FantasyHockeyLab

    NHL Fantasy Projection Methodology

    Our Approach

    FantasyHockeyLab uses a transparent, straightforward methodology for player projections. We believe in showing our work and keeping projections interpretable rather than using complex black-box models.

    NHL Skater Projections

    Historical Performance Analysis

    We calculate per-60 minute rates for key stats (Goals, Assists, Shots, Hits, Blocks) using the last 3 seasons with a weighted average:

    • Most recent season: 5x weight
    • 2 seasons ago: 3x weight
    • 3 seasons ago: 1x weight

    2. Regression to Mean

    Small sample sizes are regressed toward the league average using the formula:

    Regression Weight = 2000 TOI / (2000 + Player's Total TOI)

    Players with more ice time get less regression. This prevents overestimating breakout performances based on limited data.

    Technical Details: Per-60 Calculations

    All rate statistics are calculated as per-60-minutes of ice time:

    Goals per 60 = (Total Goals / Total Minutes Played) × 60

    This normalization allows fair comparison between players with different ice time allocations. Elite players who play 22+ minutes per game are compared on an equal footing with depth players who play 12-15 minutes.

    3. Playing Time Forecast

    Time-on-ice per game is projected using an exponential moving average (EMA) of the last 20 games. This captures recent trends in ice time allocation by coaches.

    4. Final Projections

    We convert per-60-minute rates to per-game estimates using projected TOI:

    Goals per Game = (Goals per 60) × (TOI per Game / 60)

    For example, a player with 1.5 goals per 60 minutes who plays 20 minutes per game would be projected for 0.5 goals per game (1.5 × 20/60 = 0.5).

    5. Games Played & Availability

    Each skater gets an individual games-played forecast. We measure the share of games they dressed for in each completed season, weight recent seasons more, and regress toward a league durability baseline. Durable players project near a full season, typical regulars land in the mid-70s, and players with a history of missed time project lower. The Per-82 view always uses a full 82 games.

    6. Special Teams, GWG & Plus/Minus

    Power-play points, short-handed points and game-winning goals are projected from each player's own historical share of their scoring, regressed toward league norms when the sample is small. Plus/minus is projected from past results, cut back heavily because it is one of the least repeatable stats, then scaled to projected games.

    These per-game rates are then scaled to three different contexts:

    • Per-82: Rate-based projection over a full 82-game season
    • Rest of Season (ROS): Expected totals for remaining games, adjusted for availability
    • Season Total: Current season stats + ROS projection = full season total

    Goalie Projections

    Workload (Games Started)

    Starts are projected from each goalie's share of their team's starts over recent seasons, regressed toward an average share when history is limited. Proven workhorses keep heavy usage (roughly 55–61 starts), 1A starters land in the high 40s to low 50s, and backups project fewer.

    Decisions: Wins, Losses & OT Losses

    Every projected start receives a decision, so wins + losses + OT losses always equal games started. The split follows the goalie's history adjusted for team quality — stronger teams produce more wins.

    Save Percentage, Shots & GAA

    Save percentage is regressed to the league mean weighted by shots faced. Shots against reflect the team's defensive quality, and GAA is derived from projected save percentage and shots per game. Shutouts follow historical rates adjusted for projected save percentage.

    League Leaders: Early-Season Qualification

    Goalie GAA and SV% leaders require a minimum number of games played. Early in the season the minimum scales with the league (starting at 1 GP) and rises automatically to the standard 9 GP, matching how NHL.com handles early leaderboards.

    Data Sources

    All statistics are sourced from:

    • NHL public Stats API for player and team data
    • NHL Media CDN for player headshots
    • Official NHL.com and Team PR feeds for news

    Data is refreshed daily during the season. All data points include a "last updated" timestamp for transparency.

    Limitations & Disclaimers

    Availability is modeled from injury history, but projections do not react in real time to:

    • Current injuries or day-to-day lineup changes
    • Trade impacts or coaching changes
    • Individual player development or decline
    • Team system changes

    Use projections as one tool among many for fantasy decision-making. Always combine statistical analysis with news, expert opinions, and your own hockey knowledge.