST rule: the last event before the shot is the takeaway, 7 s earlier. That is more than 4 s, so the shot is not a rush.
- Transition rule: the chain started with a takeaway in the defensive zone, the shot came 7 s later, and there was no whistle. It is a transition attempt.
- Nothing about the play changed; only the rule did. Most rushes look like this in the modern feed: a turnover, a zone exit and entry that go unrecorded, then a shot.
The shot is flagged as transition, with origin "DZ takeaway" and 7 s origin-to-shot. The timings are illustrative; the rule is the real one.
Where offence comes from
Because every chain has an origin, we can ask which starts produce the most danger. Opponent giveaways in their own zone are the most valuable start; faceoff wins outside the offensive zone are the least.
| Chain origin | xG per chain |
|---|---|
| Opponent giveaway in its own zone | 0.050 |
| Opponent giveaway in our zone (counter-attack) | 0.044 |
| Hit-forced recovery in the offensive zone | 0.040 |
| Opponent giveaway in the neutral zone | 0.039 |
| Recovery after an opponent attempt | 0.034 |
| Offensive-zone takeaway | 0.032 |
| Offensive-zone faceoff win | 0.030 |
| Defensive-zone takeaway | 0.018 |
| Neutral- or defensive-zone faceoff win | 0.009 |
The median 5v5 chain in the feed is one event and 4 s long (90th percentile: three events, 29 s). Chains are observed possession segments, not true possessions, so we anchor every definition to recorded events rather than to guessed carries.
Reliability
Transition shares are dependable at team level over a few weeks. At player level they are fair, so we always show the counts next to the rates.
| Measure | Split-half r | Sample | Verdict |
|---|---|---|---|
| Team transition share of attemptsTeam-level numbers are good. | — | About 20+ games | Moderate |
| Player transition attemptsShown with counts. | — | Half a season or more | Moderate |
| Offence source mix (team or line) | — | About 20+ games | Moderate |
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
- Zone exits, entries and carries are not in the feed. We infer the rush from where possession started, not from how the puck crossed the blue line.
- Faceoff wins in the neutral zone start many chains but rarely produce a rush shot; they are included because the rule is about possession origin, not speed.
- Legacy seasons (before 2024-25) recorded far fewer giveaways and more takeaways, so chain origins shift between eras. Compare seasons within a recording regime.
- Goal tracking gives ground truth for how scoring plays started (rush, rebound, cycle). We will check the transition flag against those labels and publish the agreement rate on the goal anatomy page.
FAQ
What is a rush shot in hockey stats?
A shot that comes quickly after a team gains the puck outside the offensive zone, before the defence is set. We call ours transition attempts: an attempt within 10 s of a turnover, recovery or faceoff win in the neutral or defensive zone, in the same play.
Why does Natural Stat Trick show so few rush shots?
Its rule needs a recorded event in the neutral or defensive zone within 4 s before the shot. The NHL feed logs about 317 events a game, so most rushes have no event in that window. In our 2025-26 sample the rule flagged 1.7% of 5v5 attempts.
Can I still see the NST rush number?
Yes. We store NST's rule as a compatibility flag on every attempt, next to the transition flag.
Are transition shots more dangerous?
Per shot on goal, yes against a set cycle: goalies saved .931 of transition shots and .896 of cycle shots in our sample. Rebounds (.849) are the most dangerous class.