The Professor

I wanted to build a Polymarket trading bot. I could find plenty of people talking about building bots. Nobody was actually saying HOW they built them. So I built one from scratch. 798 commits in 6 weeks. Here's everything I learned. A thread 🧵

Copy trading isn't dead. It's a work in progress. The problems are real — latency, adverse selection, FAK no-fills. But each one is an engineering problem with a measurable improvement path. At the same time, I started building a second strategy: weather markets. Not as a replacement, but because relying on one approach is fragile. Tomorrow: weather markets — where the edge isn't speed, it's forecasting.

The retooling worked: Selection gates (only copy when conditions are favorable), tighter age limits, depth-aware execution, per-outcome exposure caps, and the flat 5-share sizing to keep risk controlled while I learn. My early sessions were -$25.51 on 708 trades. Recent sessions have been profitable with a solid W/L rate. The next step: keep reducing latency to catch more winning trades, then scale beyond 5 shares once the execution data supports it.

708 source trades. 117 execution attempts. 45 fills. -$25.51 realised P/L. Copy trading on Polymarket has a hard ceiling. Here's the data on exactly where it breaks. 🧵

After night one, I had two feelings: 1. It works. The bot detected trades, evaluated risk, submitted orders, and some filled. The plumbing is real. 2. Something is wrong. The timing numbers don't look good. And the P/L on fills... we need to talk about that. Tomorrow: what broke. The hard ceiling of copy trading, real numbers, and why 24 seconds of latency changes everything.

The dashboard the next morning told a clear story: Every execution attempt had a full evidence trail: - When the source traded - When my bot saw it - When the order was submitted - What price, what size - Whether it filled or not - If not, why (no match, depth, slippage) This evidence trail is how you learn. Without it, you're just hoping.

The first time my bot placed a real trade, I watched the terminal like a heart monitor. After weeks of scanning, analysing, and paper trading — real money was on the line. Here's what going live actually looks like when you've built the whole thing yourself. 🧵