Two goalies. Same save percentage: .920.
One plays behind the best defense in the league. The other plays behind the worst.
Which one is actually better?
Traditional save percentage cannot tell you. But GSAx can.
GSAx — Goals Saved Above Expected — is to hockey what xERA is to baseball: a metric that separates luck from skill, and defensive support from genuine goaltending talent. For NHL bettors, it is one of the most powerful tools available.
The Problem with Save Percentage
What SV% Measures
Save percentage (SV%) is straightforward:
SV% = Saves ÷ Shots Faced
A goalie with .920 SV% stops 92% of shots. Simple enough.
Why It Misleads
The problem: not all shots are equal.
| Shot Type | Goal Probability |
|---|---|
| Point shot, no screen | ~2% |
| Slot shot, no screen | ~12% |
| Slot shot, with screen | ~18% |
| Power play one-timer | ~25% |
| Breakaway | ~35% |
A goalie facing 30 point shots has a dramatically easier night than one facing 30 slot shots — even if their final SV% looks identical.
The Example That Makes It Clear
Goalie A: - 30 shots faced, 28 saves - SV%: .933 - But: 25 shots from distance, 5 high-danger chances
Goalie B: - 30 shots faced, 27 saves - SV%: .900 - But: 10 shots from distance, 20 high-danger chances
Who had the better performance? Goalie B, and it is not particularly close.
SV% says A is better. The reality says the opposite.
Join ProbWin for free
Access our AI predictions for NBA, NHL, Soccer & Tennis. Sign up in 30 seconds.
GSAx: The True Measure of Goaltending Value
The Definition
GSAx = Goals Saved Above Expected
GSAx measures how many goals a goalie saved above (or below) what an average NHL goalie would have saved facing the exact same shots.
GSAx = Expected Goals Against (xGA) - Actual Goals Allowed
How It Is Calculated
Every shot a goalie faces gets assigned a goal probability based on:
| Factor | Impact on Probability |
|---|---|
| Shot distance | Closer = more dangerous |
| Shot angle | Center of the net = more dangerous |
| Shot type (slap, wrist, backhand) | Slap shot = more dangerous |
| Game situation (5v5, PP, PK) | Power play = more dangerous |
| Rebound or primary shot | Rebound = significantly more dangerous |
| Traffic in front | Screen = more dangerous |
| Time since last event | Quick shot = more dangerous |
Sum all those probabilities and you get the Expected Goals Against (xGA) — the number of goals an average goalie would have surrendered on that exact sequence of shots.
xGA = Sum of goal probabilities across all shots faced
GSAx = xGA - Actual Goals Allowed
A positive GSAx means the goalie saved more goals than expected. A negative GSAx means the goalie allowed more goals than expected.
Concrete Example
Game: Maple Leafs @ Canadiens
The Canadiens goalie faces 35 shots:
| Shot Group | xG for These Shots | Avg Probability |
|---|---|---|
| 20 shots from distance | 0.60 xG | 3% each |
| 10 slot shots | 1.20 xG | 12% each |
| 5 high-danger chances | 0.90 xG | 18% each |
| Total | 2.70 xGA |
The goalie allows 2 goals.
GSAx = 2.70 - 2.00 = +0.70
He saved 0.70 more goals than an average goalie would have. Excellent performance.
How to Read GSAx
The Single-Game Scale
| GSAx (per game) | Interpretation |
|---|---|
| > +1.00 | Exceptional performance |
| +0.50 to +1.00 | Very good performance |
| +0.20 to +0.50 | Good performance |
| -0.20 to +0.20 | Average performance |
| -0.50 to -0.20 | Below average performance |
| < -0.50 | Poor performance |
Cumulative GSAx Over a Season
GSAx accumulates across a full season:
| Season GSAx | Interpretation |
|---|---|
| > +20 | Elite goalie, Vezina Trophy contender |
| +10 to +20 | Very good goalie |
| 0 to +10 | Above-average goalie |
| -10 to 0 | Below-average goalie |
| < -10 | Struggling goalie |
The Elite Tier
The best goalies in the NHL typically finish with a season GSAx between +15 and +30.
