Execution Strategy: Winning the Jito Auction

Performance is not just about speed, but about mastering the protocol's constraints.

Our execution is a tactical process designed to win the Jito auction. We understand that Jito leaders run parallel auctions in ~50ms ticks, selecting the most profitable, non-conflicting bundle combinations. Our goal is not just to submit a bundle, but to make it the most attractive bid for the Block Engine in every tick.

  1. Adaptive Routing: Our engine instantly determines if the current leader is Jito-enabled. If so, we construct a Jito bundle. If not, we broadcast a raw transaction with a high Priority Fee.

  2. Tip-per-CU Optimization: We are aware that Jito prioritizes bundles based on tip-per-CU efficiency. Every transaction is pre-simulated on our Firedancer node to derive a precise CU estimate, allowing us to bid the most competitive Jito Tip without over-provisioning compute.

  3. Tactical Bundle Construction & Protection:

    • jitodontfront: We include this read-only account in the first transaction of our bundle. This signals the Block Engine to enable sandwich protection and is a standard practice for "clean" MEV.

    • Atomic Tipping: Our Jito Tip instruction is always included in the same transaction as the core MEV logic. This protects us from paying a tip on a failed execution, especially in uncled block scenarios.

  4. Multi-Region Submission: We broadcast our bundles simultaneously to all regional Jito Block Engines (Frankfurt, New York, Tokyo) to minimize geographical latency to any potential leader (approved by jito support).

Last updated