wtf is trading-backtest-kit?

Backtest Kit is a TypeScript engine for testing trading strategies against historical data and then running those same strategies live, across crypto, forex, and decentralized exchanges, spot or futures markets. Its central idea is that the exact code used for backtesting is the same code that runs in live trading, so there is no separate rewrite step where subtle bugs can creep in between testing and going live. The README frames the project around specific, common failure modes that quietly destroy trading bots, and explains how each one is designed out rather than merely warned against. The most detailed example is look-ahead bias, where a backtest accidentally reads a future candle (a unit of price data) and produces impressive but unrepeatable results. Instead of relying on the developer to be careful, the engine tracks the current simulated time automatically through the code's execution context, so there is no timestamp parameter to forget, and it physically prevents any part of a strategy from seeing data past that point, including a candle that is still in the process of forming. The documentation backs these claims with both an explanation and the underlying implementation detail so a reader, or an AI model reading it on their behalf, can verify the behavior rather than take it on faith. There are three ways to start a project: a casual mode where the command-line tool scaffolds a new project and most of the plumbing lives inside a package, a full-control mode that exposes every internal piece as editable files in the user's own repository, and a Docker-based setup meant for continuous live trading with minimal downtime. A minimal strategy is defined by registering an exchange connection, a time frame, and a signal-generating function, then starting a backtest and listening for its events. The project integrates with the ccxt library to pull real market data from exchanges like Binance. Backtest Kit is distributed as an npm package, built with TypeScript, and includes a documentation site, an article series, and a reference implementation repository. The full README is longer than what was shown.