A goalie with a +25 GSAx has literally prevented 25 more goals than an average goalie would have allowed. Over 60 games, that works out to approximately 0.4 extra goals prevented per game — an enormous impact on wins, losses, and team performance.
GSAx vs. SV%: The Comparison That Changes Everything
When They Diverge
This is where things get interesting for bettors:
| Goalie | SV% | GSAx | What It Means |
|---|---|---|---|
| A | .925 | +5 | Excellent — confirmed by both metrics |
| B | .925 | -5 | SV% inflated by weak competition |
| C | .905 | +8 | SV% depressed by poor defensive support |
| D | .905 | -8 | Genuinely struggling — no excuse |
Goalie C is the most exploitable scenario. His .905 SV% looks mediocre — but a GSAx of +8 shows he is performing well above average despite a defense that hangs him out to dry.
The market prices him as average. The data says he is elite.
A Real-World Type of Scenario
Two goalies in the same season:
| Goalie | Team | SV% | GSAx | Context |
|---|---|---|---|---|
| Goalie X | Top defensive team | .921 | +2.3 | Seeing mostly easy shots |
| Goalie Y | Poor defensive team | .908 | +12.5 | Facing dangerous shots constantly |
The public sees: "Goalie X has the better SV%, he must be better."
The data says: "Goalie Y is saving 10 more goals than expected. He is the real elite performer."
Betting opportunity: Under bets when Goalie Y starts — even though his SV% looks pedestrian to the casual bettor. The market undervalues him because it anchors on save percentage.
Using GSAx to Bet NHL Games
Strategy #1: Identify Undervalued Goalies (The Value Bet)
Look for goalies with: - Average or below-average SV% - Strongly positive GSAx
SV% < .910 but GSAx > +5 = Undervalued goalie
→ Under potential when he starts
This is pure value betting applied to goaltender evaluation. The market misprices totals because it uses the wrong metric.
Strategy #2: Identify Overvalued Goalies
Look for goalies with: - High SV% - Weak or negative GSAx
SV% > .920 but GSAx < 0 = Lucky or heavily protected goalie
→ Regression is likely — fade the Under when he starts
These are the goalies who benefit from elite team defense or playing behind a low-shot-quality defensive structure. When that support cracks (rest, back-to-back, facing a high-danger offense), their results suffer.
Strategy #3: Compare Both Goalies for the Game
For a totals bet, you need to analyze both starters:
| Goalie A GSAx | Goalie B GSAx | Signal |
|---|---|---|
| Strongly positive | Strongly positive | Under |
| Strongly positive | Negative | Team B will score |
| Negative | Negative | Over |
| Mixed | Mixed | Dig deeper |
The two-goalie comparison is often more predictive than looking at either goalie in isolation.
Strategy #4: Track Recent Form
Season GSAx is useful, but GSAx over the last 10–15 games can reveal critical trends:
| Recent Trend | What It Signals |
|---|---|
| Recent GSAx > Season GSAx | Goalie in form — lean Under |
| Recent GSAx < Season GSAx | Goalie struggling — lean Over |
| Consistent throughout | Reliable — trust the season number |
A goalie on a hot streak (high recent GSAx) is far more reliable for an Under bet than one whose season number is propped up by a strong start from two months ago.
Join ProbWin for free
Access our AI predictions for NBA, NHL, Soccer & Tennis. Sign up in 30 seconds.
Complementary Metrics to Use Alongside GSAx
GSAx is most powerful when used in combination with supporting data points.
xGA (Expected Goals Against)
xGA measures the quality of shots the goalie faces — essentially, how hard his job is.
| xGA per Game | Interpretation |
|---|---|
| < 2.00 | Excellent defensive team — easy workload |
| 2.00–2.50 | Solid defensive support |
| 2.50–3.00 | Average defensive support |
| > 3.00 | Poor defensive team — heavy workload |
Why it matters: A GSAx of +5 with an xGA of 2.0 per game is less impressive than the same GSAx with an xGA of 3.0 per game. The second goalie is doing far more work under fire.
