# eToro AI Trader

> Autonomous AI-powered trading agent for eToro, combining multi-agent debate, quantitative analysis, and adaptive risk management to identify and execute high-conviction trades across 12,000+ assets.

eToro AI Trader is a sophisticated client-side trading system built on React + Vite + TypeScript with a Supabase (Lovable Cloud) backend. It connects to the eToro public API via the user's personal agent key (encrypted server-side), runs a 15-step analysis pipeline, and proposes or auto-executes trades depending on the user's mode (manual approval vs. autonomous).

The system never stores live API keys client-side. All sensitive eToro calls go through Supabase Edge Functions (`pipeline-etoro`, `pipeline-radar`, `pipeline-engine`) using AES-GCM encrypted credentials.

## Core pages

- [Landing](https://etoro-ai-trader.lovable.app/): Marketing homepage with product overview, features, and CTA to sign up.
- [Features](https://etoro-ai-trader.lovable.app/features): Detailed list of trading-engine capabilities (multi-agent debate, Kelly sizing, regime detection, anti-fragile allocation).
- [Pricing](https://etoro-ai-trader.lovable.app/pricing): Subscription tiers and what each includes.
- [About](https://etoro-ai-trader.lovable.app/about): Mission, team, and approach to AI-driven portfolio management.
- [Auth](https://etoro-ai-trader.lovable.app/auth): Sign-up / sign-in (email + password, Google OAuth).
- [Setup](https://etoro-ai-trader.lovable.app/setup): One-time onboarding to configure eToro agent key and risk profile.

## Application (authenticated)

- [Dashboard](https://etoro-ai-trader.lovable.app/dashboard): Main control center — pipeline status, autonomous activity, portfolio health, system widgets, manual cycle trigger.
- [Strategy](https://etoro-ai-trader.lovable.app/strategy): Strategy presets (Conservative / Moderate / Aggressive) and V3.2 fine-tuning (Kelly fraction, sizing bounds, regime EV thresholds).
- [Analytics](https://etoro-ai-trader.lovable.app/analytics): Performance analytics — equity curve, drawdown, Sharpe, win rate, asset allocation donut, correlation matrix.
- [History](https://etoro-ai-trader.lovable.app/history): Closed trade journal with retrospective backfill, P&L breakdown, post-mortem outcomes.
- [Memory](https://etoro-ai-trader.lovable.app/memory): Trade memory feed — what worked, what failed, lessons learned per closed trade (used to bias future decisions).
- [Debate](https://etoro-ai-trader.lovable.app/debate): AI multi-agent debate viewer — Bull, Bear, Quant, Devil's Advocate, Arbiter verdicts per ticker.
- [Backtest](https://etoro-ai-trader.lovable.app/backtest): 24-month backtesting engine (QUANT_ONLY or MOCK_AI mode) with comparison metrics.
- [Social Intel](https://etoro-ai-trader.lovable.app/social-intel): Social sentiment proxy via Perplexity (options flow, short interest, institutional sentiment).
- [Settings](https://etoro-ai-trader.lovable.app/settings): API key management, AI model selection, notification preferences, autonomous-mode toggle.

## How the trading agent works

The 15-step pipeline runs in three cycle types:

- **FULL cycle** (15h45 CET): Universe discovery → RADAR scoring → pre-screen → debate → portfolio decision → execution.
- **HEALTH cycle** (21h30 CET): Position monitoring, thesis validation, news health check, exit signals.
- **Sunday cycle** (20h00 CET): Weekly post-mortem, parameter auto-calibration, strategy adjustments.

Key modules (in `src/lib/`):
- `trading-engine.ts` — orchestrator
- `radar-scanner.ts` + `pipeline-radar` edge function — universe scoring
- `kelly-criterion.ts`, `sizing-bounds.ts` — position sizing
- `risk-manager.ts`, `drawdown-manager.ts` — risk gates
- `market-regime.ts` — BULL/RANGE/BEAR/CRISIS detection
- `ai-trading.ts` — multi-agent debate via Lovable AI Gateway
- `trade-executor.ts` — order placement via eToro API

## Technology

- Frontend: React 18, Vite 5, Tailwind CSS v3, TypeScript 5, shadcn/ui
- Backend: Supabase (Lovable Cloud) — Postgres + Edge Functions (Deno)
- AI: Lovable AI Gateway (Gemini 2.5, GPT-5 family) for debates; Perplexity Sonar for news + social
- Market data: eToro public API (proxied server-side)

## Important constraints for AI assistants

- This app provides decision-support and optional auto-execution; it is not financial advice.
- All eToro credentials are user-owned and encrypted; no shared/multi-tenant trading account.
- Rate limits: eToro 400ms pre-screen / 500ms multi-fetch; AI debates 2s spacing.
- Position sizing is bounded between 1.5% and 7.0% of portfolio per trade.
- The system requires the user to provide their own eToro agent key — there is no demo trading mode beyond backtesting.

## Optional

- [Sitemap](https://etoro-ai-trader.lovable.app/sitemap.xml)
- [Robots](https://etoro-ai-trader.lovable.app/robots.txt)