TigerTrade Terminal

A dark, institutional-desk trading terminal with two products under one shell, switchable from the top bar. Built on the TigerTrade Terminal design system (Obsidian theme): a command palette (⌘K / /), a polar radar scope, right-side analysis drawers, a persistent watchlist, and animated SVG charts.

1. Volatility & Momentum Radar (default) — a macro-event surveillance dashboard for 2026–2027. Tracks the scheduled catalysts that drive cross-asset volatility, grouped into five regimes:

   • Central Bank & Sovereign Liquidity — FOMC, BoJ, ECB, Treasury QRA, fiscal X-dates
   • Mechanical Flow Rebalancing — quad witching, S&P / Nasdaq-100 / Russell reconstitutions
   • Growth & Tech-Thematic Catalysts — earnings seasons, NVIDIA GTC, Apple WWDC, Google I/O, mega-cap IPOs
   • High-Impact Macro Data — NFP, CPI/PPI, ISM PMIs
   • Geopolitical, Commodity & Regulatory — US midterms, OPEC+, DOJ/antitrust

   Views: Radar (next-catalyst card, radar scope + expandable event rows), Full Timeline (month-grouped), Calendar (month grid with event chips), and Playbook + AI (desk briefing + watch list). Each event opens a right-side drawer with historical reaction stats, a move distribution, and cross-asset reaction. Filter by category, minimum weight, or free-text search.

   FOMC, quadruple-witching (3rd Friday), Russell reconstitution (last Friday of June), NFP (first Friday) and the US midterms use exact dates. Items marked ~ (BoJ, ECB, QRA, CPI/PPI, ISM, earnings windows, conferences, OPEC+) are projected from standard schedules — verify against official calendars before trading.

2. Leadership Screener — a relative-strength growth screener built on the TigerTrade Leadership Model (LEADERS), our own 7-factor framework: L (Leadership / RS rank), E (Earnings momentum), A (Accumulation / volume demand), D (Durable annual growth), E (Emerging breakout / new high), R (Rising institutional sponsorship), S (Setup — market trend). Each name scores 0–100 with a letter grade. Features:

   • Screener — sortable/filterable universe with per-factor pass-fail chips, RS rank, and live buy-zone detection (Basing → Approaching → In Buy Zone → Extended vs. the pivot). The Window control (1D/1W/1M/3M/1Y) is the period everything is measured and ranked over, not just a price format: price change, the RS percentile, the leadership L chip and the score all recompute over it, so switching to 3M re-ranks the board on 3-month relative strength. It defaults to 1Y — the model's classic 12-month definition — and any column not showing 1Y is labelled with its window (RS · 3M, Score · 3M).
   • Market Health — the "Setup" gate: index trend vs. 50/200-DMA, distribution-day count, last follow-through day, and breadth. Buy only in a Confirmed Uptrend.
   • Playbook — an AI screener read plus an "actionable now" list of names in their buy zone.

   The TigerTrade Leadership Model's factors follow classic leadership-investing principles popularized by William J. O'Neil. TigerTrade is independent and not affiliated with, sponsored by, or endorsed by Investor's Business Daily; CAN SLIM® is a registered trademark of Investor's Business Daily, Inc. Educational use only — not investment advice.

   • Click any row for a right-side stock drawer: price/volume chart with pivot & buy-zone band, RS line, buy-point analysis, the 7-criteria breakdown, fundamentals, and thesis.

The site uses the dark Obsidian terminal theme (warm amber accent) with a token-based design system, grain overlay, and motion that respects prefers-reduced-motion.

A Claude-backed serverless proxy (/api/claude) is available for AI features so the Anthropic API key never reaches the browser.

Live market data (recommended before trading)

The screener attempts to load live quotes on every visit. Its data state is shown in the Leadership Screener hero line:

• Live prices · as of … · N/N live (green pulse) — price and % change are live and recompute the buy-zone status in real time (live price vs. pivot).
• Demo prices · not live (red) — the live feed is unavailable, so the screener falls back to an illustrative dataset that is clearly labeled and never presented as real.

To turn the feed on, set FMP_API_KEY in your Vercel project env vars (get a key at https://site.financialmodelingprep.com); requests route through the /api/fmp proxy (your key never reaches the browser). Single-symbol quotes work on the free tier; the dashboard tries a batch call first and falls back to per-symbol quotes automatically.

