Copy trading NHL prediction markets on Polymarket: the 84-game season

The 2026–27 NHL season opened on Tuesday, September 29, 2026 — a five-game opening night — and runs an expanded 84 games per team, 1,344 regular-season games in all, to Saturday, April 10, 2027, before the sixteen-team Stanley Cup playoffs. For a copy trader that is the second-largest regular-season sample of the four major North American leagues, behind only MLB's 2,430 games, and it comes with hockey's two particular properties: the moneylines are priced close to even, so a hit rate carries almost no information until the sample is large, and the wallets trading it are almost never hockey specialists. This guide covers where the markets live, how the season is shaped, what to check on a hockey-heavy wallet, and the graded NHL board's base rates as they stood two weeks in.

Where do NHL markets live on Polymarket?

Every NHL game is a Polymarket event carrying the nhl tag (alongside sports and games) with a series slug of nhl-2026 and between six and fourteen markets: the moneyline (the market whose slug is the event's own), puck lines, game totals and period lines. Season-long markets — the 2027 Champion market, conference and division winners, awards — sit in separate events under the same tag. They are thin by Polymarket's standards: on 2026-10-14 the champion market had taken $1.23M across 38 team markets, against $25.4M for the NBA's. Hockey on Polymarket is a game-by-game product.

CopyGrade's NHL board is built on the nhl and hockey tags, the way the NBA board is built on nba: a wallet qualifies with at least five trades on hockey-tagged markets and is ranked on its realized profit or traded volume on those markets alone. Because the hockey tag also carries other series — KHL, Liiga, SHL and AHL among them — some non-NHL fills roll into the board; the NHL series carries over nine-tenths of the settled hockey volume we track. The same tags power the NHL chip on the Event Hub, where every game gets a dossier listing the graded wallets active on it, on which side and at what price.

How is the season shaped, and why does it matter for copying?

The calendar below is the format, not a schedule. Every date was checked on 2026-10-14 against the league's own announcement of the 2026–27 schedule; re-verify before you trade one.

Three structural facts matter more than any date. First, hockey plays nearly every night for six and a half months, like baseball and basketball and unlike football, so a wallet trading the slate can settle hundreds of comparable markets in one season. Second, hockey moneylines are priced close to even: in the first two weeks of this season, 77% of the pre-game moneyline buys graded wallets placed were between 40¢ and 60¢ and only 2% were at 70¢ or above, against 12% at 70¢-plus on NBA preseason games and far more once the NBA regular season's lopsided favourites arrive. Third, hockey is low-scoring and goalie-dependent, which is why the lines are close: a hot goaltender decides games the better team should win, and a single season's variance is large even for a genuinely sharp wallet.

Put those together and the arithmetic of reading a hockey record changes. A wallet that hits 55% of 200 near-even hockey calls has beaten its prices by about ten games — a result worth reading and not yet proof. A wallet that hits 55% of 200 NBA favourites bought at 70¢ has lost to the market by thirty games. The same hit rate means opposite things, which is why the one number to read on any moneyline record is the hits its prices predicted — the recent-form guide walks the expected-hits sum — and why in hockey the sample has to be large before any record means anything at all.

What should you check on a hockey-heavy wallet?

The general vetting checklist applies in full. Hockey adds five emphases:

