jarrodwatts/jev-trader

One AI trade decision every Monad block. Jev on Kuru MON-USDC.

What it solves

jev-trader is an automated trading bot designed for the Monad blockchain. It solves the challenge of capturing the "spread" (the difference between the buy and sell price) by acting as a market maker rather than a taker, placing limit orders that are filled when other traders hit them.

How it works

The bot operates on a high-frequency loop, making a decision every single block (approximately every 300ms).

  1. Market Analysis: It reads the Kuru MON-USDC order book using a single optimized RPC call.
  2. AI Prediction: It uses a TypeSafe Jev model (or a momentum-based mock model) to predict whether the price will move up or down over a specific horizon (default 100 blocks).
  3. Order Execution: Based on the AI's decision, it places a "post-only" limit order one tick inside the current market price. It simultaneously cancels any previous resting orders to ensure only one order is active at a time.
  4. Efficiency: To fit within the 300ms block time, the bot minimizes network round trips, avoiding gas estimation and synchronous transaction calls.

Who it’s for

Quantitative traders and developers interested in high-frequency trading (HFT) on the Monad network using AI-driven price prediction.

Highlights

  • High-Frequency Execution: Optimized to make a decision and place an order within a single block window.
  • AI-Driven: Integrates with the TypeSafe Jev model for real-time buy/sell probabilities.
  • Market Making Strategy: Focuses on earning the spread via post-only limit orders.
  • Dry-Run Mode: Includes a simulation mode that uses real market data but simulates fills without requiring private keys.

Related

  • Project
  • Project
  • Project
  • Project