Harvard course demos bots

A Harvard-linked algorithmic-trading project showed bots placing trades on Polymarket, turning a classroom-style demo into a live prediction-market example. (github.com, x.com) The code repository tied to “Harvard Algorithmic Trading with AI” says it teaches students to “Research, Backtest, Implement,” and includes separate folders for research, backtesting, and live implementation. GitHub showed 327 stars and 175 forks when it was crawled this week. (github.com) Prediction markets work like yes-or-no contracts priced between $0 and $1, with the price standing in for an implied probability. Polymarket’s documentation says those prices come from a central limit order book, where users trade against each other rather than against the platform. (docs.polymarket.com, polymarket.com) That matters for a bot demo because Polymarket offers an application programming interface, or software hook, that can both read prices and place orders. Its documentation says public endpoints expose order books and prices, while authenticated endpoints handle order placement, cancellations, and other trading operations. (docs.polymarket.com, docs.polymarket.com) Polymarket’s trading system is a hybrid setup: orders are matched offchain, then settled on Polygon, a blockchain network, after users sign them cryptographically. The company says official software libraries are available in TypeScript, Python, and Rust for developers building trading tools. (docs.polymarket.com, docs.polymarket.com) The order-book mechanics are simple enough for a classroom demo but real enough to carry risk. Polymarket says every order is expressed as a limit order, and a “market order” is just a limit order priced aggressively enough to execute immediately against the best quote. (docs.polymarket.com, docs.polymarket.com) The repository itself includes a warning against skipping testing and says past performance “never guarantee[s] future performance.” That framing matches the course pitch: teach automation as a process, then deploy with risk controls and monitoring rather than improvising trades by hand. (github.com) The clip circulated as a small social post rather than a formal Harvard announcement, but it landed on a live venue where outside traders also provide liquidity. That leaves the demo sitting in two worlds at once: a teaching exercise on one side, and an open market with executable prices on the other. (x.com, docs.polymarket.com)