NHL Totals: Expected Goals, Special Teams, and the Empty Net

College basketball totals span forty-five points. NFL totals span twenty. Hockey totals operate inside a single goal, which makes every half-goal worth more here than anywhere else in American sports.

Yesterday's post covered the sides. This one covers the market where hockey's compressed scoring produces the strangest structure in betting.

Nearly every NHL total is 5.5, 6, or 6.5. That is the entire range. And because 6 is the modal outcome in a sport averaging roughly six goals a game, the difference between those numbers is enormous rather than marginal.

01The Narrowest Range in Sports

Set the comparison properly, because it explains everything that follows.

The final column understates the difference, because hockey's range is not continuous. There are effectively three numbers, and one of them appears far more often than the others.

So when a book's view shifts, it usually cannot move the number without a large jump in win probability. It moves the price instead. You will see 6 at -140 for the under and +120 for the over rather than a move to 6.5.

This is the same phenomenon as the fixed puck line from yesterday, appearing in the totals market: read the price, not the number.

02Why Line Shopping Matters More Here

Post 77 argued that shopping is the highest-return habit in betting. Hockey totals are where that argument is strongest.

Two situations, both common:

Different numbers. One book has 6, another has 5.5. If you like the under, that half goal is a large probability difference, not a rounding error. Taking under 6 instead of under 5.5 changes your outcome on the single most common total in the sport.

Same number, different price. Over 6 at +105 versus over 6 at -115 is twenty cents on an identical bet. Across a season of hockey volume, capturing the better side of that is worth more than most handicapping improvements available to you.

The practical rule: never take a hockey total without checking at least three books. The dispersion is wider than in football or basketball because the number cannot absorb disagreement, so the disagreement shows up in prices instead.

03Building a Total From Scratch

The framework, structurally similar to the possession approach from Post 65 but with different components.

Total goals = shot volume × shot quality × (1 − goaltender performance)

Broken into working parts:

1. Shot generation. How many attempts each team produces. Roughly 30 shots on goal per team per game is typical, with meaningful variation by style.
2. Shot quality. Where those attempts come from. Ten shots from the slot are worth far more than twenty from the perimeter.
3. Goaltending. The subtraction, and the largest single term, per the goaltending post.
4. Special teams. A separate scoring channel with a much higher conversion rate.
5. Empty net. A structural addition covered in Section 07.

Note what is absent compared to basketball: there is no meaningful pace negotiation. Hockey teams do not control possession length the way a deliberate basketball team can drain a shot clock. Shot volume varies, but not with the 61-to-76 spread that made tempo the master variable in college basketball.

04Expected Goals and Shot Quality

Raw shot counts mislead in hockey the same way raw points per game mislead in basketball, and for the same reason: they treat unequal events as equal.

Expected goals models assign each attempt a scoring probability based on distance, angle, shot type, whether it followed a pass across the slot, whether it came off a rush, and the game state. Summing them gives the goals an average team would score from that specific set of chances.

Why this matters for totals:

• A team generating 35 low-danger shots is a lower-scoring proposition than one generating 26 with heavy slot presence.
• Shot suppression is a real skill. Some teams concede volume but almost nothing dangerous.
• Shooting percentage regresses hard. A team converting at an unusual rate over ten games will not sustain it, and the total priced off that run is a fade candidate.
• Expected goals stabilize faster than actual goals. Which makes them more useful early in the season, when everything else is noise.

Public expected goals data is widely available. As with the goaltending metrics, you do not need to build anything. You need to use the adjusted number instead of the raw one.

05The Goalie Matchup Dominates

Everything from the goaltending post applies to totals, and arguably applies more.

Two elite goaltenders facing each other is the clearest under signal in hockey. Two struggling ones is the clearest over signal. Neither is subtle, and both are priced, but the magnitude is frequently underestimated by bettors focused on offensive matchups.

The practical sequence:

1. Wait for confirmation. A total posted before starters are announced is a total priced across possibilities.
2. Use multi-year adjusted performance, not this month's save percentage.
3. Check the defensive context. A good goalie behind a team that concedes chances still faces a difficult night.
4. Weight the goalie term heavily. It moves the total more than either offense does.

06Special Teams: The Direct Mechanism

Power play and penalty kill performance is the most directly actionable input in hockey totals, because the causal chain is short and visible.

A team on the power play converts at roughly one opportunity in five across the league, against a far lower rate at even strength. Special teams therefore account for a substantial share of all scoring, and the gap between the best and worst units is wide.

The two numbers to know for any team:

• Power play percentage. Goals per opportunity.
• Penalty kill percentage. Opponent opportunities defended.

