All work
AI Trading System · Quant Platform2025In Progress
MDPCMA Pro+.
A research and backtesting platform for systematic and discretionary traders that combines multi-timeframe technical analysis, AI-driven pattern recognition, and historical strategy validation in a single workbench.
/ 01
Key features.
- Multi-strategy backtester with realistic slippage and commission modelling
- AI pattern detection across price action, volume profiles, and order flow
- Walk-forward analysis with Monte Carlo robustness checks
- Real-time data ingestion from broker APIs (MT5, Binance, IBKR)
- Strategy versioning with reproducible runs and equity-curve diffing
/ 02
Architecture.
Event-driven microservices: a market-data ingestion service streams ticks into TimescaleDB, a strategy-runner worker pool executes backtests in parallel, and a results-aggregation service serves the API. Clean separation between data, compute, and presentation lets users scale a single component without rebuilding the platform.
/ 03
Tech stack.
- Frontend
- Next.js 14 · TypeScript · Tailwind · Recharts · D3
- Backend
- Python (FastAPI) · Celery for async backtest jobs
- Database
- PostgreSQL · TimescaleDB for tick-level data
- AI
- PyTorch · transformers for pattern models · scikit-learn for signal screens
- Infrastructure
- Docker · Kubernetes · Redis cache
/ 04
Target users.
Quant retail traders, prop-firm desks, hedge fund analysts, trading educators
/ 05
Unique selling points.
- Built-in AI signal pre-screening cuts hours of manual chart-reading
- First-class reproducibility — every backtest run is versioned and diffable
- Realistic execution model captures slippage that's invisible in mainstream tools
/ 06
Monetization.
Tiered SaaS subscription (Starter / Pro / Enterprise) plus on-demand strategy review services