Optimism Playbook: Execution Guide with Fees & Endpoints

Answer first — A working Optimism MEV operation in 2026 stacks three disciplines together. Fees: every send accounts for both the L2 execution component (normal EIP-1559) and the L1 data component (priced through EIP-4844 blobs since Ecotone) — re-query eth_getL1Fee immediately before signing because blob demand is bursty and the L1 portion can shift 30–50% inside a block. Endpoints: Optimism has no traditional public mempool; reads come from a well-connected op-node (your own or commercial), writes go through a private RPC, and the canonical "fastest available" stack is a self-hosted op-node co-located near us-east-2 paired with a tier-1 commercial RPC. Atomicity: there's no Flashbots-equivalent bundle relay on Optimism, so multi-leg trades pack into a single executor-contract call that reverts on partial fill. Tie all three together with a runbook that gates every send on fresh fees, fork-simulation against current head, and tight slippage caps, and you have a system that doesn't quietly bleed margin during blob-fee storms.

Mastery path

• Arbitrum playbook: private bundles
• Base MEV strategies (2026)
• Optimism MEV: fees, endpoints & execution tips
• Optimism playbook (current)
• Best Optimism WSS endpoints
• Lowest-latency Reth RPC for MEV

Pre-flight: the environment

Before any live capital:

• Endpoint set verified. Tier-1 commercial RPC sub-50 ms p95, plus self-hosted op-node if volume justifies. Run the WSS latency test from your actual deployment region.
• L1 fee querying wired in. Confirm eth_getL1Fee round-trips fast against your write endpoint. Stale L1-fee scalar is the most common silent loss on OP Stack chains.
• Executor contract deployed. Multi-leg routes pack into one call; the contract reverts on partial fill. Without this, two top-level sends can be ordered apart by the sequencer and one leg lands without the other.
• Strategy in simulation mode with three consecutive clean wins on the same route at >5% margin gap before going live.

Fee model that doesn't lie

The Optimism fee model has both an L2 component (priced like normal Ethereum gas) and an L1 component (the cost of posting transaction data to Ethereum mainnet). Since Ecotone the L1 portion is priced through blobs, which are roughly an order of magnitude cheaper than the calldata era — but blob fees are bursty.

The fee discipline:

1. Estimate L2 execution gas via eth_estimateGas against the candidate transaction.
2. Re-query eth_getL1Fee against the exact bytes you're about to sign. Not against an earlier estimate; not against current head; against the actual signed-tx bytes.
3. Total cost = L2 gas × L2 effective gas price + L1 fee.
4. Reject the trade if simulated profit doesn't clear total cost plus your safety margin.

Step 2 is the one operators skip. Skipping it is how trades go quietly negative during blob-demand spikes when multiple OP Stack chains compete for blob space simultaneously.

Atomicity through executor contracts

Optimism has no Flashbots-style bundle relay. There is no eth_sendBundle to submit to. So how do you guarantee multi-leg trades land together?

The pattern is a custom executor contract — a small contract you deploy that:

• Takes the multi-leg route as parameters in a single call.
• Executes both legs sequentially within the contract's execution context.
• Reverts the entire call if any leg fails.

This guarantees atomicity at the EVM level rather than via relay-side coordination. The downside is gas overhead from the contract dispatch — usually small relative to the trade economics, especially on post-Ecotone Optimism where L1 data is cheap.

The FRB Agent deploys an executor contract pattern by default for OP Stack chains.

Runbook: shipping a reliable Optimism trade

1. Verify endpoint health. eth_blockNumber round-trip; reject route if p95 > 100 ms.
2. Subscribe to op-node mempool/gossip via WSS. Confirm steady cadence; alarm on staleness > 500 ms.
3. Detect target tx and compute backrun route against current pool state.
4. Re-query L1 fee via eth_getL1Fee against the candidate signed tx.
5. Simulate against an Optimism fork at current head. Reject unless simulated profit clears (L2 gas + L1 data fee + priority fee + safety margin).
6. Submit through your private RPC at competitive priority fee. 2-second blocks; window is short.
7. Confirm within 2 blocks. If not landed, kill — state has moved.
8. Log realised vs simulated PnL plus fee breakdown.

Common Optimism execution mistakes

• Stale L1-fee scalar at sign time. During blob-demand spikes the L2 fee shifts faster than most caches refresh. Under-bid happens silently.
• Sequencer outage masking model error. When inclusion latency balloons, "the strategy isn't working" — but actually the chain is degraded. Pause new sends; don't push through.
• Forgetting the executor contract. Two separate top-level sends can be sequenced apart. If your strategy needs both legs to land together, encode them into one executor call.
• Cross-region endpoint failover quietly halving your edge. us-east → eu-west failover adds ~80 ms each direction. Alarm on latency, not just hard failures.
• Treating L2-final inclusion as L1-final. A confirmed Optimism tx is L2-final, not L1-final. For settlement-sensitive flows, wait for the L1 challenge window or accept fast-bridge trust assumptions.

When Optimism is temporarily uneconomic

• Extreme blob demand. When the L1 portion of an OP Stack fee approaches the L2 execution cost, your margin of safety collapses. Watch blob fees on mainnet alongside Optimism's L1 scalar; if both are spiking, sit out for an hour.
• Sequencer instability. The status page shows current health. During degraded periods, simulation goes stale before the tx lands.
• L1 calldata storms. Pre-Ecotone-era issue, now mostly historic — but during specific events ETH gas can briefly make Optimism economics worse than a quick L1 trade. Re-evaluate.

Reporting cadence for production

• Daily: inclusion rate per route, average L1 fee paid, realised vs simulated gap.
• Weekly: endpoint p95 latency histogram, per-relay (write-side) inclusion attribution if relevant, strategy PnL.
• Monthly: fee-regime review — has L1 base fee shifted enough to recalibrate the model?

References

• Optimism developer documentation: fees
• Ecotone upgrade overview
• Running an op-node