The matchups that matter:

Elite power play against a poor penalty kill is the cleanest over lean available in hockey. The mechanism is direct: penalties will be called, and one side converts them at a much higher rate than the other defends them.

Two strong penalty kills removes a scoring channel from both sides and is a genuine under signal.

The variable underneath both is penalty rate. A game between two disciplined teams generates fewer opportunities for anyone. A game between two teams that take penalties frequently opens the channel wider for both. Team discipline is a stable property worth tracking.

07The Empty Net Effect on Overs

The same structural feature from yesterday, viewed from the totals side.

When a trailing team pulls its goaltender in the final minutes, two things become more likely at once: the trailing team scores with the extra attacker, and the leading team scores into the empty net. Both outcomes add a goal to the total.

This creates a systematic upward pressure on hockey totals that has no equivalent in other sports. Football's endgame kills clock. Basketball's endgame adds points through fouling, as covered in Post 65, but hockey's adds them through an entirely different mechanism that operates in a large share of close games.

The practical translations:

• A close game late is over-friendly. If you hold an under and the game is within one goal in the final two minutes, you are not safe.
• A blowout late is under-friendly. A three-goal deficit does not produce an empty net sequence, and both teams coast.
• Your side projection and your total projection are correlated. This is the same relationship established for basketball in Post 65, arriving through a different mechanism.
• Live unders in close games are worse than they appear. Covered further in tomorrow's post.

08Score Effects

A well-documented hockey phenomenon that changes how you read team statistics.

Teams that are trailing generate more shot attempts. They push, take more risks, and spend more time in the offensive zone. Teams that are leading sit back, defend, and concede volume while protecting the dangerous areas.

This means raw shot statistics are contaminated by game state. A team that trails frequently will look better at generating shots than it is. A team that leads frequently will look worse.

Adjusted metrics correct for this, and it is worth using the score-adjusted versions where available. The practical point for a totals bettor is that a team's shot numbers tell you partly about their ability and partly about how often they have been losing.

09Period Totals and Team Totals

Derivative markets worth knowing, with the usual caveat about wider pricing.

Period totals. Usually 1.5 goals for a single period. Worth noting that scoring is not evenly distributed across periods: third periods tend to produce more, partly because of the empty net effect and partly because of score effects pushing trailing teams.

Team totals. One side's scoring in isolation. Useful when your read is directional, such as an elite power play against a weak kill, rather than about the game as a whole.

Regulation-only totals. Exclude overtime, which removes a small amount of scoring and the shootout goal that most books credit for settlement.

As with all derivatives, the tradeoff is pricing and limits. These are instruments for expressing a specific view, not defaults.

10Your Hockey Totals Workflow

1. Wait for confirmed goaltenders. Non-negotiable, and the largest term in the projection.
2. Pull expected goals for both teams, for and against, score-adjusted where available.
3. Apply the goaltending adjustment using multi-year performance rather than recent save percentage.
4. Check special teams. Both power plays against both penalty kills, and both teams' penalty rates.
5. Consider game script. Do you expect a close game, which is over-friendly late, or a mismatch, which is not?
6. Produce a number. Then compare it to the market.
7. Shop three books minimum. Both the number and the price.
8. Log it, including the closing number, per Post 85.

11Common Hockey Totals Mistakes

• Treating goaltending as an adjustment. It is the largest term in the projection.
• Betting before starters are confirmed. The same error as on the sides, and it costs more here.
• Using raw shot counts. Contaminated by score effects and blind to shot quality.
• Projecting off recent shooting percentage. It regresses hard, exactly like the three-point shooting problem in Post 65.
• Ignoring special teams. The most direct and most underused mechanism in the sport.
• Forgetting the empty net. Structural upward pressure that operates in most close games.
• Not shopping. A half goal here is worth what a touchdown is worth in football.
• Betting marginal disagreements. A tenth of a goal is not an edge in a market this compressed.

12The Bigger Picture

Hockey totals reward a specific discipline that most bettors find unnatural: accepting that the market is usually right and betting rarely.

The compressed range means there is simply less room for disagreement. In college basketball you can plausibly be five points away from the market and have a defensible reason. In hockey, being a half goal away requires an identifiable structural cause, and most nights there is not one.

What that produces is a market where the correct number of bets is small and the ones you do make are well-founded. A goalie matchup the market underweighted. A special teams mismatch. A total priced off an offensive reputation that the shot quality data does not support.

Those spots exist. They are just less frequent than the sixty-plus games on a college basketball Tuesday might have trained you to expect, and treating hockey totals as a volume market is the fastest way to give back whatever the goalie window earns you.