Trade$im

TradeSim AI is an innovative platform designed to assist retail traders and finance students in testing their trading strategies using historical data without the risk of losing real capital. The subscription-based service, priced at $10 per month, simplifies complex backtesting software by integrating traditional technical indicators such as Simple Moving Averages (SMA) and Relative Strength Index (RSI) with an open-weight language model acting as the decision-making layer.

This unique feature allows users to craft and test their trading logic through natural language commands.

At its core, TradeSim operates as a lightweight paper trading engine, constantly pulling in market data to compute indicators and prompting the language model for trading decisions—whether to BUY, SELL, or HOLD. The system then simulates virtual trades, tracks portfolio value in real-time, and provides users with an intuitive dashboard to monitor performance. The platform's architecture is modular, ensuring seamless integration with newer language models as they become available.

The underlying code, developed in Python, leverages the Qwen 2.5 7B model, running locally through Ollama, to handle the LLM decision-making process. This approach not only reduces dependency on costly third-party APIs but also ensures that users' proprietary trading strategies remain private and processed offline. The founder, Anupam Jain, emphasizes the importance of open innovation for side hustles, highlighting how using open-weight models enables unlimited backtesting for a flat monthly fee while maintaining user privacy.

One of the platform's earliest users, a friend inspired by the prototype, successfully backtested a momentum-based crypto strategy. Impressed by the simulated portfolio's 12% growth within a single test month, he became the first paying subscriber, opting for the $10/month tier to access extended historical datasets. This real-world application underscores TradeSim's potential to democratize access to sophisticated trading tools, making advanced strategy testing accessible to a broader audience.