DEV Community

Agent Workshop
Agent Workshop

Posted on

How to Compare NHL Goalies Before a Game (2026): Starter Certainty, Rest, Defense and Chance Quality, Kept Separate

Full disclosure: I built a goalie-comparison tool on Capafy, so I'm biased. Everything below works with a notes app and the box scores you already read. You don't need the tool to use the method.

TL;DR

Most pre-game goalie takes go wrong in the same way: four different questions get blended into one vibe. "He's hot." "He's tired." "That team gives up nothing." "He's been lucky." Each of those may be true, but they are separate claims with separate evidence.

The fix is a one-page card with four boxes you fill in order, then a short list of angles you are allowed to write, and a list of things you still need to verify before you publish.

  1. Starter certainty: who is actually in net, and how sure are you?
  2. Workload and rest: back-to-backs, recent starts, travel.
  3. Defensive context: what the skaters in front of each goalie usually allow.
  4. Chance quality: how dangerous the shots faced have been, kept apart from results.

Then: angles that each point to a box, scenarios for "if X starts", and a verification queue.

Why keep the four boxes separate?

Because they change on different clocks and come from different sources.

Starter news can flip an hour before puck drop. Rest is a calendar fact. Defensive context moves slowly over weeks. Chance quality is noisy over a handful of games. If you blend them, one late lineup change quietly breaks your whole take, and you won't notice which part broke.

Separating them also keeps you honest. "His save percentage is down" is a result. "He's been facing more dangerous shots" is context. "He's playing worse" is a conclusion. Readers can tell when a writer jumps from the first to the third without the second.

Step 1: Starter certainty

Write each team's expected starter and a certainty label:

  • Confirmed: the team or a reliable beat reporter said so today.
  • Likely: strong pattern (for example, the backup started last night), but nothing official.
  • Uncertain: no word, or conflicting reports.

Write down your source and the time you saw it. If the label is "Likely" or "Uncertain", everything below for that team gets a small asterisk in your head. Don't write "X gets the start" unless the label is Confirmed.

Step 2: Workload and rest

For each probable starter, list:

  • Days since last start.
  • Starts in the last 7 and 14 days.
  • Whether either team played yesterday (a back-to-back).
  • Notable travel, if you know it.

Only record what you can check. "He looked gassed" is not a workload fact; "third start in four nights" is. If you don't know something, leave the line blank and add it to the verification queue instead of guessing.

Step 3: Defensive context

This box is about the team, not the goalie. Note what you know about the skaters in front of him: injuries on the blue line, how many shots and how many good chances they usually give up, and whether the opponent's offense is strong or weak.

The point is to stop crediting or blaming a goalie for things his defense does. A goalie behind a team that gives up few clean chances may look great on paper for reasons that aren't his.

Step 4: Chance quality

Now look at the shots themselves, if you have a source for them: shot locations, rebounds, rush chances, power-play time. Write it as a description of the shots, separate from the goals.

Two goalies can both have allowed three goals last game. One faced several clean looks from the slot; the other let in a long shot. Those are different stories. Keep "what he faced" and "what went in" on separate lines.

If you don't have chance data, say so on the card. "No chance-quality source" is a fine entry. Inventing it is not.

Step 5: Angles, each tied to a box

Now write two to four coverage angles. Every angle must point at the box that supports it. For example:

  • "The rested goalie vs. the one on a back-to-back" (Box 2).
  • "A goalie whose numbers dipped while facing harder shots" (Boxes 4 and results).
  • "A blue line missing two regulars" (Box 3).

If an angle can't point to a filled box, it isn't an angle yet. Move it to the verification queue or drop it. This one rule removes most of the confident-sounding takes that turn out to be wrong.

Step 6: Conditional starter scenarios

For any team whose starter isn't Confirmed, write short "if" versions:

  • If the usual starter plays: which angles still hold?
  • If the backup plays: which angles flip, and which new one appears?

This is the part that saves you when the lineup changes late. You already have the second version written; you just pick it.

Step 7: The verification queue

End the card with a checklist of things you must confirm before you publish:

  • Official starter announcement for each team.
  • Any injury news on defense.
  • Rest facts you marked as unknown.
  • Any number you plan to quote, checked against its source.

Publish only what has cleared the queue. Anything still open is either labeled as uncertain in the piece or left out.

A worked example (with made-up placeholder teams)

To show the shape, here's a card for two placeholder teams. These are example entries, not real players or stats.

  • Starter certainty: Team A starter Confirmed (team post, 10:05). Team B: Likely the backup, since the starter played yesterday; nothing official.
  • Workload: A's goalie on two days' rest. B's probable starter last played four days ago; B is on a back-to-back as a team.
  • Defense: A missing one regular defenseman. B: no known changes.
  • Chance quality: A's goalie faced several rebound chances last game. No source for B.
  • Angles: "Rested backup vs. a team on a back-to-back" (Box 2). "A thinner blue line in front of A" (Box 3).
  • Scenarios: If B's usual starter plays instead, the rest angle mostly disappears.
  • Queue: B's official starter; A's defense lineup; chance data for B.

That card takes ten minutes and keeps every claim in the piece traceable.

Common mistakes

  • Treating "Likely" as "Confirmed." Write the label every time.
  • Using results as context. Goals allowed belong next to chance quality, not instead of it.
  • Blaming the goalie for the defense. Fill Box 3 before you judge Box 4.
  • Filling blanks with guesses. A blank plus a queue item beats a confident wrong number.

Where my tool fits

I built NHL Goalie Matchup Brief for this exact step. You paste one game's goalies, workload and rest notes, and chance-context notes, and you get a goalie comparison card with starter certainty, rest, defensive context and chance quality kept separate, plus evidence-bound coverage angles, conditional starter scenarios and a verification queue. Uncertain starters and unsupported angles are marked uncertain, never presented as fact. It works from the notes you paste. On Capafy it's $9.99/week, $24.99/month or $149.99/year.

You can do all of the above by hand, though. The four boxes are the method; the tool just makes the card faster.

Disclosure: I built NHL Goalie Matchup Brief and sell related tools on Capafy.

Top comments (0)