Flames PDO: Analyzing Luck & Regression

Flames PDO: Analyzing Luck & Regression

Executive Summary

This case study examines the role of PDO—a foundational advanced statistic measuring a combination of shooting and save percentage—in evaluating the performance trajectory of the Calgary Flames during the 2023-24 NHL season. By analyzing the Flames' PDO fluctuations, this report dissects the interplay between sustainable process, statistical luck, and the inevitable regression that shapes outcomes in a professional hockey campaign. The analysis reveals how an understanding of this metric provides critical context for the club's results, informs evaluations of individual performances from players like Jonathan Huberdeau and Jacob Markström, and offers a data-driven lens through which to view the strategic decisions of GM Craig Conroy and head coach Ryan Huska. Ultimately, this study demonstrates that while PDO is not a predictor, it is an essential diagnostic tool for separating signal from noise in the complex narrative of a season.

Background / Challenge

The Calgary Flames entered the 2023-24 NHL season amidst a period of significant transition. Following a campaign that fell short of playoff expectations, the organization embarked on a recalibration under new leadership. GM Craig Conroy and head coach Ryan Huska were tasked with implementing a system that emphasized pace, possession, and defensive structure, aiming to return the club to contention in a highly competitive Pacific Division and Western Conference.

The primary analytical challenge emerged as the season unfolded. The Flames experienced stretches of both pronounced success and frustrating adversity. Pundits and the C of Red alike often attributed these swings to "luck"—hot goaltending, cold shooting streaks, or an inability to "get the bounces." Such narratives, while emotionally resonant, lack empirical rigor. The hockey operations staff required a more objective framework to answer pressing questions: Was the team's process better than its results indicated during losing skids? Were early-season winning streaks built on a stable foundation or due for a correction? Accurately assessing the team's true performance level, separate from the volatile noise of random variance, was crucial for making informed decisions on lineup configurations, tactical adjustments, and long-term planning.

Approach / Strategy

To navigate this challenge, we employed a season-long tracking and analysis of the Flames' PDO. PDO is the sum of a team's (or player's) on-ice shooting percentage (Sh%) and on-ice save percentage (Sv%) at 5-on-5, typically expressed as a whole number (e.g., 100.0). The core premise is that, over a large enough sample, these percentages will regress toward the league mean, which historically stabilizes near 100.0. A PDO significantly above 100 suggests a player or team may be benefiting from positive variance (puck luck), while a figure below 100 may indicate unsustainable misfortune.

Our strategy involved a multi-layered analysis:

1. Team-Level Tracking: Monitoring the Flames' overall 5-on-5 PDO on a rolling 10-game basis to identify sustained trends and inflection points.


2. Contextual Segmentation: Isolating PDO data during specific scenarios: early-season adaptation, post-trade deadline roster changes, and key segments of the Battle of Alberta.


3. Individual Player Correlation: Examining the PDO of key drivers, such as top-line forwards Jonathan Huberdeau and Nazem Kadri, rookie Connor Zary, and starting goaltender Jacob Markström, to assess their influence on team results and their own personal sustainability.


4. Process vs. Outcome Comparison: Contrasting PDO data with underlying process metrics, such as Corsi For% (shot attempt share) and Expected Goals For% (xGF%), available in our broader Flames Advanced Stats Explained library. This determined if results were aligned with performance.

This approach allowed us to move beyond simplistic win-loss records and identify periods where the Flames' fortunes were disproportionately impacted by variance, providing a clearer picture of the efficacy of the systems implemented by Huska and his staff.

Implementation Details

Data was sourced from publicly available, reputable advanced statistics databases, focusing exclusively on 5-on-5 play to eliminate the distorting effects of special teams. Analysis was conducted in phases corresponding to the natural breaks and pivotal moments of the Flames' schedule.

Phase 1: Baseline Establishment (First 20 Games): The initial quarter of the season served to establish a performance baseline. During this period, the Flames' system was in its nascent stage. We tracked how PDO interacted with the team's possession metrics, noting if early wins or losses were "deserved" from a process standpoint.
Phase 2: Mid-Season Regression & Adjustment (Games 21-60): This extended phase captured the heart of the schedule where regression most commonly manifests. We paid particular attention to segments where the PDO deviated sharply from 100.0 for 10-15 game stretches. For instance, a prolonged PDO surge above 102.0 would be scrutinized against shot quality data to see if it was driven by elite finishing/saving or mere randomness.
Phase 3: Roster Stability & Final Push (Post-Deadline): Following the trade deadline, with the roster largely settled, we analyzed whether the team's PDO stabilized, indicating a "true" performance level for the group assembled by Conroy. Individual focus intensified on players like Huberdeau, whose personal on-ice PDO is a critical component of his offensive output evaluation, and Markström, whose save percentage is half of the equation.

