How to frame a hockey analytics question
The step great analysts never skip: turning a vague hunch into an answerable question.
Vague hunch
- "Our PK is bad"
Sharpen it
- Metric?
- Sample?
- Compared to what?
Answerable question
- xGA/2min on the PK
- last 20 games
- vs league median
Decision
- Change personnel?
- Change scheme?
The 30-second version
- Most bad analysis starts with a bad question, not bad math.
- Make it specific, measurable and decision-linked before touching data.
- Name the decision the answer will change — or don't run it yet.
Start from the decision, not the data
Great analysts work backwards from the choice a coach or GM has to make. If no decision changes based on the answer, the analysis is trivia. Ask: who will act on this, and what would move them?
Make it specific and measurable
Replace "our power play is struggling" with "our power-play expected goals per two minutes over the last 15 games versus the league median." Now it has a metric, a sample, and a benchmark — three things every good question needs.
Pre-commit to what would change your mind
Decide up front what result would confirm or reject the hunch. This stops you from fishing the data for a story and keeps the analysis honest.
Newsletter
Weekly hockey analytics
Data-driven takes on performance, scouting, and team strategy. No fluff — just the numbers that matter.
No spam. Unsubscribe anytime.