NHL Equivalency (NHLe)

League Translation Factors

NHL Equivalency (NHLe) is a method for translating a player's performance in one league to what they might produce in the NHL. By analyzing how players perform before and after NHL transitions, we can calculate league-specific translation factors.

import pandas as pd
import numpy as np
from sklearn.linear_model import LinearRegression

# Load player data across leagues and their subsequent NHL performance
player_transitions = pd.read_csv('league_to_nhl_transitions.csv')

# Calculate NHLe translation factors for each league
leagues = ['AHL', 'KHL', 'SHL', 'Liiga', 'OHL', 'WHL', 'QMJHL', 'NCAA']

nhle_factors = {}

for league in leagues:
    league_players = player_transitions[
        player_transitions['previous_league'] == league
    ]

    # Linear regression: NHL PPG = factor * League PPG
    X = league_players[['league_ppg']].values
    y = league_players['nhl_ppg'].values

    model = LinearRegression()
    model.fit(X, y)

    nhle_factors[league] = model.coef_[0]

    # Calculate R-squared for quality of translation
    r_squared = model.score(X, y)

    print(f"{league}: {nhle_factors[league]:.3f} (R² = {r_squared:.3f})")

# Create comprehensive NHLe conversion table
print("\n=== NHL Equivalency Factors ===")
nhle_df = pd.DataFrame.from_dict(nhle_factors, orient='index',
                                  columns=['NHLe_Factor'])
nhle_df = nhle_df.sort_values('NHLe_Factor', ascending=False)
print(nhle_df)

# Example: Convert player stats to NHLe
def calculate_nhle(points, games, league):
    """Calculate NHL-equivalent points"""
    ppg = points / games
    factor = nhle_factors.get(league, 0.5)
    nhle_ppg = ppg * factor
    return nhle_ppg * 82  # Project to 82-game season

# Test with sample prospects
prospects = pd.DataFrame({
    'name': ['Player A', 'Player B', 'Player C', 'Player D'],
    'league': ['AHL', 'KHL', 'OHL', 'NCAA'],
    'points': [65, 48, 92, 54],
    'games': [68, 60, 62, 38]
})

prospects['NHLe_Points'] = prospects.apply(
    lambda x: calculate_nhle(x['points'], x['games'], x['league']),
    axis=1
)

print("\n=== Prospect NHLe Projections ===")
print(prospects)

# Age-adjusted NHLe for junior players
def age_adjusted_nhle(points, games, league, age):
    """Adjust NHLe based on player age"""
    base_nhle = calculate_nhle(points, games, league)

    # Junior leagues: younger = higher projection
    if league in ['OHL', 'WHL', 'QMJHL']:
        age_adjustment = 1 + (19 - age) * 0.08
        return base_nhle * age_adjustment

    return base_nhle

# Compare raw vs age-adjusted projections
junior_prospects = prospects[prospects['league'] == 'OHL'].copy()
junior_prospects['age'] = [18, 19, 20, 17][:len(junior_prospects)]
junior_prospects['Age_Adj_NHLe'] = junior_prospects.apply(
    lambda x: age_adjusted_nhle(x['points'], x['games'],
                                x['league'], x['age']),
    axis=1
)

print("\n=== Age-Adjusted NHLe (Junior) ===")
print(junior_prospects[['name', 'age', 'NHLe_Points', 'Age_Adj_NHLe']])

library(tidyverse)
library(broom)

# Load player transition data
player_transitions <- read_csv("league_to_nhl_transitions.csv")

# Calculate NHLe factors for each league
leagues <- c("AHL", "KHL", "SHL", "Liiga", "OHL", "WHL", "QMJHL", "NCAA")

nhle_results <- leagues %>%
  map_dfr(function(league) {
    league_data <- player_transitions %>%
      filter(previous_league == league)

    # Linear regression model
    model <- lm(nhl_ppg ~ league_ppg - 1, data = league_data)

    # Extract coefficient and R-squared
    coef_value <- coef(model)[1]
    r_squared <- summary(model)$r.squared

    tibble(
      league = league,
      nhle_factor = coef_value,
      r_squared = r_squared,
      n_players = nrow(league_data)
    )
  })

cat("=== NHL Equivalency Factors ===\n")
print(nhle_results %>% arrange(desc(nhle_factor)))

# Function to calculate NHLe points
calculate_nhle <- function(points, games, league, factors_df) {
  ppg <- points / games
  factor <- factors_df %>%
    filter(league == !!league) %>%
    pull(nhle_factor)

  if (length(factor) == 0) factor <- 0.5  # default

  nhle_ppg <- ppg * factor
  return(nhle_ppg * 82)  # 82-game projection
}

# Example prospect projections
prospects <- tibble(
  name = c("Player A", "Player B", "Player C", "Player D"),
  league = c("AHL", "KHL", "OHL", "NCAA"),
  points = c(65, 48, 92, 54),
  games = c(68, 60, 62, 38)
)

prospects <- prospects %>%
  rowwise() %>%
  mutate(
    nhle_points = calculate_nhle(points, games, league, nhle_results)
  ) %>%
  ungroup()

cat("\n=== Prospect NHLe Projections ===\n")
print(prospects)

# Visualize NHLe factors
ggplot(nhle_results, aes(x = reorder(league, nhle_factor),
                         y = nhle_factor, fill = r_squared)) +
  geom_col() +
  geom_text(aes(label = sprintf("%.3f", nhle_factor)),
            hjust = -0.2, size = 3) +
  coord_flip() +
  scale_fill_gradient(low = "lightblue", high = "darkblue",
                      name = "R²") +
  labs(title = "NHL Equivalency Factors by League",
       subtitle = "Conversion factors for projecting NHL performance",
       x = "League", y = "NHLe Factor") +
  theme_minimal()

# Quality of translation analysis
cat("\n=== Translation Quality ===\n")
print(nhle_results %>%
  select(league, nhle_factor, r_squared, n_players) %>%
  arrange(desc(r_squared)))

Using NHLe Effectively

NHLe provides a useful baseline for comparing players across leagues, but it's not a perfect predictor. Factors like age, playing style, physical readiness, and team context all affect NHL translation beyond raw statistics.