Scoring Chances (SCF) in Hockey

Nobody owns the definition. "If you were to ask 30 coaches the definition of a scoring chance, you will get 20 or more different answers," Garret Hohl wrote at Hockey-Graphs in 2017, and he noted that the blogging community's usual shortcut was the "home plate" area in front of the net. A 2022 analytics column on the Seattle Kraken's NHL.com site made the same point, noting that teams define chances their own way, and walked through the personal checklist broadcaster JT Brown uses.

The public, repeatable version came from War-on-Ice. On December 31, 2014, the site published its definition zone by zone, built on its own research into where goals come from and on David Johnson's earlier work on rush shots. War-on-Ice's run ended about two years later, after NHL teams hired its founders.

Natural Stat Trick carries the definition today. Its glossary describes a scoring chance "as originally defined by War-on-Ice," restates it as a points system, which is how it's usually explained now, and publishes SCF, SCA and SCF% for every team and skater back to 2007-08.

MoneyPuck, the source this site uses, doesn't define a scoring-chance stat in its glossary, so the league numbers below come from its closest equivalent, the medium- and high-danger shots its expected goals model rates at 8% or better.

Natural Stat Trick puts a points value on every shot attempt in the offensive zone and keeps the ones worth 2 or more. Attempts from the neutral or defensive zone aren't scored at all. The value has four parts:

• Where the attempt was taken: 1 point for the outer areas of the zone, 2 for the high slot and the two wedges beside the net-front box, 3 for the net-front box itself
• Rebound bonus: +1 for an attempt within 3 seconds of another blocked, missed or saved attempt, with no stoppage in between
• Rush bonus: +1 for an attempt within 4 seconds of any event in the neutral or defensive zone, with no stoppage in between
• Block penalty: −1 if the attempt was blocked, since the league logs a blocked shot where it was blocked

Here's a worked example with invented plays. Over one shift a team takes four attempts. The point shot is worth one point and the wrister from the high slot two. A backhand from the outer edge of the zone two seconds after a neutral-zone takeaway is worth one plus one for the rush, and a blocked attempt from the net-front box is worth three minus one for the block. Three of the four are scoring chances. None is high-danger, which takes a three.

War-on-Ice's original post put the same rule a simpler way. From the low-danger area, only unblocked rebounds and rush shots count. From the medium-danger area, every unblocked attempt counts, and from the high-danger area everything counts, blocked shots included.

SCF counts how often a team gets a look worth having. It's a step up from Corsi, which counts a point shot through three bodies the same as a rebound at the crease, and a step short of expected goals, which gives every attempt its own price. Chances sit in between. That's the appeal. You can count them, and you can picture what each one looked like.

Natural Stat Trick's five-on-five numbers for four Red Wings forwards, pulled for this site's research in September, show what the scale looks like. Detroit got 47.0% of the scoring chances with Dylan Larkin on the ice, 46.3% with Emmitt Finnie, 47.4% with Andrew Copp and 49.0% with Marco Kasper. All four sat under break-even.

League-wide benchmarks come from MoneyPuck's medium- plus high-danger shots, the nearest thing in this site's data. At five-on-five in 2025-26 that cut was 24.0% of unblocked attempts and produced 53.3% of the goals. Ottawa led the league with 56.4% of those shots in its games, and Vancouver finished last at 42.1%. Detroit was 21st at 49.2%, just under the median team's 49.9%. The Wings generated 9.40 of them per 60 minutes (21st) and allowed 9.71, exactly the median.

Why do chances matter more than attempts? A MoneyPuck chance went in 13.6% of the time at five-on-five last season, and everything else 3.8%. Hohl found a similar conversion rate for the home-plate area, where shots scored about 14% of the time and produced 78% of the goals on 46% of the attempts.

Checking expected goals against the chance count

Copp and Kasper both finished above 50% in NST's five-on-five expected goals share last season (51.9% and 52.0%) and below it in scoring-chance share (47.4% and 49.0%). SCF% counts a two and a four the same, while xG gives each attempt its own value, so the two shares can split. So which number do you trust? For a forward's process, expected goals, since it answers the question a split raises (what kind of chances each side got).

Measuring who creates chances

Individual scoring chances (iSCF) count only the shooter's own attempts. In NST's five-on-five numbers, Kasper created 129 in 963 minutes, about 8.0 per 60, more than Copp (6.5) or Finnie (6.7) managed in more ice time. He scored nine goals all season. A forward who keeps getting chances and isn't scoring raises a finishing question, and that's a different conversation from one who never gets there.

Ranking the league's chance creators

MoneyPuck's version ranks shooters the same way. Connor McDavid led the league with 168 medium- or high-danger shots in all situations, about two and a half times the 68 of the median forward who played at least 60 games. Alex DeBrincat was ninth with 138, and Larkin tied for 21st with 121. Detroit's two best chance creators were its two leading goal scorers, with 41 and 34.

Nobody agrees on the definition

Hohl's 30 coaches would give you 20 or more definitions, and NHL teams keep their own. NST's formula is a public convention that anyone can reproduce from the play-by-play, and a team's internal chance count for the same game can come out somewhere else entirely.

A two counts the same as a four

Every qualifying attempt's worth one chance, so a high-slot wrister and a rebound at the crease get equal credit. Hohl's case against binning applies here too, since the line between a one and a two is a cliff with nothing on either side of it.

Sites draw different lines

NST had Larkin at 47.0% of the scoring chances, and MoneyPuck's closest cut had him at 46.6%. For Copp the same two numbers were 47.4% and 52.3%, a gap big enough to flip him from the wrong side of break-even to the right one. Compare chance shares from one source at a time.

Location and timing are all it sees

The formula's only inputs are where the attempt came from and how soon it followed another event. A screened point shot stays a one-pointer, and a one-timer off a cross-ice pass gets no extra credit, because the league's feed doesn't log screens or passes.

Does every goal count as a scoring chance?

No. A goal is a shot attempt like any other in NST's count, so a point shot that beats a screened goalie goes in the books as a goal but not a scoring chance. NST tracks the goals that came from chances separately (SCGF), along with a shooting percentage on scoring-chance shots.