Summary
All About AI shares week-one live results from a Kalshi prediction market trading bot that uses Google DeepMind’s weather forecasting model to bet on New York City temperature buckets. The bot was built and documented in a prior video; this follow-up tracks real-money performance over roughly five days of automated operation.
Actual results shown: a loss on day one due to a position issue, a $12 gain on September 11, a $9 gain the following day, and an unrealized position of approximately $14 — bringing the five-day total to around $34 profit after fees. The creator walks through the bot’s decision logic: predictions are frozen about 15 minutes before market open, and the bot selects the side with the largest “conservative edge” — requiring at least a 5-cent advantage before entering. Entry edges across the week ranged from 7 to 27 cents, averaging around 18 cents.
The video is candid about statistical limitations: five days is far too short to distinguish edge from luck, and the creator explicitly plans a 30-day follow-up if viewer demand warrants it. For those building or monitoring AI-powered prediction market strategies, this provides a transparent early look at real-money performance and the practical realities of running a 24/7 automated weather-market bot on a VPS.
📺 Source: All About AI · Published September 14, 2026
🏷️ Format: Workflow Case Study
