Predict & Profit
Weather + inflation bots for Kalshi. Full source code.
Start new thread
Steve Farmer

Predict & Profit - Automated weather trading bot for Kalshi prediction markets

by•
Predict & Profit is an automated trading bot that uses GFS 31-member ensemble weather model data to find mispriced contracts on Kalshi prediction markets. The bot scores every trade on ensemble confidence, model agreement, and fee efficiency. It only fires when the edge is real. Includes full Python source code, setup guide, and trading strategy documentation. No guessing. No emotion. Just signal vs price.

Add a comment

Replies

Best
Steve Farmer
Hunter
📌
I've been fascinated by prediction markets for a while, specifically the idea that public data could give you a real edge if you knew how to use it. Weather markets on Kalshi caught my attention because the data is free, the signal is measurable, and most traders are just guessing. The GFS ensemble weather model runs 31 forecast simulations every 6 hours. When all 31 agree, the probability is real. When the market hasn't caught up yet, that gap is the edge. So I built a bot to find it automatically. It scans Kalshi temperature markets every 5 minutes, scores each opportunity on ensemble confidence, model agreement, and fee efficiency, and only trades when all three line up. The whole thing runs unattended on my home machine while I do other things. I packaged it up with full source code, setup guide, and strategy docs so anyone with basic Python experience can run it themselves. Would love feedback from anyone in the algo trading or prediction market space. Happy to answer questions about the model, the edge scoring, or the Kalshi API integration.
Steve Farmer
Hunter

Built a second version of Predict & Profit over the last couple of weeks.

The weather bot now runs on a 62-member hybrid ensemble instead of the old 31-member setup. It combines 31 GFS runs with 31 AI-powered AIGEFS runs, and only trades when the models line up with enough confidence. That made the filtering stricter, which is exactly what I wanted.

I also launched an econ bot focused on CPI and PCE contracts. That one uses Cleveland Fed nowcast data plus FRED energy signals to look for mispriced opportunities in Kalshi’s macro markets.

Everything is still built around the same idea: small size, hard data, transparency, and no hype. I show the losses too, not just the wins. The source code package includes the Python bot, setup guide, Kalshi API integration, PostgreSQL logging, and the ebook.

If you’re into prediction markets, algo trading, weather models, or macro data, I’d love your feedback:
predictandprofit.io

Steve Farmer
Hunter

Built a much stronger second version of Predict & Profit.

The weather bot now uses a 62-member hybrid ensemble instead of the old 31-model setup. It combines 31 GFS runs with 31 AI-powered AIGEFS runs, which lets the bot be much stricter about confidence before it ever places a trade.

I also added an econ bot for CPI and PCE markets. That one uses Cleveland Fed nowcast data and FRED energy signals to look for contracts that may be mispriced.

What I care about most is not making the bot busier. It is making it more selective.

Less noise. Better filters. More transparency.

The full package includes the Python source code, Kalshi API integration, PostgreSQL logging, setup docs, and the ebook. Existing buyers also get the v2.0 upgrade path.

Would love feedback from anyone interested in prediction markets, algo trading, weather models, or macro signals.

Steve Farmer
Hunter

Shipped v2.2 this week. Here's what actually changed.

The weather bot had a bug where it was silently rejecting all NO-side edges and only ever placing YES trades. Fixed. The econ bot was treating submitted orders as confirmed fills, which made P&L, budget accounting, and the kill switch all unreliable. Fixed.

Also added 8 new cities to the weather bot (Dallas, Houston, Seattle, San Francisco, Philadelphia, New Orleans, Las Vegas, Minneapolis), rebuilt the dashboard with 6 tabs including a live Kalshi reconciliation tool, and added a self-check script (bot_doctor.py) that validates your entire setup before you go live.

Full changelog at predictandprofit.io. Live results at predictandprofit.io/results.

This is the reliability release. The bots now account for reality more honestly.