coco
7.4 KiB
title, source, author, published, created, description, tags
| title | source | author | published | created | description | tags | ||
|---|---|---|---|---|---|---|---|---|
| How to Build a Claude Powered Polymarket Trading Bot for $28 a Month | https://x.com/2xnmore/status/2047238933218521182 |
|
2026-04-23 | 2026-04-26 | An Anthropic engineer just turned $200 into $13,679 on Polymarket using a Claude powered trading bot that costs $28 a month to run.This is n... |
|
An Anthropic engineer just turned $200 into $13,679 on Polymarket using a Claude powered trading bot that costs $28 a month to run.
This is not a headline. The dashboard is real. The compounding curve is real. And the entire system was built using tools that anyone reading this can access today.
This article gives you the complete breakdown of how it works and the exact steps to build your own version.
What the Bot Actually Does
Most people lose money on prediction markets because they trade on opinion. This bot does not have opinions.
It analyses 86 million historical trades on Polymarket, ranks 14,000 wallets by win rate and actual profit over time, identifies the wallets that consistently beat the market across hundreds of trades, monitors what those wallets are entering in real time, and only places a trade when multiple top wallets agree on the same position within a 30 minute window.
It makes a maximum of 10 trades per day. That discipline is not a limitation. It is the entire edge. Overtrading is how most bots and most humans destroy their accounts.
The system ran for 27 days in the video. It closed with a 74% win rate across 259 trades and turned $200 into $13,679. The infrastructure cost for the entire run was under $30.
Step One: Get the Data
Go to github. com/warproxxx/poly_data.
This is a completely public dataset containing every single trade ever made on Polymarket. 86 million entries. Every wallet address. Every timestamp. Every market. Every outcome. It costs nothing to access, and it is the foundation on which the entire system is built.
Download it and store it locally or on a basic cloud server.
Step Two: Build the Wallet Ranking System
Open Claude and use this prompt:
"I have a dataset of 86 million Polymarket trades. Help me build a Python script that ranks wallets by win rate, total profit, and consistency over the last 90 days. Filter for wallets with a minimum of 50 trades and an above 60% win rate. Output the top 100 wallets with their stats."
Claude writes the complete script. Run it against the dataset. You now have a ranked list of the 100 smartest, most consistent wallets on Polymarket. These are your signal sources.
Step Three: Build the Real Time Monitor
Once you have your top wallet list, you need to watch what they are doing live.
Ask Claude: "Build a script that monitors these top wallets in real time using the Polymarket API and sends me an alert when 2 or more of them enter the same market position within a 30 minute window. Include the market name, current odds, and which wallets triggered the signal."
That alert is your entry signal. You are not copying one wallet. You are waiting for consensus among multiple proven wallets before touching a trade. That filter alone eliminates most bad entries.
Step Four: Build the Exit Logic
Knowing when to exit is more important than knowing when to enter.
Ask Claude: "Add an exit rule to the system that automatically closes the position when it hits 15% profit or when the top wallets tracking this market begin reducing their position size, whichever condition is met first."
Exiting before the whales do is what separates this from simple copy trading and what protects profit on winning positions before the crowd reverses.
Step Five: Build the Dashboard
The Matrix themed dashboard in the video was built entirely inside Claude Code in approximately 20 minutes.
Ask Claude: "Build me a terminal style trading dashboard with live position tracking, a whale wallet monitor panel, a real time trade log, Claude consensus voting display, and a bankroll compounding curve chart. Use green text on black background in Matrix movie styling. Connect it to my Polymarket positions via API."
Claude builds the entire interface. You connect your API keys and it runs live.
Step Six: Deploy It
The complete infrastructure cost is $28 a month.
A basic VPS to run the scripts continuously costs $8 a month. Claude API access costs $20 a month. Polymarket API access is free. The GitHub dataset is free.
That is the entire budget.
The Honest Part
The 74% win rate over 27 days is impressive. Whether it holds over 6 months is a completely different question and you should approach it as such.
Polymarket is a prediction market with relatively thin liquidity compared to financial markets. An edge that works when only a few people are using it gets crowded and compressed the moment everyone sees it. Past performance of top wallets does not guarantee they continue performing. Markets change. Information edges get arbitraged away.
Front running whale wallets sounds like a clean strategy until you realise that other bots may be watching the same wallets, slippage increases as more capital chases the same signals, and Polymarket can adjust rules or flag heavy automated activity.
Start with an amount you are fully prepared to lose. Paper trade the system for at least two weeks before putting real money in. Never size positions in a way that a losing streak destroys your account.
This is not financial advice. It is a technical breakdown of what someone built and how you can learn from it.
What This Actually Means
Five years ago building a system like this would have required a quantitative developer, a data engineer, a front end developer, and months of work. The cost would have been six figures minimum.
Claude made it possible for one person with $28 a month and a few hours to build the same thing.
That compression of capability is not a Polymarket story. It is not a trading story. It is the defining story of what AI is doing to the gap between people who build and people who watch others build.
The barrier is gone. The tools are available right now. The only thing left is whether you spend the next hour building or scrolling.
Quick Reference: The Full Prompt Sequence
Prompt 1 for wallet ranking: "I have a dataset of 86 million Polymarket trades. Help me build a Python script that ranks wallets by win rate, total profit, and consistency over the last 90 days. Filter for wallets with minimum 50 trades and above 60% win rate. Output the top 100 wallets."
Prompt 2 for real time monitoring: "Build a script that monitors these top wallets in real time using the Polymarket API and alerts me when 2 or more of them enter the same position within a 30 minute window."
Prompt 3 for exit logic: "Add an exit rule that closes the position at 15% profit or when whale wallets begin reducing their position, whichever comes first."
Prompt 4 for the dashboard: "Build a terminal style trading dashboard with live position tracking, whale monitor, trade log, consensus display, and bankroll curve. Green on black Matrix styling. Connect via Polymarket API."
Total cost: $28 a month. Total setup time: under 3 hours.
NFA DYOR. Start small. Paper trade first. The edge is real but no edge lasts forever without iteration.
$200 became $13,679 in 27 days. The question is not whether Claude can build this.
The question is what you build with it next.