How to read NHL matchup analysis
Read expected-goal numbers in context, keep goalie uncertainty visible, and distinguish a statistical advantage from a prediction.
In this guide
Check the matchup and goalie status first
A hockey preview is a snapshot. Start with its game date and the teams, then read the goalie wording carefully. “Listed,” “reported” and “confirmed” convey different levels of information. A name appearing in a preview is not independent confirmation that the goalie will start.
Check the latest available information before treating an earlier article as current. A missing goalie report is an information gap; it does not mean that the team has no available goalie. The same distinction applies to injury information: absence from a report is not proof that everyone is healthy.
Expected-goal share is not a win probability
Expected goals assign estimated scoring values to shot opportunities. The values depend on the provider’s model and inputs. They help describe the chances represented in a sample; they are not goals that must eventually be scored. MoneyPuck’s model explanation provides one published example of how an expected-goals model works.
xGF: expected goals for
The expected-goal value credited to a team over the selected sample. Read the situation and date filters alongside the total.
xGF%: share of expected goals
Expected goals for divided by expected goals for plus expected goals against, expressed as a percentage. It measures a share within the sample. Definition: Evolving-Hockey.
A team’s xGF% is not its probability of winning the next game. Nor should similarly named figures from different providers be assumed identical. In WP articles, a “WP-estimated” label distinguishes a WP estimate from an unqualified provider figure.
Read the denominator before the headline number
6 expected goals for
4 expected goals against
60% expected-goal share
The calculation is 6 ÷ (6 + 4) × 100 = 60%. This describes the team’s share of expected goals in that sample. It does not establish a 60% chance of winning tonight.
Now ask which games and situations produced those numbers. Five-on-five data and all-situations data answer different questions. A ten-game stretch and a season total also describe different samples. Compare like with like before interpreting one team’s percentage as stronger evidence than another’s.
How to read a WP hockey article
- Check the preview date. A historical matchup belongs to its original date, even when you find it through search during the offseason.
- Read the statistical scope. Look for the source, season, game window and situation. If a source does not supply an exact through-date, do not assume the latest completed game is included.
- Keep missing information visible. An unavailable goalie or player update should remain an uncertainty, not become a confident statement about availability.
- Separate comparison from prediction. A stronger descriptive statistic does not explain every part of the model’s selection. A confidence label is not a guarantee or a substitute for an explicitly stated probability.
WP’s current writing process arranges paragraphs built from available source records, with selections added separately. When supported information is sparse, an article may be shorter. More words would not resolve missing evidence, and the writing process cannot independently establish that every upstream record is correct.
Use the archive with its original date in mind
The current daily overview offers a public briefing. Complete current matchup articles and detailed selections require an eligible pass. Historical articles remain available under the current archive policy; they are records of earlier analysis, not fresh picks for an upcoming game.
Visit the article list for current coverage and clearly dated archives. Use the track record to examine recorded results over a broader sample, and the FAQ for product questions. This public guide contains no current selections.