Replacement Level (RL)
The performance baseline representing a freely available player -- typically a waiver claim or AHL call-up. All "above replacement" metrics (GAR, WAR) measure value relative to this baseline.
Replacement level is the conceptual foundation of all "above replacement" metrics in sports analytics. It answers the question: what is the minimum performance level a team can expect from a player they could acquire for essentially nothing? In hockey, this means a player available on waivers, a minor-league call-up, or a league-minimum free agent.
Setting the replacement level is both critically important and inherently subjective. Set it too high, and average players appear to have negative value. Set it too low, and even below-average players look valuable. Hockey Alchemy calibrates it from the replacement-caliber population itself rather than from a chosen percentile of regulars, and the result puts roughly a quarter to a third of qualified NHL regulars below the line in a given season. That is deliberate and it is in line with the other public models -- Evolving Hockey and HockeyStats land in the same range. A much lower share below zero would be a sign the baseline had been set too generously.
The replacement-level concept is borrowed from baseball's WAR framework and serves the same purpose: it provides a meaningful zero point. Saying a player is "worth 10 goals above average" is less useful than "worth 15 goals above replacement" because teams do not replace players with league-average talent -- they replace them with the best available option, which is typically far below average.
Frequently Asked Questions
Who qualifies as a replacement-level player?
A replacement-level player is the caliber of talent a team could acquire for minimal cost: waiver pickups, AHL call-ups, or league-minimum free agents. These are typically the 13th forward, 7th defenseman, or third-string goaltender -- players on the fringe of the NHL roster.
Why use replacement level instead of league average?
Because teams do not replace players with league-average talent. When a player gets injured or traded, the replacement is usually a much cheaper, less skilled option. Measuring against replacement level provides a more realistic assessment of what a team actually loses when a player is unavailable.
Does replacement level change from season to season?
Yes. Hockey Alchemy uses adaptive replacement levels calculated from each season's data, accounting for changes in the league's scoring environment and talent distribution. The xG calibration is also normalized per season to maintain consistency.
Can a player be below replacement level?
Yes, and it is far from rare. A player with negative GAR performed below replacement level -- theoretically, the team would have been better served by a freely available alternative. Roughly a quarter to a third of qualified NHL regulars finish a season below zero. Negative GAR is a normal feature of the scale, not an outlier or a small-sample artifact.