LINHAC 2026 — Conference Report
Conference Report · June 2–4, 2026
LINHAC 2026
5th International Ice Hockey Analytics Conference · Linköping · 36 sessions
Six big ideas from the conference
01 — Data is infrastructure Analytics is no longer a competitive edge — it is the operating system. Lalonde: the clubs still treating it as a side project are already behind.
02 — Own your stack Own your video, events, models, and raw data — not a vendor dashboard. Every step you outsource is a step you lose when the game changes. “Own your data. Control your AI.”
03 — Data → behavior Getting the data right is the easy part. The hard part is never finished — how you communicate it, whether it lands, whether behavior actually shifts. You have to keep testing new ways to get the message through.
04 — Test, learn, iterate Hockey is starting to work the way business has for decades: hypothesize, test, learn, iterate. That is a hard sell for any vendor offering a standard solution.
05 — Story at scale Hockey already has the data. One model, millions of personal stories — for players, fans, and sponsors. The only question is whether your org decides to use it.
06 — Design the org, not just the team Hockey keeps recycling the same coaching profiles. Other major sports are hiring economists, data scientists, and academics with no traditional sports background. The question nobody wants to answer: how much of that hiring logic survives AI?
Key numbers from research
OR 5.05 — Pull the goalie earlier Down one with more than 2:30 left (vs 2:00 or less): 5x better odds to score, and 0.5–3.2 extra standings points per season.
+4.3% — Home ice matters most on the power play +4.27% Corsi on the PP vs +2.02% at even strength, and about 10% better odds to score. The PP does not help road teams catch up.
r = .437 — EPV passing tracks with coach ratings Best match with how coaches rate passing smarts (n=120). Raw shot volume barely matters (r=0.08).
Δ +0.001 — Keep career models simple Age, size, games played, draft spot, and region explain almost everything. Fancy social inputs add almost nothing.
From data to behavior on the ice
MODO U20 (Felix Dahlin): data sets the bar; players choose the behaviors to hit it. Five win factors run the season — slot shots, blocked shots %, shift length, penalties — talked through in small groups, not handed down by coaches.
Färjestad BK (Erik Wilderoth): weekly reports beat game-by-game noise. SIF (Veronica Eriksson): own the full pipeline — video → events → metrics — in-house. Spiideo: most clubs track games but not practice. The same chain — capture → data → insight → decision — needs to cover both.
Beyond the ice and the analytics room
Data Talks (Stefan Lavén): stop running one-off campaigns. Build a fan OS loop — collect, unify, segment, activate, measure. One fan profile across tickets, merch, digital, and sponsors, with churn and lifetime value driving what you send next.
49ing (Andreas Hänni): hockey already has the data. The question is whether it becomes league highlights or personal stories. AI on match data means “one model, millions of stories.” With Claude, approved users can build their own views.
Ray Lalonde: analytics is moving from edge to core infrastructure. The Tampa Bay Rays compete without a top payroll because data is how the org runs — not a side project. Timo Seppa: garbage in, garbage out. Fix your data before you add more dashboards.
What hockey can learn from AI
David Radke (Chicago Blackhawks): use game sims (FC 24, RoboCup) to test tactics. Game Plan (DeepMind × Liverpool FC) points to three building blocks — stats, game theory, and computer vision — as a hockey AI roadmap. His ranking paper with Troy Mulholland shows that Elo, social choice, and basic stats can rank the same teams differently. How you rank is a choice, not just math.
Personal reflection: Data for the whole organization
My read on LINHAC — shaped by the sessions and by what I see daily in a large enterprise organization. Hockey is not there yet, but it is on the same change journey business is running through now, only faster. One argument in four steps.
1 · The shift — less dashboards, more leadership
LLMs have made analysis a commodity. Anyone can prompt their way to a dashboard. In business, this is already everyday reality — dashboards everywhere, insight everywhere, change nowhere. The bottleneck is no longer getting dashboards out; it is finding people who can drive data-driven change. Hockey is a few steps behind, but heading for the same wall. Next year’s conference will have more graphs, data, and AI than anyone can absorb. Dashboards are not the bottleneck anymore.
2 · The response — data engineering as foundation
The bottleneck is the data platform — how it is built and continuously improved. That shapes not just game and opponent analysis, but how efficiently the organization runs on the business side too. Enterprise is investing heavily in this now across the whole operation: marketing, sales, finance. One shared platform, structured data, engineers who build and maintain it — the base for efficient AI flows and lasting efficiency, with data that stays when people move on. Hockey is still on the old model: each analyst arrives with their own tools and takes them to the next club. Clubs need to build that capability in the organization.