1. Run the farming check first, and expect it to fire often. On 2026-10-14 the NHL board was 56.8% flagged, the twelfth-highest share of our 26 category boards and well above the 34.9% site-wide rate. Every board sits above the site-wide rate — a board requires five trades in one category, and concentration is what the detectors read — so read the figure as a board-level base rate, not as a verdict on hockey. A farming-risk flag is a veto whatever the board's base rate.
2. Find out whether it is a hockey record at all. Of the 1,184 wallets on the NHL board that day, four traded hockey alone; the median NHL wallet qualifies for nine other boards and three in four are also on the NBA board. A wallet's NHL profit is usually a small slice of a generalist book, so open the verdict's category-edge breakdown before the board column: a whole-record edge built in soccer with a hockey footprint is a different copy from a hockey edge, and the score is computed on the whole record.
3. Price every hit, and demand sample. Near-even lines mean a hockey hit rate is close to a coin flip's until the count is in the hundreds. Where a wallet's trade ledger shows the prices it paid, sum them — that is the hit count the market expected — and read the record as the difference. The season-versus-bracket guide has the binomial table: at 100 settled markets a 55% rate is inside the noise of 50%.
4. Check when it trades. Starting-goalie news lands in the hours before face-off and moves hockey lines; in-play prices move on every goal. A wallet whose fills cluster after the puck drops has a live-trading record a bot copying it seconds later cannot reproduce. In the first two weeks of this season the clean cohort placed 70% of its fills before face-off and the severe cohort 32% — the first NHL data cut has the split. Pre-game positioning is the only part of a record a copier inherits at a comparable price.
5. Read the market's overtime rules before reading the record. Hockey games can end in overtime or a shootout, and whether a moneyline, puck line or total counts them is a per-market resolution rule. A wallet's "wins" on a market that excludes overtime are a different record from one that includes it; check the rule on the market page, not the headline.

How do you use the board and the tools during the season?

• The NHL board ranks graded wallets by hockey-attributed profit or volume, with each wallet's score, verdict and farming level beside it. Sort by volume first: it surfaces the wallets with sample, which is what a season is for.
• Scout filters the whole graded set by verdict, farming level, real edge, drawdown, minimum trades and volume, and by sport through its Topic filter — the shortlisting step before any single verdict is opened.
• Event dossiers on the Event Hub show, for one game or one futures market, which graded wallets are in it, on which side, at what price — the pre-game positioning a board cannot show.
• Board watch (Pro) fires when a tracked wallet you do not already follow opens a buy above your threshold on hockey markets — checked hourly, so a pre-game position reaches you within the hour. Off until you switch it on, like every notification here.
• The Graded Slate and the per-wallet recent form card currently cover the NFL, college football and MLB — the leagues with a slug-matched moneyline per game. Their rules are published, and any extension to hockey would follow them.

Nothing on any of those surfaces is a pick. CopyGrade never executes a trade; every surface describes what a set of graded wallets did.

The base rates, dated

All wallet figures are a point-in-time pull from CopyGrade's live scored set via the read-only database on 2026-10-14, two weeks into the season. The board recomputes with every sync and will grow through the winter; treat every figure as a dated level, never as a trend, and compare it only with figures carrying the same date. The first NHL data cut carries the full tables.

• Flagged share: 56.8% (672 of 1,184; 41.4% severe) — twelfth-highest of the 26 boards on that date, against 34.9% site-wide across 13,039 graded wallets.
• Negative real edge: 83.7% of the board, eighth-highest of the 26. The clean cohort is underwater more often (88.9%) than the severe cohort (79.0%) and by less (median −1.00% against −2.70%). Quote either half alone and the finding inverts.
• The clean cohort is net negative — 56% of the board's volume for a $14k loss — while the two flagged cohorts took all of the net profit, and the top wallet by hockey profit, severe-flagged, took 49% of it alone.
• Positive verdicts: 14, four of them observed for 60 days or more, with $2,271 of hockey volume between them. 147 wallets had 50 or more hockey trades — the population where a season record can begin to be read.
• Median CopyGrade Score on the board: 45.

"Real edge" is our post-fee, post-latency estimate of what a copier would have kept across a wallet's whole record, not its hockey record alone, per the published methodology; the hockey-attributed volume and profit are the board's own. A farming-risk flag is an algorithmic risk assessment, not an accusation of conduct, and a positive verdict is where diligence starts, not a recommendation — our own validation of whether the score predicts copier returns is still null.

CopyGrade is analysis-only — it never executes trades, holds funds, or custodies keys. A CopyGrade Score is a documented research opinion. Not financial advice.