BackTrade — Trading Backtesting Platform

TypeScriptReactNode.jsExpressPrismaZodJWTStripePostgreSQLRedisClickHouseRabbitMQMinIODockerKubernetesNginxCloudflareViteTanStack QueryTurborepopnpmJest

BackTrade is a trading backtesting platform that replays historical markets as if they were live: start a session at any past timestamp, trade in simulated real time and review detailed performance analytics.

Type: SaaS web platform
Role: Sole developer — architecture, development and infrastructure
Timeline: Sep 2025 – Jan 2026
Live: backtrade.damien-reichhart.fr

Overview

Built for traders and quantitative analysts, BackTrade provides a deterministic simulation environment. Each session is configured with an instrument (XAUUSD, EURUSD…), a time range, a starting balance, leverage, and fixed spread, slippage and commission models — then played back candle by candle, exactly as the market unfolded.

Key features

Market replay — play, pause and change speed (0.5× to 10×) on interactive candlestick charts
Concurrent sessions — run several simulations side by side across instruments and timeframes
Position management — market orders sized in lots, take profit and stop loss with live edits, close one or all positions with real-time P&L
Performance analytics — detailed metrics for every session
Subscription tiers — User, Trader and Expert plans with role-based session limits and Stripe billing
Dataset pipeline — admins upload OHLCV CSV files that are split, validated and ingested asynchronously

Architecture

A TypeScript monorepo (pnpm workspaces + Turborepo) split into independently deployable services that share typed packages for data access, validation, caching, queueing, object storage, mailing and logging.

Service	Responsibility
Web	React + Vite single-page app served by Nginx
API	Express REST API — authentication, sessions, trading, billing
Worker	RabbitMQ consumers that split and process uploaded datasets
Scheduler	Scheduled tasks and failed-job retries
PostgreSQL	Core relational data, accessed through Prisma
ClickHouse	Historical candlestick time series
Redis	Caching layer
MinIO	S3-compatible storage for dataset files
RabbitMQ	Message broker between the API and workers, with retries and a dead-letter queue

Engineering highlights

Security — JWT access and refresh tokens, Argon2 password hashing, Helmet headers, role-based authorization and Zod validation on every request
Production infrastructure — Docker Compose stack with non-root containers, dropped capabilities, health checks, resource limits and three isolated networks, exposed only through a Cloudflare Tunnel; Kubernetes manifests included
Code quality — strict TypeScript, ESLint, Prettier, Jest, pre-commit hooks and Conventional Commits
Documentation — 50+ design documents: use cases, sequence and activity diagrams, data dictionary, API reference, error handling and deployment guides (available below)

Tech stack