NHL goals saved above expected: how to read goalie GSAx

NHL goals saved above expected, often shortened to GSAx, compares the expected-goal value of shots a goalie faces with the goals actually allowed. It aims to add shot quality to a simple save count. The statistic is useful context, but it is not an official single-number verdict on a goalie: data providers can build different expected-goal models, and cumulative totals depend on how much the player has played.

This Canadian English guide explains how to read the metric without treating it as a stand-alone forecast. See the match analysis hub for more performance measures and the BetSigy Canadian English blog for other hockey analysis.

How GSAx is calculated

A provider estimates the goal probability of each unblocked shot faced. The probabilities are added to create expected goals against. The model then compares that total with the goals the goalie allowed on those shots. A positive result means fewer goals were allowed than the model expected over that sample; a negative result means more. The exact treatment of empty-net attempts, shootouts, rebounds and shot locations depends on the provider's methodology.

The current NHL goals-saved-above-expected table describes its GSAx measure as expected goals for unblocked shots faced minus goals allowed. A Canadian data page such as PlayerWON's 2026–27 advanced NHL goalie stats displays xG faced, goals allowed and the difference. Read the definition beside the table because the same acronym can sit on top of different shot models.

GSAx is a count, so check workload

A goalie with more starts faces more shots and has more chances to accumulate a large positive or negative GSAx. Compare games played, starts, shots faced and minutes alongside the total. If the provider offers a per-game or per-60 rate, it can help describe performance relative to workload, but rates can be noisy in a small sample. Do not rank a five-game sample against a full season without showing that difference.

Separate regular season and playoffs, and verify the date range. A goalie traded mid-season may appear under a current team label while the season total includes games for an earlier club. Backups may face a different mix of opponents and situations than starters. When a table uses a minimum-start filter, check what it is and whether the goalie you are evaluating is included.

Compare it with familiar goalie stats

Save percentage counts saves relative to shots on goal, treating each shot in that calculation as one attempt. Goals-against average measures goals allowed per unit of playing time. GSAx instead uses a model's estimate of shot difficulty. These measures answer different questions, so disagreement does not automatically mean one is broken. A goalie can have a strong save percentage and a modest GSAx if the shots faced were easier on the chosen model, or a low save percentage with a better-than-expected result against difficult chances.

Look at high-danger chances, rebounds, screens and defensive support if a provider includes them. Those factors can help explain a run of results, but the model may not capture every screen, pass or deflection cleanly. Team defence affects both the quality and volume of shots faced. GSAx is an attempt to account for shot quality, not proof that the goalie alone caused the result.

Why provider rankings differ

Expected goals depends on inputs such as shot location, angle, type, traffic and rebounds. Providers may have different tracking coverage, event classifications and model weights. Some include only unblocked attempts; others may add or exclude specific situations. As a result, a goalie can rank differently across two GSAx tables even when both calculations are internally consistent. Check the provider's methodology and keep one source fixed for a season-to-date comparison.

Also avoid reading the metric as a prediction of the next start. A recent positive GSAx stretch could reflect strong saves, team defence, opponent shot selection, a small sample or some combination. Confirm who is expected to start, consider workload and rest, and compare the opposing offence with the goalie’s previous opponents. The stat describes shots already faced; it does not say what the next game will produce.

GSAx checklist for a matchup review

  • Confirm the table’s season, date range and provider definition.
  • Read GSAx with starts, minutes and shots faced so workload is visible.
  • Check whether the figure is cumulative or normalised per game or time.
  • Keep regular-season and playoff data separate.
  • Compare save percentage and goals-against average as different measures.
  • Verify the starter and team context before drawing a matchup conclusion.

NHL goals saved above expected adds a shot-quality lens to goalie review, which makes it more informative than a raw goals-against count on its own. Its value depends on the underlying model, the workload and the sample. Use GSAx as one description of past performance, compare it with other measures and avoid turning a ranking into certainty about a future start.