ThetaMind

Professional US Stock Option Analysis — Five Specialists. Deep Research. One Report.

ThetaMind is a production-grade option strategy analysis platform that delivers desk-style research in one click. Build strategies with real option chains and fundamentals, run a multi-agent AI pipeline powered by Google Gemini 3.0 Pro, and get a single long-form report with full audit trail—analysis and research only; we do not execute trades.

• Strategy Lab — Pick symbol and expiration; option chain (Tiger) and fundamentals (FMP) load automatically. Build iron condors, straddles, strangles, spreads, or custom combos. Real-time Greeks, IV, max loss, break-evens.
• Multi-Agent + Deep Research — Five specialists run in sequence (Greeks → IV → Market → Risk → Synthesis); then Deep Research answers four questions via live Google search and folds them into one long-form section. One report, not five fragments.
• Reports & Audit Trail — Every report is stored with full task history: stage-by-stage outputs, exact prompt, and final memo. Revisit, compare runs, or export.
• Real-Time Charts — Candlesticks (TradingView-style), interactive payoff diagrams, Greeks curves. Pro: 5s refresh; Free: delayed (15 min cache).
• Risk & Greeks — Per-leg and portfolio max loss, max gain, break-evens; delta, gamma, theta, vega. AI explains the numbers in plain language.
• Nano Banana — AI-generated option strategy images (payoff-style diagrams, strategy layout) for sharing or presentation.

Data

• Tiger Brokers OpenAPI — Real option chains (bids, asks, volume, OI, computed Greeks). Same numbers you’d see in a broker.
• FMP (Financial Modeling Prep) — Fundamentals, earnings, ratios, debt, profitability. AI references real financials in the report.
• Google Search (Gemini) — Deep Research uses live web search for grounded, up-to-date synthesis.

Backend

• FastAPI (async), PostgreSQL (SQLAlchemy async + Alembic), Redis (cache + rate limit).
• Resilience: circuit breakers for Tiger/Gemini; tenacity retries; stale cache fallback so the app stays up when external APIs fail.
• Auth: Google OAuth2 + JWT. Payments: Lemon Squeezy (checkout + webhooks, idempotent, audit-logged).
• AI: Google Gemini 3.0 Pro via a provider abstraction (swap to DeepSeek/Qwen if needed). Context filtering (e.g. ±15% strike) to keep token usage and noise under control.

Frontend

• React 18, Vite, TypeScript. Shadcn/UI + Tailwind CSS.
• React Query for server state; lightweight-charts (candlesticks), recharts (payoff/area).
• All times UTC in backend; US/Eastern on the UI.

Deployment

• Docker Compose: PostgreSQL, Redis, FastAPI backend, React (static), Nginx gateway. Health checks, env-based config, single-command run.

User (Browser)
       │
       ▼ HTTPS
   Nginx Gateway
       │
       ├── /api  ──► FastAPI Backend ──► PostgreSQL, Redis
       │                  │
       │                  ├── Auth (Google OAuth2)
       │                  ├── Market Data (Tiger + cache; circuit breaker)
       │                  ├── Strategy Engine
       │                  ├── AI Service (Gemini 3.0 Pro, multi-agent + Deep Research)
       │                  ├── Payment (Lemon Squeezy)
       │                  └── Scheduler (daily picks, quota reset)
       │
       └── /     ──► React Frontend (static)

Report pipeline

1. Strategy Lab — User builds strategy (symbol, expiration, legs). Option chain (Tiger) + fundamentals (FMP) loaded automatically.
2. Five specialist agents — Greeks → IV → Market → Risk → Synthesis. Each writes a focused memo; results feed the next stage.
3. Deep Research — Four questions, live Google search; answers synthesized into one long-form section.
4. Report & task history — Stored with full prompt, stages, and audit trail. Replay any run.

End users

1. Sign in with Google (or try the Demo).
2. Open Strategy Lab: pick symbol, expiration, add legs. Chain and fundamentals load automatically.
3. Click Deep Research (or equivalent “one-click report”). Wait for the pipeline (agents + Deep Research).
4. Read the report, check payoff charts and Greeks. Export or revisit via task history. We never execute — you trade (or not) in your broker.

Developers (local)

• Prerequisites: Docker & Docker Compose (or local Python 3.11+, Node 18+, PostgreSQL, Redis).
• Copy .env.example to .env and set at least: DB_PASSWORD, GOOGLE_CLIENT_ID, VITE_GOOGLE_CLIENT_ID, backend AI keys (e.g. Gemini/ZenMux), optional Tiger/FMP/Lemon Squeezy.
• Run stack: docker-compose up -d (db, redis, backend, frontend, nginx). Backend: http://localhost:5300, Frontend: http://localhost:3000 (or via nginx on 80/443).
• Backend only: cd backend && pip install -r requirements.txt (or poetry), set DATABASE_URL and REDIS_URL, then uvicorn app.main:app --reload.
• Frontend only: cd frontend && npm install && npm run dev.

Quota (daily, UTC reset)

• Free: 5 units/day (1 Deep Research run; 1 run = 5 units).
• Pro Monthly: 40 units/day (8 runs).
• Pro Yearly: 100 units/day (20 runs). See in-app Pricing for details and to upgrade.

ThetaMind/
├── backend/          # FastAPI app (Python)
│   ├── app/
│   │   ├── api/      # Routes, deps, schemas
│   │   ├── core/     # Config, security
│   │   ├── db/       # SQLAlchemy, session
│   │   ├── services/ # Agents, AI, market, payment, storage, etc.
│   │   └── main.py
│   ├── alembic/      # Migrations
│   └── Dockerfile
├── frontend/         # React app (Vite + TypeScript)
│   ├── src/
│   │   ├── components/
│   │   ├── features/ # Auth, etc.
│   │   ├── pages/    # Dashboard, Strategy Lab, Pricing, etc.
│   │   └── services/
│   └── Dockerfile
├── nginx/            # Gateway config
├── docker-compose.yml
├── spec/             # PRD, tech spec, design docs
└── README.md         # This file

• No execution — ThetaMind does not execute trades, place orders, or hold customer funds.
• Secrets — API keys and PII are not logged; webhook signatures (e.g. Lemon Squeezy) are verified.
• Data — Backend stores and processes times in UTC; frontend displays in US/Eastern.

ThetaMind is for analysis and educational use only. It is not a trading platform or investment advisor. Options involve substantial risk; past results do not guarantee future outcomes. Always do your own due diligence and consult a qualified financial advisor before any investment decision.

ThetaMind — Option analysis that thinks like a research desk.