RKL Stats Glossary

GEM (Geometric Mean Rank)

GEM is an alternative way of averaging a player's rankings to reduce the impact of outliers. GEM is calculated by taking the nth root of the product of all the player's gameday ranks, where n is the total number of values.

GEM = (x1 * x2 * ... * xn) ^ (1/n)

PAM (Points Above Median)

PAM represents the cumulative number of points that a team has scored in excess of the leaguewide median on their gamedays. PAM is officially used as the tiebreaker in league standings.

PAM = (total points scored) - (total gameday medians)

apPAM (Average Percentage Points Above Median)

Closely related to PAM, apPAM represents the *average* amount, in percentage terms, by which a team exceeds the leaguewide median score on a given day. Because different days have different scoring opportunities, teams with identical PAM values may not have the same apPAM.

pPAM1, pPAM2, ... , pPAMn: percentage points above the median score in games 1, 2, ... , n apPAM = (pPAM1, pPAM2, ... , pPAMn) / n

REL (Relative Performance Score)

REL measures a player's raw point generation relative to the league's mean or median on days they played. It's a way to normalize scores across games that might have different scoring opportunities. An REL above 1.000 indicates above-average performance, while a score below 1.000 is below average.

REL Median = (Player's Raw Scores / League Median Scores) REL Mean = (Player's Raw Scores / League Mean Scores)

tREL (Team REL)

tREL is the weighted average REL score of all the players on a team's roster. Players who have started more games are given more weight in the formula (ex: if Player A has started six games and Player B has started two games, Player A's REL counts 3x as much as Player B's when calculating tREL).

REL1, REL2, ... , REL10: the REL scores for players 1, 2, ... , 10 GP1, GP2, ... , GP10: the number of games played for players 1, 2, ... , 10 tREL = [(REL1 * GP1) + (REL2 * GP2) + ... + (REL10 * GP10)] / (GP1 + GP2 + ... + GP10)

WAR (Wins Above Replacement)

WAR attempts to quantify a player's total contribution to their team in a single number. It compares a player's performance to that of a "replacement-level" player—a hypothetical player who is readily available and typically performs at a league-minimum level. The resulting value represents the number of additional wins a team would get by having that player in their lineup instead of a replacement-value one. WAR is a cumulative stat, increasing with every above-replacement level performance.

The average margin of victory in a S7 RKL game is equal to 92% of the leaguewide median player score on a given day. Therefore, a player is considered to add 1.0 WAR if they produce (.92 * median) more karma than a replacement level player. For the purposes of WAR calculations, a "replacement level" performance is defined as 80% of the leaguewide median player score. This is approximately equal to the 180th-best REL in the league (30 teams * 6 starters).

WAR = (Player Scores - Replacement-Level Scores) / (Avg MoV)