Every hockey fan has a theory about attention: win, and the internet follows. We tested it on a full year of data.
We pulled 365 days of public social conversation about all 32 NHL clubs and the league itself with the LunarCrush CLI, 12 November 2025 to 11 November 2026, across X, Reddit, TikTok, YouTube, Instagram and 10K+ news sources. That is 27.20B interactions (views included) on the 32 club series and 21.93B on the NHL topic.
The full study, with interactive charts and a free dataset, is at lunarcrush.com/research/nhl-fan-attention-2026.
Winning barely moves attention
For each club we took its win share in the window (completed regular-season and playoff games; overtime and shootout losses counted as losses; the window covers the end of 2025-26 and the start of 2026-27) and its total interactions for the year. We measured attention two ways: the main keyword collection, and a stricter one that counts full club names only.
The correlation between win share and log annual interactions is 0.00 on the main measure, 0.27 counting full club names only, across 32 clubs. Swap in NHL points percentage (regular-season games, a point for an overtime or shootout loss) and it is 0.02 and 0.25. With 32 clubs, all of those are weak.
Two clubs show it on both measures:
- Carolina won the Stanley Cup and ranks 9th on the main measure (991.8M), 6th counting full club names only.
- Toronto won 32 of 86 games in the window and ranks 2nd on the main measure (2.05B), 3rd on full names.
The ranking
Montreal leads on the main measure; San Jose leads on full names. Montreal drew 2.26B interactions, ahead of Toronto (2.05B) and San Jose (1.83B). Nashville is 32nd at 275.9M. Counting full club names only, San Jose leads at 1.55B.
Each club series is a keyword collection: the full club name, its one-word form and its nickname. Ordinary-word nicknames (Hurricanes, Jets, Rangers, Kings, Stars, Lightning and the rest) only count with league, hockey or home-city context, and relevance on each club's 25 top posts ran 80% to 100%. Unique nicknames like "Canadiens" and "Canucks" match alone, which can lift those clubs; Vancouver is 5th on the main measure and 21st on full names. That is why the dataset carries both ranks. 8 of the top 10 hold their place within 3 ranks.
The top three clubs' largest days
The top three got there through very different days.
Montreal. Its three largest days were playoff nights: Game 7 in Buffalo (57.9M on 19 May UTC), Game 7 against Tampa Bay (42.7M on 4 May UTC) and the next round's opener (38.5M on 22 May UTC). French-language posts are visible in its top posts, including the club's own bilingual publishing.
Toronto. Its two largest days were lottery night (36.9M on 6 May UTC, after the 5 May ET draw) and the draft's first round (34.9M on 27 June UTC; the first pick was made on 26 June ET). Neither was a game night.
San Jose. Its largest day was 19.6M. Most of its total is a steady daily floor of posts from the club's own accounts and a large fan-edit community around Macklin Celebrini and Sharks forward Will Smith.
The league year, day by day
The largest day in NHL conversation followed the Olympic men's final: 23 February 2026 UTC, 147.2M interactions. The ten largest days were four Olympic days around the men's final, a league-news and Ovechkin milestone day, Utah's Zammoth day (a Mammoth-themed ice resurfacing machine), the draft's first round (125.4M), the draft lottery (111.2M) and two trade-deadline days.
No Stanley Cup Final game makes the list. The Cup-clinching date, 15 June UTC, drew 81.7M, less than the draft's first round ten days into the offseason by our calendar.
The season still matters in aggregate. In-season days average 70.5M interactions and offseason days (16 June to 28 September) 34.3M, a ratio of 2.05x.
Attention is concentrated
The top five clubs draw 33.0% of the 32 clubs' combined 27.20B interactions (views included; a post naming two clubs counts for both). The bottom five draw 5.7%. Montreal alone draws 8.2 times Nashville's total.
In interactions per home-market resident, with all 32 clubs normalized, Buffalo leads at 869, then Edmonton (742), Winnipeg (516) and Montreal (491). Interactions come from anywhere, so this is a scale index, not a count of local fans. New York's three clubs share one market, as do Los Angeles and Anaheim, which puts them at the bottom of that view.
Reproduce one number
Every series on the page is a public LunarCrush topic or collection. Here is how to get the largest league day yourself:
curl -fsSL https://lunarcrush.ai/install.sh | bash lunarcrush login # a year of daily metrics for the NHL topic lunarcrush topic nhl time-series --interval 1y --start 2025-11-12 --end 2026-11-12 --json > nhl.json import json from datetime import datetime, timezone rows = json.load(open("nhl.json"))["data"] day = lambda r: datetime.fromtimestamp(r["time"], timezone.utc).strftime("%Y-%m-%d") year = [r for r in rows if "2025-11-12" <= day(r) <= "2026-11-11"] peak = max(year, key=lambda r: r["interactions"]) print(day(peak), peak["interactions"]) # 2026-02-23, about 147.2M; totals move slightly as late engagement lands Each club works the same way with its collection id, for example Montreal:
lunarcrush topic lc1sQWOkyy time-series --interval 1y --start 2025-11-12 --end 2026-11-12 --json Public topic pages follow the same pattern: lunarcrush.com/topic/lc1sQWOkyy. All 32 collection ids, plus the full-name-only ids, are columns in the team table download.
The shipped club series remove two items you will see in a raw pull: activity credited to an older post beyond its lifetime interactions (4.0% of the combined club total), and a recurring non-hockey weather video matched by Tampa Bay's collection (302.1M interactions).
How the data works
- Interactions include views alongside likes, comments and shares, so they are not unique people. They are summable across days.
- Active contributors and active posts are unique counts per day. We report them as daily means and never sum them.
- Sentiment is the daily 0 to 100 score, weighted by each day's interactions for the annual figure.
- A post naming two clubs counts for both, so the 32 club series sum to more than the league topic. Every share uses the 32 clubs' combined total.
- Correlations are across the 32 clubs, not across games. All dates are UTC; an evening game in North America often lands on the next UTC date.
Take the data
The dataset is free on the study page: the 32-club table with both ranks, win share and points percentage, every club by day (11,680 rows), the league by day, in CSV and Parquet, plus an AI guide with the schema and validation anchors so an assistant reproducing the numbers gets ours.
The same read works for any club and its rivals, a sponsor, a player, or any brand or ticker, hourly, through the API, the CLI, bulk files or the LunarCrush MCP server for Claude, ChatGPT and other assistants.
Social data: LunarCrush daily series, 12 November 2025 to 11 November 2026, pulled 12 and 13 November 2026. Results: ESPN completed regular-season and playoff games. Population: 2025 U.S. Census CSA and Statistics Canada CMA.
Independent research by LunarCrush. Not affiliated with or endorsed by the NHL or any club.
