Podcast Episode #5 – Josh Pohlkamp-Hartt – Boston Bruins

In our fifth episode of the Sport Analytics Podcast, host Amrit Vignesh sits down with Dr. Josh Pohlkamp-Hartt, the Associate Director of Hockey Analytics at the Boston Bruins. From designing novel neutral-zone metrics with the Kingston Frontenacs to sharpening in-game analytics for one of the NHL’s most storied franchises, Josh has seen firsthand how data transforms pro hockey decision-making.

Josh explains how he translates complex statistical models into quick-hit insights for coaches, scouts, and front-office staff—from daily player evaluations to long-term trade and contract strategies. He also draws on his PhD training in statistics (and prior experience at Apple) to highlight the importance of building user-friendly dashboards, fostering trust with skeptical stakeholders, and doubling down on communication skills to make your analytics truly actionable.

Whether you’re an NHL hopeful looking to automate your own analytics workflow or a data enthusiast intrigued by real-time puck- and player-tracking, this episode offers a deep dive into how advanced stats can give a storied hockey club its competitive edge.

🔔 Subscribe to our channel for more episodes featuring leaders in sports analytics, career advice, and technical breakdowns!
📧 For inquiries or collaborations, contact Dave Yount at dave@sportanalytics.com.

Amrit (Host): Welcome to the fifth episode of the Sport Analytics Podcast. Today, we’re joined by Josh Pohlkamp-Hartt, who is the Associate Director of Hockey Analytics at the Boston Bruins. Josh, how are you doing?

Josh: I’m good. It’s a nice Thursday, the team isn’t playing, so it’s a relaxing day.

Amrit (Host): Great! Could you give us an overview of your role with the Bruins—your day-to-day responsibilities and which areas of hockey operations you spend the most time supporting?

Josh: My work varies by season, but right now I focus heavily on supporting coaches and management with decisions like who plays, what strategies we use, and player acquisitions. With the trade deadline coming up, there’s a lot of analysis around player performance and projected value. I also maintain models like our expected goals system, incorporating new data to improve accuracy. A typical day might involve coding and model-building, plus writing reports that go directly to our staff.

Amrit (Host): We hear you touch everything from scouting prospects to helping with injury rehab. How does analytics inform these diverse areas of hockey operations?

Josh: We have a mindset that anywhere there’s a decision to be made and data is available, we want to harness it. Scouts accumulate reports, and we gather that info to provide both predictive analytics—like the probability a prospect becomes an NHL player—and descriptive stats, like how all the scouts graded a player’s skating. On the rehab side, we track certain metrics privately through our sports science group and give them tools to analyze that data efficiently. Overall, it’s about organizing and making data actionable rather than prescribing one “right” move.

Amrit (Host): How do you translate complex models into insights that coaches, players, and execs can use quickly?

Josh: It’s about meeting them on their terms. If you’re talking about expected goals, keep the units in “goals” so they have an immediate frame of reference. Use color-coding or percentiles—they get 100 vs. 50 out of 100. Don’t overload them with model internals; build trust by proving the metrics align with what they already see on the ice. Over time, once they see that data is consistent with their experience, they’re more likely to rely on it.

Amrit (Host): You earned a PhD in statistics. How did that blend of academic rigor and sports passion prepare you for the NHL?

Josh: A PhD teaches you to handle data end-to-end: cleaning, model-building, and communicating results. My advisor hammered home clarity in writing—labeling something “significant” vs. “important” is crucial. That said, academia can be verbose, while pro sports requires concise, executive-style communication. It’s a big shift, but the analytical foundation is incredibly useful.

Amrit (Host): You established an analytics program with the Kingston Frontenacs. How did that come about, and what were your first big breakthroughs in OHL hockey data?

Josh: I pitched the idea of measuring neutral-zone metrics, and they let me track data in exchange for providing basic stats like Corsi. A group of students and I would watch games, track shot share, and analyze neutral-zone transitions. We discovered some teams, like Sault Ste. Marie, had unique systems—like constant regroups—that broke our models. It taught me the importance of matching metrics to real styles of play, not just theoretical ones.

Amrit (Host): You spent three years at Apple Maps. How did large-scale data projects and autonomous systems influence your approach to hockey analytics?

Josh: Apple had immense resources; you could throw hundreds of staff at a single data-cleaning task. In the NHL, you can’t hire 500 people to manually fix all your data. So you learn how to automate efficiently and work with a smaller team. I also learned how to deliver projects faster—speed matters a lot more in sports, where you need insights quickly for roster or tactical decisions.

Amrit (Host): How do you balance quick-turnaround analytics (for daily tactics) with more complex, future-oriented projects like trades or contracts?

Josh: It’s always a juggling act. One approach is automation: if we can make a report or dashboard self-serve for coaches or scouts, we do it. That frees us up to work on deeper models. But you have to discern whether a request is actually high-priority or just a random curiosity. Saying “no” diplomatically is a skill. Ultimately, communication and setting expectations are vital.

Amrit (Host): Which programming languages and databases do you rely on most, and how has your tech stack evolved since joining the Bruins?

Josh: Our data science team uses R and Python, though our engineers prefer Python for easier productionizing. We host everything on AWS and have gone through multiple iterations of our front-end. It’s a balance between adopting modern tools (like React) and keeping things stable. Our engineers handle the web dev side with Java or Python, while stats folks might prefer R for deep analysis.

Amrit (Host): With player and puck tracking ramping up, how do you handle massive real-time data, and what new metrics does it unlock?

Josh: It’s a challenge. We use distributed systems and AWS solutions for huge data volumes, which requires new skills like Spark or GPU computing. Once you know where everyone is at all times, concepts like space ownership and defensive attribution become possible. A huge defender like Zdeno Chara might never let opponents near the net, which is a big reason no “bad” events happen in front of him. Now we can actually measure that without inferring from pure on-ice events.

Amrit (Host): Coaches and players can be skeptical. How do you build trust and prove data is a competitive edge, not an extra burden?

Josh: Simplify their workflow. If a coach wants to see all of an opponent’s turnovers, we can filter that in seconds. Provide reliable data and speak in their language. One minor error can ruin credibility if the dataset is small, so accuracy is crucial. But once they see that analytics saves them time and confirms what they observe on the ice, they’re usually all-in.

Amrit (Host): For students aiming for an NHL analytics role, which technical or domain skills matter most?

Josh: “Coding to completion”—taking a project from idea to actionable insight. That includes data cleaning, model design, accuracy evaluation, and final reporting. Practicing that full cycle is key. Many teams recruit from hackathons or competitions like the Big Data Cup because those projects mirror real in-house tasks.

Amrit (Host): Are there gaps in current sports analytics curriculums, and any final words for aspiring analysts?

Josh: Overall, I’m impressed with what students know, but they need more reps. Completing projects fully and honing communication are often overlooked. You can’t just produce a fancy model; you have to explain it succinctly and match it to real-world hockey decisions. Practice that, and you’ll stand out.

Amrit (Host): Thanks so much, Josh, for sharing your experiences. This has been incredibly insightful.

Josh: Thank you. I really appreciate it.