What's live vs. analyst-maintained: prices, % moves, off-52-week-high and relative volume come live from FMP. The slower-moving Leadership-model inputs — earnings growth, RS rank, institutional read, pivot, base pattern and catalyst — are maintained in the editorial dataset and labeled as model inputs, not live quotes. Verify against your broker before trading.

Precompute snapshot + durable storage (optional, for scale)

/api/snapshot computes the whole universe's quotes + momentum signals server-side and serves them in one request, so the browser doesn't fetch every name. A daily cron (vercel.json, weekdays 22:00 UTC) refreshes it after the US close.

By default the snapshot is edge-cached. To make it durable (the cron writes it once; user requests only read it, never recompute — needed as the universe grows to hundreds), connect a Vercel Blob store:

1. Vercel dashboard → the Storage tab in the top nav (it's alongside Overview / Integrations / Activity — not inside the project's Settings, which is where people look first and don't find it) → Create Database → pick Blob → name it → Create. Then open it, go to Projects, and Connect it to this project.
2. Vercel auto-adds the BLOB_READ_WRITE_TOKEN env var. Redeploy — env vars only apply to new builds.
3. The snapshot now reads/writes Blob automatically; without it, it falls back to compute-on-demand + edge cache (zero setup, still works).

Connect Blob before you consider paying for a bigger data plan. Without it every cache miss recomputes the whole universe, and each recompute spends ~13 FMP calls on the macro board, VIX and the earnings calendar before it touches a single quote. That is how those three go blank together: the API quota is spent, not the code broken. With Blob connected the cron computes once per weekday and normal reads cost nothing upstream.

Check which mode you're in: load /api/snapshot and look at the top level — "blob": true and "served": "blob" is right; "blob": false with "served": "compute" means every visit is burning quota.

Report dates for names outside the S&P 500

The shared snapshot is the same for everyone, so it can't carry report dates for your holdings outside the tracked universe. /api/earnings?symbols=NBIS,ASML fills that gap, trying each source in turn and stopping at the first real answer:

#	Source	Cost	Notes
1	Blob cache	0 requests	A date resolved once is served to everyone afterwards.
2	Finnhub calendar/earnings	1 request	Set FINNHUB_API_KEY and this becomes the primary source. A documented, supported API that takes a symbol filter, so one request answers for one name — and it ships the forward EPS/revenue estimate with the date. Skipped entirely without a key.
3	Yahoo chart events=earn	1 request	No cookie, no crumb. The same endpoint the snapshot already reaches ~500×/night, so it's the Yahoo path known to work from this deployment.
4	Yahoo quoteSummary	3 requests	Richer (projected-date flag, EPS history, revenue) but cookie+crumb gated, so it costs a handshake first. Asked only when the chart returns no date or no reported quarter.
5	Stale cache	0 requests	A date from an earlier read, flagged stale, rather than a blank.

Getting the Finnhub key (free, ~2 minutes, no card):

1. Go to https://finnhub.io/register and sign up.
2. The dashboard shows your API key immediately — copy it.
3. Vercel → your project → Settings → Environment Variables → add FINNHUB_API_KEY, paste the value, tick all environments, Save.
4. Redeploy (Deployments → ⋯ → Redeploy). Env vars only apply to new builds.
5. Check it: /api/earnings?symbols=NBIS&debug=1 should show a {"step":"finnhub","status":200} line and src: "finnhub" on the result.

The free tier allows 60 calls/minute, and the Blob cache means a symbol is asked at most twice a day, so a normal portfolio uses a handful of calls per day.

Residential proxy for the Yahoo calls (optional)

Yahoo's endpoints are unkeyed, so they're defended by IP reputation instead — datacenter ranges get rate-limited, and Vercel's are heavily used. That's the likeliest reason the cookie/crumb handshake fails in production while the same code works from a laptop. Setting MASSIVE_PROXY_URL to the full proxy URL including credentials, exactly as Massive gives it (http://user:pass@host:port), routes those calls through a residential IP.

Only the small, IP-blocked calls go through it by default — the crumb handshake, quoteSummary and chart-events lookups in /api/earnings, a few KB each. Keyed APIs (Finnhub, FMP) authenticate by token and don't care about your IP, so proxying them would spend bandwidth for nothing and they're left alone.

MASSIVE_PROXY_BULK=1 additionally routes /api/yahoo and the snapshot's ~500 one-year history pulls through the proxy. That's tens of megabytes per full compute, and a residential proxy bills by the gigabyte — turn it on only if Yahoo starts refusing the snapshot, and only with a Blob store connected so that compute runs once a weekday instead of on every request.