High-Danger Save Percentage (HDSV%)
HDSV% measures save percentage on high-danger chances only — the most dangerous shots in the game.
| HDSV% | Interpretation |
|---|---|
| > .850 | Excellent under pressure |
| .820–.850 | Good |
| .800–.820 | Average |
| < .800 | Weak in dangerous situations |
A goalie with elite HDSV% delivers when it counts most. This is the clutch factor in goaltending — easy to praise, hard to fake over a full season.
Quality Start Percentage (QS%)
QS% is the percentage of starts where a goalie posts SV% above .913.
| QS% | Interpretation |
|---|---|
| > 60% | Very reliable |
| 50–60% | Reliable |
| 40–50% | Variable |
| < 40% | Unpredictable |
A goalie with a strong GSAx but low QS% is volatile — he has brilliant nights and disasters, not the consistent above-average performer you want to bet on repeatedly.
The Confirmed Starter Factor
Why This Is Critical in Hockey
Unlike baseball, where the starting pitcher is typically confirmed days in advance, NHL teams often do not name their starting goalie until morning skate — or sometimes only at warmups.
| Timing | Reliability |
|---|---|
| 2 days before | Estimate based on rotation |
| Morning of the game | Probable but not confirmed |
| Warmups (~1 hour before puck drop) | Generally confirmed |
This timing gap is one of the reasons sharp NHL bettors stay patient and do not rush to bet totals early in the day.
Impact of a Goalie Change
When a starter changes unexpectedly:
| Scenario | Impact on Total |
|---|---|
| Elite starter → Backup | +0.5 to +1.0 goal |
| Backup → Elite starter | -0.5 to -1.0 goal |
| Tired goalie (back-to-back) | +0.3 to +0.5 goal |
A one-goal swing in a low-scoring sport is enormous. A 6.5 total might be a clear Under with the starting goalie — and might shift to a coin flip with the backup.
Best Sources for Starter Confirmation
| Source | Timing | Reliability |
|---|---|---|
| Daily Faceoff | Morning | High |
| Local beat reporters on X/Twitter | Variable | Very high |
| NHL.com | 1 hour before game | Confirmed |
Our AI NHL predictions wait for goalie confirmation before finalizing picks. This is not optional — it is baked into the process.
Full Example: GSAx Analysis in Practice
The Game
Golden Knights @ Avalanche — January 15, 2026
The Goalies
| Team | Goalie | Season SV% | Season GSAx | GSAx (10 games) | xGA/Game |
|---|---|---|---|---|---|
| VGK | Hill | .912 | +8.5 | +4.2 | 2.65 |
| COL | Georgiev | .898 | -6.2 | -3.1 | 2.45 |
The Analysis
Hill (Vegas): - SV% of .912 looks average at first glance - But GSAx of +8.5 = excellent performer - Facing high xGA (2.65) means he's seeing genuinely dangerous shots - Strong recent form (+4.2 over last 10 games) - Verdict: Significantly undervalued. Performing much better than his SV% suggests.
Georgiev (Colorado): - SV% of .898 = below average - GSAx of -6.2 confirms it — he's genuinely struggling - Despite a manageable xGA of 2.45 (decent defensive support) - Worsening recently (-3.1 over last 10 games) - Verdict: Truly struggling. No mitigating context to explain it away.
Estimating the Total
xGA for Hill (Colorado will face): ~2.7
GSAx adjustment for Hill (+): -0.3
Expected Colorado goals: ~2.4
xGA for Georgiev (Vegas will face): ~2.5
GSAx adjustment for Georgiev (-): +0.4
Expected Vegas goals: ~2.9
Estimated total: 2.4 + 2.9 = 5.3 goals
The Market Line
Total: O/U 6.5
Over: -105
Under: -115
Verdict
- Our estimate: 5.3 goals
- Market line: 6.5 goals
- Gap: 1.2 goals in favor of Under
Recommended bet: Under 6.5
The market appears to be overrating Colorado's offense or underrating Hill's true quality. Either way, our GSAx-driven model sees clear value on the Under.