3 · From generalist to specialist
Once the shared platform is in place — with today’s tools, that is 3–6 months of work — the analyst role changes. The generalist who collected data, built dashboards, and ran every report does not scale anymore. AI handles the basic analytics. What the org needs instead are deep specialists: pass data, shot data, movement patterns — people extreme enough in one domain to actually move the needle. That will reshape who works in hockey analytics and open the door to subject matter experts who may not come from hockey at all.
4 · Small clubs can punch up
In business right now, small AI-native challengers are taking on giants that were never built to automate from scratch. Hockey has the same window — but only if a club invests in data foundation, runs the organization more efficiently, and finds new revenue streams — which is what data makes possible. Smaller clubs can challenge the big ones the way startups are challenging incumbents across every industry. That takes thinking differently and hiring educated people from different backgrounds — not the buddy or former teammate.
Appendix A — References
- Kumagai, B. (Teamworks). New Frontiers in Hockey Analytics with Player Tracking Data. LINHAC 2026, session 1, 2 June 2026, Linköping.
- Eriksson, V. (Swedish Ice Hockey Federation). Coach’s challenge: Own your data. Control your AI. LINHAC 2026, session 17, 3 June 2026, Linköping.
- Lalonde, R. (Ray Lalonde Sports+). From Competitive Edge to Core Infrastructure: The Global Rise of Sports Analytics. LINHAC 2026, session 24, 3 June 2026, Linköping.
- Seppa, T. (Hockey Data Show / NHL Network Radio). Garbage in, garbage out: What’s in your analytics? LINHAC 2026, session 23, 3 June 2026, Linköping.
- Dahlin, F. (SIF & MODO Hockey). Behavior-driven development — Fueled by analytics. LINHAC 2026, session 19, 3 June 2026, Linköping.
- Wilderoth, E. (Färjestad BK). Raise the floor — Learnings from Data-Driven Player Development. LINHAC 2026, session 18, 3 June 2026, Linköping.
- Palvalin, M. Validating EPV Against Expert Coach Assessments of Passing Intelligence in Elite Youth Ice Hockey. LINHAC 2026, session 3, 2 June 2026. n=120. PDF
- Davis, J., Bransen, L., Devos, L. et al. (2024). Methodology and evaluation in sports analytics: challenges, approaches, and lessons learned. Machine Learning, 113, 6977–7010. doi:10.1007/s10994-024-06585-0
- Radke, D. & Mulholland, T. Non-Transitivity in the NHL: Ranking Teams, Lines, and Players using Game Theory. LINHAC 2026, session 2, 2 June 2026. Paper
- Goldstein, E. J. & Pearson, J. A. (ORRO AI GENIUS). Multi-Horizon Career Longevity Prediction for NHL Skaters. LINHAC 2026, session 6, 2 June 2026. Dataset: 5,754 skaters, 1979–2024. PDF
- Ali, S. A. & Mohamed, H. (University of Waterloo). Closing the Gap: A Longitudinal, Covariate-Adjusted Analysis of NHL Goalie Pull Timing. LINHAC 2026, session 5, 2 June 2026. Dataset: 7,047 pulls across 10,250 games, 2015–16 to 2024–25. PDF
- Riccardi, N., Campbell Jr., T. & Paul, R. J. (Syracuse University). NHL Home-Ice Offensive Advantage and the Power Play. LINHAC 2026, session 7, 2 June 2026. Dataset: 4 seasons, 2021–22 to 2024–25. PDF
- Zou, Y. & Schuckers, M. A Bayesian Approach to Estimating an NHL Draft Value Pick Chart with Bounds. LINHAC 2026, session 4, 2 June 2026. Dataset: NHL drafts 2009–2018, tracked for 7 seasons each. PDF
- Lavén, S. (Data Talks). From followers to revenue — what hockey orgs get wrong about fans. LINHAC 2026, session 22, 3 June 2026, Linköping.
- Hänni, A. (49ing). Data use beyond analytics. LINHAC 2026, session 20, 3 June 2026, Linköping.
- Radke, D. (Chicago Blackhawks). What Can Hockey Analytics Learn from AI? LINHAC 2026, session 25, 3 June 2026, Linköping.
- Tuyls, K. et al. (DeepMind & Liverpool FC). (2021). Game Plan: What AI can do for Football, and What Football can do for AI. Journal of Artificial Intelligence Research, 71, 41–88. doi:10.1613/JAIR.1.12505
- Paul, R. J., Riccardi, N., Pauline, G. & Weinbach, A. Climate and Income Effects on U.S. Hockey Participation. LINHAC 2026, session 8, 2 June 2026. Panel data 2009–2023, state level. PDF
- Cook, D. & Zeba, M. (Spiideo). From Capture to Decision: Building the Modern Sports Intelligence Stack. LINHAC 2026, session 21, 3 June 2026, Linköping.
