TraderBot
A chat-first research tool for markets, with backtesting and paper trading
What it is
TraderBot is built around a conversation. The main surface is a chat: you ask about an instrument, a strategy or a position, and the answer arrives as streamed text with the working shown rather than as a number on a dashboard. Conversations are kept, so a line of enquiry can be picked up later.
Behind the chat are three things worth building. A backtest runner sits behind a port, so a strategy can be replayed over historical data and read back as an equity curve and a trade log. A set of model services produces signals from market data, news and fundamentals. And a broker integration carries an explicit trading mode, so an account is a paper account until somebody deliberately makes it something else.
The risk machinery is the part that took the longest. A daily loss tracker and a circuit breaker sit in front of the broker adapter rather than inside it, so a failing or erratic upstream cannot be retried into a hole. Anything raised in the chat passes a discipline check that can refuse it outright, and the refusal is shown to the user instead of being logged quietly.
What it is not: advice. TraderBot does not tell anyone what to buy, it makes no claim about returns, and nothing in it is a recommendation. It is a tool for research, for testing a strategy against history, and for running that strategy against a paper account to see how it behaves.