Methodology
How we benchmark
Enso Shield asks every major DeFi aggregator for a quote on the same swap, then simulates each quote to see what the swap would actually produce. The gap between the quote and the simulation is the core measure of quote honesty. This page walks through the test setup, the cycle, and what each metric means.
Overview
What we test
Trade sizes
- Small$500 – $15K
- Medium$15K – $200K
- Whale$200K – $2M
Chains
How a benchmark cycle works
Step 1 - Fetch quotes
Request a quote from every supported aggregator in parallel, with a fixed 1% slippage tolerance. Record the quoted output and the request latency.
Step 2 - Simulate the quote
Re-simulate each quote via Enso Shield immediately after the quote response arrives, against the latest block at that moment - it measures what the quote would have produced if executed as fast as possible. The simulation records the actual amount out and the gap versus what was quoted.
What we measure
Overquote rate
Share of quotes where the simulated output came back lower than what the aggregator promised - the user would have received less than quoted.
Underquote rate
Share of quotes where the simulated output exceeded what was promised. Conservative quoting - the user gets more than expected.
Quote latency
Time from quote request to response, measured from US East.
Simulation failure rate
Share of quotes that fail when re-simulated via Enso Shield. Flags routes that look fine in the API response but wouldn't execute successfully.
Pinned-block overquote
For aggregators that report the block their quote was built against, we re-simulate at that block in addition to the latest one. Separates quote honesty from chain-state drift since the quote - if the pinned-block rate is much lower than the default rate, the gap is drift, not the aggregator.
Data integrity
Trade amounts are randomized per cycle within each tier's band, so no single liquidity level dominates the dataset. Within a cycle, every aggregator is quoted the identical amount per route and tier, so no aggregator is systematically handed a harder or easier trade.
Enso Shield is run by Enso, but no aggregator gets preferential treatment in the test setup. The same routes, sizes, slippage, and simulator are used for everyone.
Simulations that fail with RPC errors are not included in the data.