Started with $1,000. Hit $50. Now $600. Polymarket weather-trading bot with AI, posting the real numbers - wins, losses, lessons. No hype. Not financial advice.
So this week I'm trying something new. A team of AI agents goes through every city, one at a time: two analysts, a red-team whose only job is to kill ideas, and a manager who decides. I'll post what they find. This is the part I can't predict. Follow along.
Two weeks ago my weather bot was in profit for the first time since March: lifetime +$153. I almost posted a victory lap.
Today it's -$264. $417 gone in 13 days. Same bot, same cities.
The gap: it now hits 30% of bets at 40c each. Over the summer it hit 45% at 37c.
Zero lottery wins out of 28+ tickets. Worse: the hedge orders were 3 shares, and Polymarket's minimum is 5. Most hedges never went through, so the bets ran unprotected.
That afternoon the bot got a line I still consider the most important in the codebase: an absolute dollar cap per contract that no signal can override.
One bad Padres game cost $14.70. Then, by April 6-7, the scanner found zero opportunities at a 6%+ edge. The gaps everyone writes about close in seconds.
Then came six weeks of worse ideas: sports arbitrage, then crypto bots on 5-minute markets. About $900 gone by May. One post a day on each, starting tomorrow. And weather? It came back.