Fantasy Hockey Tools

A set of tools for fantasy hockey managers. Currently live:

• Draft Assistant (/draft) - see below.
• Skaters (/skaters) - every skater ranked by C-Score over Last 5 / Last 10 / Season / Last Season, split into Forwards and Defense, plus Compare and a Drop & Replace flow for streaming. Drop & Replace can stage several adds as one move and shows their combined starts day by day, so two streamers can't both count the same open slot.
• Goalies (/goalies) - Start Tracker (each goalie's recent vs. season share of starts) and Spot Starts (ranked by estimated win probability).
• Deployment (/deployment) - skaters whose ice time or power-play share in their most recent game jumped or dropped against a chosen baseline. Split into Forwards and Defense tabs, 50 players per page. A Current Roles tab shows each skater's raw TOI, PP TOI and PP share (last game, last 4, or season), sortable by column - pair it with Unowned to scan free agents.
• Trade Targets (/trades) - buy-low skaters: strong C-Score but running below their own career S%, oiS% and IPP. With a Yahoo league connected it shows who owns each one, and "Build Offer" ranks your skaters as 1-for-1 offers by whether they fill a hole or thin spot in that team's lineup, whether your own lineup stays full, and how close their points per game are.

Skaters and Deployment can filter to unowned players. With a Yahoo league connected they use that league's real free agents; otherwise they use a manual roster.

Getting Started

npm run dev

Open http://localhost:3000.

Draft Assistant

At /draft. Workflow:

1. Import a ranking — upload a CSV of your player rankings (or a projections export) and map its columns to the app's fields. Common header names (Name, Team, Pos, Rank, ADP, G, ...) are auto-detected. Only a name column is required; if there's no Rank column, the row order in your file is used as the ranking. Grab the "Download an example CSV" link on the upload screen if you want a template to start from.
2. Set league settings — team count, your team, roster slots, draft rounds.
3. Draft live — the Cheat Sheet shows players in ranked order. Click "Draft" as picks happen (yours or opponents') to advance the snake draft board. My Team and Pick Suggestions update as you go. Its Schedule column shows a Low/Med/High badge per player for how much their team's schedule overlaps with players you've already rostered at a shared position (dual/tri-eligible players are checked against all of their positions, and UTIL counts too) — hover a badge to see which of your teams it's comparing against. Nothing rostered at a shared position yet? Always Low. This resets to 0 relative to your current roster, not a fixed per-team stat, so it's most useful once you've got a few picks in at a position.
4. Schedule tab — shows which teams play the most on league-wide "off-nights" (light game-volume dates), pulled live from the NHL API — useful for streaming-friendly picks late in the draft.
5. Schedule Fit tab — pick a position, and it ranks the remaining players there by how much their team's schedule overlaps with your already-rostered players at that same position (computed live from the full NHL season schedule, same source as the Schedule tab). Lower overlap means more distinct game-days covered if you're rostering multiple players at that spot — useful for deciding between two similarly-ranked players at a thin position.

Stat projections and scoring are optional and layered on top: if your CSV also includes stat columns (goals, assists, etc.), configure points-per-stat under Settings and the Cheat Sheet switches to ranking by computed Fantasy Points and Value Over Replacement (VOR) instead, with auto-generated tiers.

All draft state (players, scoring, picks) persists to localStorage, so a page refresh mid-draft won't lose anything. Use "Reset draft picks" or "Clear all players" in Settings to start over.

CSV format

One row per player. At minimum needs a name column; team, position, rank, ADP, and stat columns are all optional and mapped manually on import. See src/lib/csv.ts (FIELD_ALIASES) for the column keys recognized out of the box — you can add any other stat as a custom scoring entry.

Yahoo positions (optional)

If your ranking CSV's positions don't exactly match Yahoo's fantasy eligibility (e.g. a dual-eligible C/LW player), the app can apply Yahoo's own position data automatically on every import - no login, no per-user step, for anyone using the app. This comes from src/data/yahoo-positions.json, a bundled dataset built once (and refreshable anytime) from a CSV of Yahoo player data:

npm run build:yahoo-positions -- path/to/your-file.csv

It auto-detects the Name/Position/Team columns (same alias matching as the main ranking import - see src/lib/csv.ts) and writes src/data/yahoo-positions.json. Since positions rarely change mid-season, you'd typically only need to do this once before drafting. For one-off touch-ups afterward (a player who was missed, or whose eligibility changed), Settings also has a manual paste box: copy a small Name + Position table and paste it in to update just those players in your current session (this one doesn't touch the bundled file).

If a player's name in your ranking CSV doesn't match the bundled dataset (different spelling, nickname, etc.), positions just won't be applied for them - no error, they keep whatever position was in your CSV. Tell me the mismatched name pairs and I can add them as aliases in src/data/yahoo-positions.json.

ADP (optional)

Same idea as positions, for Average Draft Position: src/data/adp.json, built from a Name,ADP CSV via:

npm run build:adp -- path/to/your-file.csv

Unlike positions (which always override), this only fills in ADP for players who don't already have one from their own ranking CSV - it won't clobber ADP you already imported. Rows with a non-numeric value (e.g. "N/A" for undrafted players) are skipped automatically.

Both datasets use the same accent/case/whitespace-insensitive name matching, and Settings shows how many players are covered by each.

Comparing multiple ranking sources (optional)

Beyond your main imported list, Settings has an Additional Ranking Sources box for pasting in other rankings (a friend's list, another site's rankings) purely to compare - paste Name + Rank rows, give it a label, and it shows up as its own column on the Cheat Sheet, matched to your main list by name. An Avg Rank column also appears, averaging each player's own rank plus every source that has them (so a player only one source ranked still just shows their own rank - the average isn't dragged down by sources that don't cover them). Unlike the bundled Yahoo/ADP datasets, these live in localStorage with the rest of your draft state, not a file in the repo - add or remove them anytime per session.

Stack

Next.js (App Router) + TypeScript + Tailwind CSS, deployed on Vercel. Zustand for Draft Assistant state (persisted to localStorage), Papaparse for CSV import.

• Database: Postgres via Drizzle ORM (Neon in production). Note that .env.local's DATABASE_URL points at a local Postgres, so a local drizzle-kit push does not reach production.
• Auth: NextAuth with the Drizzle adapter, plus a separate Yahoo OAuth connect flow for league rosters and free agents.
• Data sources: Natural Stat Trick (skater/goalie stats), the NHL API (schedule, standings), and the Yahoo Fantasy API (leagues, rosters, free agents, position eligibility).
• Caching: two daily Vercel crons (vercel.json) refresh DB-backed caches for Yahoo position eligibility and the NST-backed tools, so page loads never wait on a live NST or Yahoo call.

Roadmap

See ROADMAP.md.