PDO Regression (Luck Factor)

Overview

This pillar analyzes a team's PDO, a key hockey statistic combining shooting and save percentage, to identify teams that are overperforming or underperforming due to luck. It provides a contrarian signal, highlighting teams likely to regress to the mean.

What It Does

It calculates a team's PDO by adding its shooting percentage and save percentage. Since the league-wide average always trends towards 100, teams with significantly higher or lower numbers are flagged. This analysis helps determine if a team's current winning or losing streak is skill-based or just a run of good or bad fortune.

Why It Matters

PDO is a powerful predictor of future performance because luck is temporary. Teams with a high PDO are often overvalued in betting markets, while those with a low PDO are undervalued, creating profitable opportunities to bet on their eventual regression.

How It Works

First, we gather 5-on-5 shooting percentage and save percentage data for each NHL team. We then calculate the PDO for each team using the formula S% + Sv%. Finally, we compare this value against the league average of 100 to identify teams with unsustainably high (over 102) or low (under 98) PDOs, who become targets for regression-based predictions.

Methodology

The core calculation is PDO = (Team Shooting Percentage * 100) + (Team Save Percentage * 1000). The analysis primarily focuses on 5-on-5 even-strength situations, as this removes the noise from special teams play. We analyze both season-long PDO and rolling 10-game averages to identify long-term trends and short-term streaks.

Edge & Advantage

This pillar provides an edge by systematically identifying teams the market has mispriced due to short-term luck, allowing for bets against hot streaks and on cold streaks breaking.

Key Indicators

• Team PDO (S% + Sv%)

  high

  The core metric combining shooting and save percentage. A value far from 100 indicates luck.

• 5v5 PDO Trends

  high

  PDO calculated during even-strength play, which is the best indicator of a team's true baseline.

• Rolling 10-Game PDO

  medium

  Tracks recent performance to identify teams on hot or cold streaks driven by luck.

Data Sources

• Natural Stat Trick

  Provides comprehensive, free advanced NHL statistics including PDO at various game states.

• Evolving-Hockey

  Offers advanced models and data tables for NHL analytics, including historical PDO data.

• Hockey-Reference

  A reliable source for team-level shooting and save percentages needed for basic PDO calculation.

Example Questions This Pillar Answers

• → Will the Vancouver Canucks win fewer than 5 of their next 10 games?
• → Will the Columbus Blue Jackets have a positive goal differential next month?
• → Which team will have a better record in the second half of the season: Team A (high PDO) or Team B (low PDO)?

Tags