Can AI replace TradingWizard?

Build me a local personal market-analysis bot inspired by TradingWizard. Requirements:

- Node 22, TypeScript, Next.js App Router, better-sqlite3, technicalindicators,
  and node-cron. Bind to localhost only.
- watchlist.json holds up to 10 stock or crypto symbols plus a 15m, 1h, or 1d
  timeframe. Fetch candles from Twelve Data with the key in .env.local.
- Cache responses, respect provider limits, and show source plus last-updated time.
  Stale or missing candles must block a new setup.
- Calculate EMA 20/50, RSI 14, ATR 14, recent support and resistance, and volume
  change. Keep every setup rule in one documented rules.ts file.
- For each symbol return WAIT, LONG SETUP, or SHORT SETUP with entry zone, stop,
  target, invalidation, confidence, and the exact rule hits behind the result.
- Use an OpenAI-compatible API only to turn the calculated facts into two plain
  sentences. The model cannot change levels or invent facts. Work without it too.
- Add a paper journal. Confirm each simulated entry manually, size by a fixed risk
  percentage, and update open trades from fresh candles. Never connect a broker.
- Dashboard: watchlist state, latest setup, stale-data warning, open paper trades,
  closed results, and a per-symbol timeline. No profit promises.
- SQLite tables: candles, analyses, paper_trades, and scan_runs. Keep immutable scan
  facts so every result can be explained later.
- Add npm run scan and a configurable cron. One failed symbol must not stop the rest.
- No accounts, cloud sync, telemetry, payments, live orders, or investment advice.
  Do not scrape consumer websites.
- Include seeded fixtures and tests for stale-data blocking, rule calculations, risk
  sizing, manual confirmation, and duplicate-scan idempotency.
- README with setup, API limits, backup/restore, cron examples, and the exact limits
  of a personal build compared with a maintained multi-market service.