- Boulet, L. (LB-Hockey). Defining Physicality and Skaters’ Ability to Play Through It. LINHAC 2026, session 12, 2 June 2026. PDF
Appendix B — Session index
All 36 sessions. Bold rows have detailed personal notes.
| # | Speaker | Title | Paper |
|---|---|---|---|
| June 2 — Academic Day | |||
| 01 | Brendan Kumagai · Teamworks | New Frontiers in Hockey Analytics with Player Tracking Data | — |
| 02 | Radke, Mulholland | Non-Transitivity in the NHL: Ranking with Game Theory | |
| 03 | Miikka Palvalin | Validating EPV Against Expert Coach Assessments | |
| 04 | Zou, Schuckers | Bayesian NHL Draft Value Pick Chart with Bounds | |
| 05 | Ali, Mohamed · U. Waterloo | Closing the Gap: NHL Goalie Pull Timing | |
| 06 | Goldstein, Pearson · ORRO AI GENIUS | Multi-Horizon Career Longevity Prediction for NHL Skaters | |
| 07 | Riccardi, Campbell Jr., Paul · Syracuse | NHL Home-Ice Offensive Advantage and the Power Play | |
| 08 | Paul, Riccardi, Pauline, Weinbach | Climate and Income Effects on U.S. Hockey Participation | |
| 09 | Quinton J. Krueger | Who’s In by Game 15? NHL Playoff Qualification Forecasting | |
| 10 | Sezgin, Quinn | Gaussian Rink Control-Based Expected Threat Framework | |
| 11 | Krueger, Velte, Plocki, Carone | Net Man-Games Lost as Coaching and GM Evaluation Metric | |
| 12 | Louis Boulet · LB-Hockey | Defining Physicality and Skaters’ Ability to Play Through It | |
| June 2 — Student Competition | |||
| 13 | Arrestam, Bertmar | SHL Powerplay Efficiency — Sequences to Styles | |
| 14 | Man, Li, Fan | Zone Entry to Danger — Post-Entry Sequences | |
| 15 | Riemenschneider | Power-Play Re-Entries (ML) | |
| 16 | Kiran Roye | Shot Surface Distributions from Zone Entry | |
| June 3 — Hockey Day 1 | |||
| 17 | Veronica Eriksson · SIF | Coach’s challenge: Own your data. Control your AI | — |
| 18 | Erik Wilderoth · Färjestad BK | Raise the floor — Data-Driven Player Development | — |
| 19 | Felix Dahlin · SIF & MODO | Behavior-driven development — Fueled by analytics | — |
| 20 | Andreas Hänni · 49ing | Data use beyond analytics | — |
| 21 | Cook, Zeba · Spiideo | From Capture to Decision: Sports Intelligence Stack | — |
| 22 | Stefan Lavén · Data Talks | From followers to revenue | — |
| 23 | Timo Seppa · Hockey Data Show | Garbage in, garbage out — What’s in your analytics? | — |
| 24 | Ray Lalonde · Ray Lalonde Sports+ | Global Rise of Sports Analytics | — |
| 25 | David Radke · Chicago Blackhawks | What Can Hockey Analytics Learn from AI? | — |
| 26 | Panel · Brecht, Hamann, McMillan, Radke, Weaver | Analyzing NHL data and workflow | — |
| 27 | Mike Kelly · Sportlogiq | Reception — NHL play-offs update | — |
| June 4 — Hockey Day 2 | |||
| 28 | Erik Lignell · HC Fribourg-Gottéron | Successful analysis is communication — Not Location | — |
| 29 | Panel · Arponen, Morkes, Bezdek, Lignell, Malmquist | Analytics workflows in European teams | — |
| 30 | Ola Lidmark Eriksson · Playmaker AI | 10 Years of Football Analytics Meets Hockey | — |
| 31 | Hänni, Paterlini · 49ing / SCL Tigers | From Goalie-Driven Wins to System Stability (NL) | — |
| 32 | Larsson, Bibic · TV4 | Analytics in broadcasts | — |
| 33 | Sjöholm, Gullbrand | The Hidden Map to Gold | — |
| 34 | Nordfjell, Almqvist Andersson · d-fine / SIF | Skating from Raw Data to Clear Insights | — |
| 35 | Neil Lane · Stathletes | The next frontier of hockey analytics | — |
| 36 | Organisers | Closing — Best paper & student competition winners | — |