Test a supervised Kalshi trading agent with a tiny budget

An agent traded a live $20 Kalshi account for nine days while a human watched, and ended with more incentive payouts than trading losses.

Done with Claude Code· Advanced· Oct 2026

What they did
Mirabilis, an autonomous agent running on Claude Code, was given a live Kalshi account funded with $20. It handled trades on its own for nine days and took part in Kalshi’s Liquidity Incentive Program, with a person supervising the process the whole time.
How it went
Over the run, it earned $20.30 in LIP payouts and logged $36.06 in trading losses. A toolkit tied to the program describes rewards per snapshot as size times d to the power of ticks below the reference price, and daily payouts only apply when the price is at least $1.00.
Tips
Keep the budget small and watch every order closely, since the agent can still lose money while chasing incentives. If you use a similar setup, ask for a clear record of positions, fees, and P/L before letting anything execute.

Try it yourself with Claude Code

Act as a careful Kalshi research assistant. Each day, review [market category] markets, rank the best opportunities for a [paper/live] account with [budget amount], and propose up to [number] trades with the price, reasoning, and worst-case loss for each one. Keep a running log of open positions, fees, payouts, and total profit or loss. Before any order is sent, any funds are moved, or any account setting is changed, stop and show me the exact action for approval. After [number of days], summarize the full result, including fees, payouts, and what you would do differently next time.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on x.com ↗

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.