Execution Price Improvement Requires a Frozen Reference
By DX Research Group · · Execution mechanics
Measure favorable execution against a declared pretrade reference. A worked ledger separates price improvement, fees and subsequent market movement.
Price improvement is meaningful only against a reference chosen before execution. We would report side-aware improvement in quote currency alongside its reference timestamp. Hyperliquid describes limit execution at the selected price or better. That defines a constraint, while the analyst still chooses the economic comparator used to judge the execution.
Freeze the comparator before the fill
Illustrative case: an agent buys 4 units against a frozen decision reference of $100. Execution is $99.80 per unit. Gross improvement is 4 × ($100 − $99.80) = $0.80. An illustrative $0.30 execution fee leaves $0.50 favorable difference after fees. If the market subsequently rises to $102, that additional $8.80 of marked position change belongs to holding-period economics relative to the fill, rather than the pretrade execution improvement measure.
Use signed quantities with a declared convention
For buys, improvement is reference minus fill price; for sells, it is fill price minus reference. Apply quantity weights consistently and keep fee currency conversion in a separate ledger. Compare decision reference and arrival reference side by side when diagnosing delay. A $100 decision reference and $99.70 arrival reference make the same $99.80 buy look favorable by $0.20 per unit on one measure and adverse by $0.10 on the other. Both are useful when their meanings remain visible.
Keep the unresolved state visible
Preserve negative improvement values. A buy above the reference has adverse improvement even if the holding eventually profits. Summing signed values gives an execution comparison across the entire eligible population; reporting only favorable fills selects winners. Also retain the reference source for every row so a switch between midpoint and executable quote cannot manufacture a change in performance.
Our execution and reconciliation framework provides the broader mandate-to-outcome trace. The state and memory contract explains how the verified result should enter the next decision. These notes narrow those published methods to one execution boundary; the worked amounts above are illustrative and the test is a proposed local fixture. The source inspected for the venue-specific statements is the official Hyperliquid order types.