Common Mistakes to Avoid
Mistake #1: Trusting SV% Alone
Wrong: "He has a .920 SV%, he must be good"
Right: ".920 SV% but GSAx -5 — his defense is protecting him"
Mistake #2: Ignoring Team Context
The same goalie posting GSAx +10 could be: - Excellent behind a good defensive team - Spectacular behind a terrible defensive team
Always check xGA per game to understand how difficult his workload actually is.
Mistake #3: Small Sample Sizes
Early in the season, GSAx is highly volatile:
| Games Played | GSAx Reliability |
|---|---|
| < 10 | Low — do not draw strong conclusions |
| 10–20 | Moderate |
| 20–40 | Good |
| > 40 | High |
Wait until a goalie has 15–20 starts before acting strongly on GSAx data. Early-season outliers — positive and negative — tend to regress.
Mistake #4: Forgetting the Back-to-Back
A goalie playing the second night of a back-to-back is often the backup, not the starter. Verify before betting. This is one of the most common and costly errors in NHL totals betting.
Mistake #5: Ignoring the Other Goalie
Your goalie can be elite — but if the opponent also has an elite goalie, the Under is not guaranteed. You might get a 2-1 game or a 1-0 game. Both teams' goaltending situations need to be analyzed together.
Where to Find GSAx Data
Free Sources
| Source | URL | Data Available |
|---|---|---|
| MoneyPuck | moneypuck.com | GSAx, xGA, HDSV% |
| Natural Stat Trick | naturalstattrick.com | GSAx, advanced stats |
| Hockey Reference | hockey-reference.com | Traditional stats + some advanced |
| Evolving Hockey | evolving-hockey.com | GSAx, goalie WAR |
How to Navigate MoneyPuck
1. Go to moneypuck.com
2. Click "Goalies"
3. Sort by "GSAx" or "GSAx/60"
4. Also check "xGA", "HDSV%", and "QS%"
This takes about two minutes before a game. For the edge it provides, that is two of the most valuable minutes in your betting preparation.
How ProbWin Uses GSAx
Our AI NHL predictions model treats GSAx as the primary goalie input for every analysis. Here is the full sequence:
1. Season GSAx for both goalies
2. Recent GSAx trend (last 10–15 games)
3. xGA per game (contextualizes the GSAx)
4. HDSV% for high-pressure situations
5. Confirmed starter (required before any pick is finalized)
6. Head-to-head history vs. this opponent
We never publish an NHL pick without confirmed starters. That is a hard rule. The ProbWin guide explains the full methodology if you want to understand how all the pieces fit together.
Check our NHL picks page to see this analysis applied daily, and results to track how the model performs over time.
For the possession-based context that complements goalie analysis, Corsi and Fenwick explain how shot generation reveals true team quality. A well-managed bankroll ensures you can stay in action long enough to capture the edge that value betting provides.
Summary: 7 Things to Remember About GSAx
| # | Key Takeaway |
|---|---|
| 1 | GSAx = Goals Saved Above Expected |
| 2 | GSAx is more accurate than SV% for evaluating goalies |
| 3 | High GSAx + average SV% = undervalued goalie (betting opportunity) |
| 4 | High SV% + negative GSAx = overvalued or heavily protected goalie |
| 5 | Always check xGA to understand the goalie's workload context |
| 6 | Confirm the starting goalie before placing any NHL bet |
| 7 | Require 15–20 starts before trusting GSAx strongly |
Next Steps
You now have a complete framework for evaluating NHL goalies through the GSAx lens. The next step is understanding the defensive side of the equation for entire teams — not just individual goalies.
Dive deeper into Corsi and Fenwick, the possession metrics that reveal which NHL teams are genuinely controlling play versus riding luck. Combined with GSAx, you will have a significantly sharper view of NHL totals than the average bettor.
And when you're ready to see this analysis working in real time, our NHL picks page puts it all together — every game, every night.
