Hiberius

Polymarket Bot Graveyard — Development skill for Claude Code

Development community

An honest, step-by-step post-mortem of six autonomous Polymarket trading bots built and killed over three months.

How to install Polymarket Bot Graveyard

This entry records only its repository, not the path inside it, so there is no exact command to give. Open Hiberius/polymarket-bot-graveyard and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Polymarket Bot Graveyard does

An honest, step-by-step post-mortem of six autonomous Polymarket trading bots built and killed over three months. Zero edge found — and why that 'no' was the only real win. Code, autopsy, and lessons. Built with Claude Fable 5.

Alternatives in Development

Pre Mortem — Run pre-mortem risk analysis on PRDs and launch plans 7.8k ★
Eval Skill — /eval-skill — Six-Dimension Evaluation 216 ★
Windows To Wsl2 Screenshots — Automate Windows screenshots for Claude Code workflows in WSL2. Take screenshots (Win+Shift+S, Win+PrintScreen 85 ★

README

🪦 Polymarket Bot Graveyard

Six autonomous trading bots. Three months. Zero edge found. This is the honest, step-by-step post-mortem of every one of them — what I set out to do, how each died, and the one thing that finally worked.

[![CI](https://github.com/Hiberius/polymarket-bot-graveyard/actions/workflows/ci.yml/badge.svg)](https://github.com/Hiberius/polymarket-bot-graveyard/actions/workflows/ci.yml) [![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](LICENSE) [![Docs: CC BY 4.0](https://img.shields.io/badge/Docs-CC%20BY%204.0-lightgrey.svg)](LICENSE-DOCS) [![Built with Claude Fable 5](https://img.shields.io/badge/Built%20with-Claude%20Fable%205-8A2BE2.svg)](https://www.anthropic.com/)

**This is not a "how I made money" repo. It's the opposite.** Between March and June 2026 I built, ran, and killed six generations of automated trading bots on [Polymarket](https://polymarket.com), one after another, each convinced it had found an edge the last one missed. None had. I'm publishing the whole graveyard — code, plans, and a brutally honest autopsy — because a documented failure is worth more than a hidden one, and because the *reasons* these failed are structural and reusable knowledge.

If you're about to build a prediction-market bot: **read this first.** It might save you three months.

TL;DR

6 bots in 3 months (29 Mar → 28 Jun 2026). Zero edge found. Zero profitable live trades. Ever.
Real money lost: roughly $45, all in the first two weeks (v0 + v1), after which the Polymarket wallet was banned for order-spam. Everything after that was paper / shadow only.
Every version dies on the same wall: an assumed edge that, measured honestly, does not exist for a retail trader with a few hundred dollars.
The only thing that ever worked was the shadow-first discipline of the last version — it correctly refused to promote a coin-flip strategy to live trading. The system's job was to say "there's no edge