How FRB orchestrates execution

Execution is more than “send bundle, hope for the best.” FRB keeps a live state machine of every relay, PGA lane, and refund guard so you know exactly where capital flows. Use this hub as a systems guide: private bundles, public auctions, telemetry, and rollback procedures belong in one place so the desk operates predictably.

Private bundle workflow

Private bundles are the default lane for FRB. They keep payloads sealed, respect refund policies, and provide deterministic inclusion windows. A standard workflow looks like:

1. Choose relays with clean health metrics from Relay Status.
2. Run the Latency Test and tag endpoints by region.
3. Feed bundles through FRB’s policy engine so slippage, gas, and session budgets stay in sync.
4. Stream results into Ops Pulse + telemetry for day-two audits.

If anything drifts—variance spikes, refund ratio climbs—document the cause and pause gracefully. The Flashbots vs Public PGA guide outlines how to choose new relays or escalate to support.

Public PGA fallback

Public PGAs still matter. They let you verify pricing, send tiny canary trades, and keep routes alive when relays degrade. But they must be fenced in:

• Size < 10% of your private bundle allocation unless approved.
• Attach extra logging (envelope, rejection reason, broker) for every attempt.
• Expire public attempts after a handful of blocks—never chase indefinitely.

When FRB flips to public mode, the UI surfaces a banner, and Ops Pulse alerts your on-call operator. Close the loop by noting the event in the Knowledge Base.

Telemetry-first culture

Reliable execution depends on measurement. Export metrics into Prometheus, Datadog, or your preferred stack, but also keep the human-readable summaries: Ops Pulse snapshots, refund reports, and the metrics dashboard form the backbone of every retro. When you debrief a session, ask: Which chains performed best? Did public fallbacks add or subtract value? Were there budget overruns?

Chain-focused guidance

Each chain surfaces unique quirks—Arbitrum throttling, Base sequencer bursts, BNB liquidity hygiene. Link this hub to the chain guides so your team can jump from high-level concepts to actionable checklists:

• Ethereum MEV for Flashbots-first routing.
• Base MEV for Coinbase sequencer nuances.
• BNB Chain MEV for router hygiene and tiering.
• Arbitrum MEV for bursty latency management.

Encourage teammates to add notes after every incident—execution quality improves fastest when knowledge is shared.

Bundle anatomy: what actually goes on the wire

A "private bundle" is more than a flag. It is a structured payload that hints at what the builder should preserve, what to revert on, and where to refund. The minimum viable structure for an Ethereum bundle in 2026 looks roughly like this:

{
  "txs": ["0x02...signed-search-tx-1...", "0x02...signed-search-tx-2..."],
  "blockNumber": "0x12a4b3c",      // target block (current + 1)
  "minTimestamp": 0,
  "maxTimestamp": 0,
  "revertingTxHashes": [],         // none allowed to revert
  "replacementUuid": "...",        // for idempotent retries
  "builders": ["flashbots", "titan", "beaverbuild", "rsync-builder", "buildernet"],
  "refundRecipient": "0xYourPayoutAddress",
  "refundPercent": 90              // for MEV-Share orderflow
}

The fields that operators routinely get wrong: revertingTxHashes (must include any tx you expect to soft-fail), builders array (single-builder submission cuts your inclusion probability by 60–80% vs multi-builder fan-out), andrefundRecipient (forgetting this leaks MEV-Share rebates back to the user instead of capturing them). See Flashbots Bundles Explained for the full structural walkthrough.

2026 builder market — who actually wins blocks

The MEV-Boost builder market consolidated through 2025–2026. As of mid-2026, five builders dominate Ethereum mainnet inclusion share:

Submit to all five top builders in parallel as a baseline. Per Best MEV Relays by Chain, single-builder submission costs operators 60–80% inclusion vs full fan-out. The marginal gain from adding builder #6–#15 is small; the marginal gain from monitoring which builder won each landed bundle is large.

Latency budget by chain (target RTT to relay)

Routing latency dominates inclusion probability on fast chains. Use these targets as a ceiling — anything above wastes meaningful share to faster competitors.

See Best Private RPC for Ethereum andZero-Latency RPC Mastery for chain-specific provider recommendations. For benchmarking your own setup, use the in-browserWSS Latency Test.

Failure mode catalog

The eight failure patterns that cause >90% of unexpected misses. Each maps to a diagnostic signature in your logs:

1. Simulation drift — bundle simulates clean but reverts on-chain. Usually fee-staleness or state changed in a competing transaction.
2. Stale fee — gas price assumption already obsolete by submission time. Symptom: insufficient_max_fee_per_gas.
3. Hint mismatch — MEV-Share hints don't match the actual mempool order. Symptom: bundle accepted but not included.
4. Builder rejection — single-builder submission to a builder that didn't win the slot. Mitigation: multi-builder fan-out.
5. Revert tolerance — revertingTxHashes didn't allow a tx that soft-failed. Mitigation: explicit revert allowlist.
6. Sequencer rate limit — L2 sequencer throttled your IP. Mitigation: distribute submission across multiple RPCs.
7. Refund recipient missing — captured the trade but the refund went to the user. Mitigation: always set refundRecipient.
8. Replacement collision — two competing bundles with the same replacementUuid from your fleet. Mitigation: unique UUIDs per submission.

The full diagnostic flow with log signatures and remediation steps is in Private Bundle Debugging: Common Failures and Fixing Failed Bundles Guide.

Intent-flow execution: when bundles aren't the answer

Roughly 25–40% of EVM DEX flow in 2026 goes through intent networks (UniswapX, CoW Protocol, Across, 1inch Fusion) rather than direct DEX swaps. Intent flow is captured by solver auctions, not mempool searchers. For execution-layer operators this means:

• Direct DEX MEV — your existing private-bundle workflow handles this. Half of flow remains here.
• Intent flow — requires being a solver or a filler. Different infrastructure (off-chain auction participation, multi-chain inventory). Most retail searchers can't viably onboard as solvers.
• Hybrid — professional firms operate both. Solo operators typically focus on direct DEX MEV and ignore intent flow.

See Intent-Based Trading vs MEV 2026 for the searcher-to-solver migration framework and DEX Aggregator MEV 2026 for per-aggregator MEV leakage characteristics.