Overview
Solo personal project: a full-stack cryptocurrency algorithmic trading platform covering market data ingestion, strategy backtesting with walk-forward optimization, multi-source sentiment analysis, portfolio tracking, and research tooling. Built to go beyond basic charting into systematic, data-driven trading workflows.
Challenge
Most retail crypto tools offer simple charting and manual order entry. Wanted a platform that could backtest strategies with walk-forward optimization (avoiding naive curve-fitting), incorporate sentiment signals from multiple sources, and provide structured research and journaling tools, all in a single self-hosted application.
Approach
Used MongoDB for flexible storage of heterogeneous trading data (strategies, signals, journal entries, backtests) and Redis for caching frequently accessed market state. Connected to Binance via WebSocket for real-time market data. Built the strategy engine around walk-forward optimization with ensemble strategy composition and position sizing. Added multi-source sentiment analysis pulling from Reddit, X, crypto news feeds, and the Fear & Greed Index. Used KlineCharts v10 for candlestick charting with 15+ technical indicator overlays.
Tech Stack
| Layer | Technology |
|---|---|
| Framework | Next.js 16 |
| UI Library | React 19 |
| Language | TypeScript |
| Database | MongoDB + Mongoose |
| Cache | Redis (ioredis) |
| Auth | NextAuth 5 |
| Charting | KlineCharts v10 |
| 3D | Three.js + React Three Fiber |
| State | Zustand, TanStack React Query |
| UI Components | shadcn/ui |
| Testing | Vitest, Playwright |
| DevOps | Docker, GitHub Actions |
Key Implementation Details
- Walk-forward optimized backtesting engine with ensemble strategies and position sizing across 6 trading profiles (Scalping, Day Trading, Swing, Position, Grid, DCA)
- Multi-source sentiment analysis aggregating Reddit, X, and crypto news feeds alongside the Fear & Greed Index
- Real-time Binance WebSocket market data with 15+ technical indicators (RSI, MACD, Bollinger Bands, Ichimoku, Fibonacci, ADX)
- Portfolio P&L tracking with cost-basis calculations and recurring price alerts
- Trading journal with tagging, notes, and post-trade analytics for strategy refinement
- Research notes and playbooks for documenting market hypotheses
Outcome
Running on personal VPS with Docker. The platform ingests real-time market data via Binance WebSocket, runs backtests with walk-forward optimization, and tracks portfolio performance with cost-basis P&L. Full CI/CD pipeline via GitHub Actions with containerized deployment.