FMP is deliberately not in this list. Verified twice against the live API: per-symbol earnings, income-statement and financial-estimates all answer ACCESS DENIED on this plan, and the earnings-calendar it does serve returned 21 mega-cap names for a 17-day window — names the nightly snapshot already carries. It could never answer for the off-universe holdings this endpoint exists for, so calling it was two guaranteed-failing round trips per request.

Every result is written back to the Blob store, so one success anywhere is permanent. Add &debug=1 to see the status code of every upstream hop (no cookie or crumb value is ever echoed), or &refresh=1 to bypass the cache.

If every source is silent for a listing — which happens, these are free feeds — set the date yourself on the position (Portfolio → add, or Edit position in the stock drawer). It's stored on-device with the rest of the position and shown tagged yours everywhere, never as a confirmed date.

---

Deploy to Vercel (recommended — free, ~5 minutes)

Step 1 — Get your Anthropic API key

1. Go to https://console.anthropic.com
2. Click API Keys → Create Key
3. Copy the key (starts with sk-ant-...)

Step 2 — Install Vercel CLI

npm install -g vercel

Step 3 — Deploy

cd TigerTrade-dashboard
npm install
vercel

Follow the prompts:

• Set up and deploy? Y
• Which scope? Pick your account
• Link to existing project? N
• Project name: canslim-dashboard (or anything)
• Directory: ./ (default)
• Override settings? N

Step 4 — Add your API key to Vercel

vercel env add ANTHROPIC_API_KEY

Paste your key when prompted. Select Production, Preview, and Development.

Step 5 — Redeploy with the env variable

vercel --prod

Your dashboard is now live at https://canslim-dashboard-[yourname].vercel.app 🎉

---

Run locally (for development)

npm install

Create a .env file:

ANTHROPIC_API_KEY=sk-ant-your-key-here

Then run:

npm run dev

Open http://localhost:5173

---

Alternative: Deploy to Netlify

1. Push this folder to a GitHub repo
2. Go to https://netlify.com → Add new site → Import from Git
3. Build command: npm run build
4. Publish directory: dist
5. Add environment variable: ANTHROPIC_API_KEY = your key
6. Deploy

For Netlify, rename api/claude.js to netlify/functions/claude.js and update any /api/claude fetch URL to /.netlify/functions/claude.

---

Project structure

tigertrade-dashboard/
├── api/
│   ├── claude.js            ← Claude proxy (keeps Anthropic key secure)
│   └── fmp.js               ← Live market-data proxy (FMP)
├── src/
│   ├── main.jsx             ← React entry point (imports terminal.css)
│   ├── terminal.css         ← Design tokens + all terminal styling
│   ├── App.jsx              ← Terminal shell: routing, state, drawers, watchlist
│   ├── tt.js                ← Data namespace (events, stats, CANSLIM dataset)
│   ├── liveData.js          ← Live FMP quote fetch + merge
│   ├── components.jsx       ← Top bar, hero, stat strip, radar view, watch ctx
│   ├── radarScope.jsx       ← Polar event radar scope
│   ├── charts.jsx           ← SVG chart primitives
│   ├── views.jsx            ← Calendar / Timeline / Playbook
│   ├── commandPalette.jsx   ← ⌘K command palette
│   ├── drawer.jsx           ← Drawer shell + event / stock / watchlist bodies
│   └── canslim.jsx          ← CANSLIM screener / market health / playbook
├── index.html
├── package.json
├── vercel.json
└── vite.config.js

---

Features

• TigerTrade Terminal — dark Obsidian theme, token-based design system, grain overlay, reduced-motion-aware animations
• Command palette (⌘K / /) — jump to any view, toggle filters, or open any event drawer
• Volatility Radar: next-catalyst card, polar radar scope, expandable event rows, timeline, month-grid calendar, category/weight filters, and an event drawer with historical reaction stats + move distribution
• Leadership Screener: 7-factor (LEADERS) scoring → composite + grade, per-row trend sparklines, live buy-zone detection, RS ranking, market-health cards, and a full stock drawer (price/volume chart, RS line, buy-point analysis)
• Persistent watchlist — star any event or ticker; survives reloads
• Live prices via FMP (price & % change drive the buy zone); honest live/demo state, never fake-as-real
• AI proxy (/api/claude) available for future AI features