Benjamin-Cup
New member
Hi everyone,
Over the past few months I’ve been working on several automated trading systems for prediction markets, mainly focused on crypto markets on Polymarket. I thought I’d share some of the approaches I’ve been testing and see if anyone here has been exploring similar strategies.
Prediction markets are quite interesting because they behave differently from traditional exchanges. Liquidity can be uneven, spreads can widen quickly, and pricing inefficiencies sometimes appear—especially in shorter-duration markets.
Most of my work has focused on automating strategies for short-term crypto prediction markets, such as BTC, ETH, SOL, and XRP across different intervals (5 minutes to 1 day).
Below are a few of the systems I’ve been experimenting with.
OpenSource : https://github.com/Gabagool2-2/polymarket-trading-bot-python
The system connects to the orderbook through WebSocket streams and tracks price movement over the course of the epoch. By comparing early price behavior with later price movement and reference prices from major exchanges, the bot generates potential UP or DOWN signals.
The idea is to identify moments when the market probability and the external reference price diverge slightly near the end of the market cycle.
Some variations I’ve tested include:
• Momentum entries when prices move strongly in one direction
• Contrarian entries when prices spike excessively
• Optional risk controls such as stop-loss or hedging
Because the markets are short, the system can run multiple cycles throughout the day across several assets simultaneously.
Instead of predicting outcomes directly, this bot monitors activity from selected trader wallets and automatically replicates their trades.
The system architecture looks roughly like this:
Trader Activity → Monitoring Service → Database → Execution Engine → Follower Wallets
Key features include:
• Copy trades based on percentage of the original order
• Fixed USD copy sizing
• Support for multiple follower wallets
• Preview / dry-run mode for testing
• Built-in PnL tracking
This approach is interesting because it allows traders to scale strategies used by experienced wallets while still maintaining full control of their funds and private keys.
In prediction markets, there are moments where both outcomes can be sold at prices whose combined value exceeds $1.
Example:
YES sold at 0.54
NO sold at 0.49
Total received = 1.03
Since creating the YES/NO pair costs $1, the spread becomes the profit before fees.
After the market resolves:
• the winning token redeems to $1
• the losing token becomes $0
The bot automatically redeems the winning token and repeats the process.
This type of strategy behaves more like automated liquidity provision, capturing inefficiencies in the orderbook rather than trying to predict the market direction.
• BTC
• ETH
• SOL
• XRP
Across multiple durations:
• 5 minutes
• 15 minutes
• 1 hour
• 4 hours
• 1 day
The infrastructure is built mainly with:
• Python and TypeScript
• WebSocket orderbook streaming
• Polymarket CLOB API
• Gamma API for market discovery
• Builder Relayer for gasless execution
Email: [email protected]
Telegram : https://t.me/BenjaminCup
X: https://x.com/benjaminccup
Over the past few months I’ve been working on several automated trading systems for prediction markets, mainly focused on crypto markets on Polymarket. I thought I’d share some of the approaches I’ve been testing and see if anyone here has been exploring similar strategies.
Prediction markets are quite interesting because they behave differently from traditional exchanges. Liquidity can be uneven, spreads can widen quickly, and pricing inefficiencies sometimes appear—especially in shorter-duration markets.
Most of my work has focused on automating strategies for short-term crypto prediction markets, such as BTC, ETH, SOL, and XRP across different intervals (5 minutes to 1 day).
Below are a few of the systems I’ve been experimenting with.
OpenSource : https://github.com/Gabagool2-2/polymarket-trading-bot-python
1. End-Cycle Sniper Strategy (5-Minute Markets)
One of the bots I built monitors 5-minute markets in real time and looks for trading opportunities near the end of each cycle.The system connects to the orderbook through WebSocket streams and tracks price movement over the course of the epoch. By comparing early price behavior with later price movement and reference prices from major exchanges, the bot generates potential UP or DOWN signals.
The idea is to identify moments when the market probability and the external reference price diverge slightly near the end of the market cycle.
Some variations I’ve tested include:
• Momentum entries when prices move strongly in one direction
• Contrarian entries when prices spike excessively
• Optional risk controls such as stop-loss or hedging
Because the markets are short, the system can run multiple cycles throughout the day across several assets simultaneously.
2. Copy Trading Bot
Another system I built focuses on copy trading.Instead of predicting outcomes directly, this bot monitors activity from selected trader wallets and automatically replicates their trades.
The system architecture looks roughly like this:
Trader Activity → Monitoring Service → Database → Execution Engine → Follower Wallets
Key features include:
• Copy trades based on percentage of the original order
• Fixed USD copy sizing
• Support for multiple follower wallets
• Preview / dry-run mode for testing
• Built-in PnL tracking
This approach is interesting because it allows traders to scale strategies used by experienced wallets while still maintaining full control of their funds and private keys.
3. Arbitrage / Ladder Trading Bot
The third strategy is less about predicting outcomes and more about capturing spread between YES and NO tokens.In prediction markets, there are moments where both outcomes can be sold at prices whose combined value exceeds $1.
Example:
YES sold at 0.54
NO sold at 0.49
Total received = 1.03
Since creating the YES/NO pair costs $1, the spread becomes the profit before fees.
After the market resolves:
• the winning token redeems to $1
• the losing token becomes $0
The bot automatically redeems the winning token and repeats the process.
This type of strategy behaves more like automated liquidity provision, capturing inefficiencies in the orderbook rather than trying to predict the market direction.
Markets and Infrastructure
The bots currently support crypto prediction markets such as:• BTC
• ETH
• SOL
• XRP
Across multiple durations:
• 5 minutes
• 15 minutes
• 1 hour
• 4 hours
• 1 day
The infrastructure is built mainly with:
• Python and TypeScript
• WebSocket orderbook streaming
• Polymarket CLOB API
• Gamma API for market discovery
• Builder Relayer for gasless execution
Contact
If anyone would like to discuss strategies, automation, or see the bots running live, feel free to reach out.Email: [email protected]
Telegram : https://t.me/BenjaminCup
X: https://x.com/benjaminccup