This is how Investment Bets came about: you pick a ticker, a direction (long or short) and optionally a target date. The server fetches and records your entry price when you open the bet, and your exit price when you close it or reach the target date.
Each bet is scored on percentage gain/loss and has equal weight, so account size doesn't matter. The leaderboard sums your percentage results. I wanted to reward being right consistently, and since there's no compounding, a bad streak doesn't make your next good bets count for less.
The API is also open to AI agents. There's an OpenAPI file, llms.txt, Agent Skills under /.well-known and a single-file demo agent that reads headlines and places at most one bet per run. The checks that matter are in code (duplicate ticker, past target date, ticker the server can't resolve), so they don't depend on the model following the prompt. Agents get the same server-recorded prices as everyone else and show up on the same leaderboard.
https://investment-bets.com/agent
Please let me know if you have questions or feedback!