Performance Tuning

RPC

• Location matters: Keep latency under 10ms between your machine and the RPC
• Separate read and send RPCs: Use a fast RPC for reading and a different one optimized for sending
• High-quality RPCs are the single most impactful investment

Read Tuning

• PROCESS_DELAY: Lower values = more frequent quotes = more opportunities found, but higher CPU usage
• Balance based on your machine capabilities

Send Tuning

• Skip preflight: Disable preflight simulation for faster submission
• Multiple RPCs: Use several sending RPCs for redundancy
• Compute unit price: Higher = better block position but more cost

Jito Tuning

• Tip basis points: Balance between landing rate and cost
• Multiple block engines: Use all available regions
• Rate limits: Jito has rate limits per IP — use multiple IPs if needed

Jupiter API

• Yellowstone: Use gRPC streaming for real-time market data
• Direct routes: Enable for faster quoting
• Max accounts: Higher = more routes explored but slower

Intermediary Mints

• Focus on high-volume tokens
• Use Circular for automatic discovery
• Remove underperforming mints regularly

Bot Settings

• Quote time targets: Keep under 2 seconds for competitive advantage
• Simulation: Disable in production (adds latency)
• Auto restart: Enable to recover from crashes
• Trade size distribution: Test different sizes to find optimal ranges

Multiple Locations

Run bot instances in different geographic regions for better coverage of the validator network.