This phased implementation, integrated with our ongoing Flames Player Possession Metrics tracking, created a dynamic model for assessing performance sustainability throughout the campaign.

Results

The Flames' 2023-24 PDO journey provided a clear numerical narrative of a team grappling with and ultimately being leveled by regression.

Early-Season Unsustainability: Through the first 25 games, the Flames posted a collective 5-on-5 PDO of 101.8. This was buoyed by an on-ice shooting percentage of 9.1% and a save percentage of .927. This period included several one-goal victories. However, underlying metrics showed only a modest 50.2% Corsi For% and a 48.7% xGF%. The results were outperforming the process, signaling an impending correction.
The Regression Window: From games 26 through 50, that correction arrived decisively. The Flames' PDO plummeted to 98.4, driven by a shooting percentage drop to 7.8% and a save percentage decline to .906. Despite the process metrics improving slightly (Corsi For% rose to 51.0%), the results suffered. This period starkly illustrated how a PDO swing of over 3 points directly translated to the loss of standings points.
Individual Case Studies:
Jonathan Huberdeau: His on-ice PDO through the first half was a paltry 97.1, among the lowest on the team for regular forwards. This quantified the "bad luck" narrative surrounding his line, as they generated chances but saw a low percentage go in. In the latter half, this regressed positively toward 100.2, correlating with an uptick in his point production.
Connor Zary: The rookie's early impact was reflected in a stellar on-ice PDO of 103.5 through his first 30 games, indicating he was on the ice for highly efficient scoring. As the season progressed, this regressed to a more sustainable 100.8, a natural progression for a first-year player.
Jacob Markström: His personal Sv% at 5-on-5 mirrored the team's PDO arc. He posted a .934 Sv% during the team's high-PDO start, which regressed to .912 during the mid-season downturn before stabilizing near his career average later on.
Final Stabilization: Over the final 30 games, the Flames' metrics converged. The PDO normalized to 99.9, almost exactly at the league mean. Crucially, their process metrics held firm at approximately 51.5% Corsi For% and 50.5% xGF%. This indicated that by season's end, the team's record was a far more accurate reflection of their actual performance level—a mid-tier team in the West fighting for a playoff berth.

Key Takeaways

1. PDO Validates the "Process Over Results" Mantra: The most significant takeaway is that the Flames' underlying process, particularly in the second half of the season, was more stable and promising than the volatile results suggested. The extreme highs and lows in win-loss records were significantly amplified by unsustainable percentages.


2. Regression is a Powerful, Inevitable Force: The mid-season PDO correction of over 3 points was a textbook example of regression to the mean. It serves as a caution against overreacting to both hot streaks and cold snaps, a vital lesson for management and fans alike.


3. Individual Performance Context: PDO provides essential context for evaluating players. Huberdeau's early struggles were exacerbated by poor puck luck, while Zary's hot start was partially fueled by good fortune. True evaluation lies in their play-driving metrics, which PDO helps to isolate.


4. Goaltending is Half the Story: A team's PDO is exceptionally sensitive to goaltending performance. Markström's fluctuations were the single largest driver of the Flames' PDO swings, underscoring how reliant any team's fortune is on its last line of defense.


5. A Tool for Roster Construction: For GM Craig Conroy, understanding PDO trends aids in distinguishing which players' performances are sustainable and which may be market inefficiencies to target or avoid in trades and free agency.

Conclusion

The 2023-24 season for the Calgary Flames serves as a compelling case study in the practical application of PDO analysis. It moves the discourse from abstract "puck luck" to quantifiable variance. The data reveals a team that, under head coach Ryan Huska, established a reasonably sound, possession-oriented identity—an identity that was often masked by the extreme swings of shooting and save percentage variance inherent to a single NHL season.

For the Flames' management, the lesson is clear: long-term success is built on consistently strong underlying process metrics—controlling shot attempts, generating quality chances, and suppressing them against. While the bounces will always play a role in the short-term outcomes of any single game at the Scotiabank Saddledome or on the road, they are not a viable strategy. The work of Conroy and his staff must focus on reinforcing the structural and talent-based foundations that keep a team's PDO from needing to be unsustainably high to win. As this analysis shows, when the regression comes—as it always does—it is the strength of that foundation that determines whether a team sinks or stays afloat in the relentless tide of the National Hockey League season.

For further statistical analysis and metric deep dives on the club, explore our hub for Flames Stats & Metrics Analysis.