Concept
Execution Evidence
Evidence generated through active and completed execution that reveals how decisions behaved, not just how they finished.
Research Area
Intelligence
Status
Emerging
Version
0.1
Last Updated
July 2026
Related Discipline
Trade Analysis, Behavioral Finance
Implementation
Orqis
Execution Evidence encompasses all evidence generated through active and completed execution — both outcome-oriented (how a decision finished) and path-oriented (how a decision behaved while arriving there). Two strategies with identical return, Sharpe, and win rate can present materially different execution-path profiles: one with shallow adverse excursion and modest favorable excursion, the other with deep adverse excursion and large favorable excursion followed by reversal. Aggregate outcome metrics obscure these differences. Execution Evidence makes them visible.
Precise Definition
Execution Evidence is the broader evidence generated through active and completed execution, encompassing both Performance Evidence (outcome-oriented: return, Sharpe, win rate, drawdown) and Behavioral Evidence (path-oriented: MAE distribution, MFE distribution, holding-duration patterns, recovery characteristics, capture efficiency, exit-mechanism distribution). The distinction is analytical — some measures span both categories.
Key Properties
- Conditional distributions — winner-only, loser-only, and all-trade distributions answer different questions. Mixing them produces misleading summary statistics.
- Cohort dependence — every contextual statement must specify its comparison cohort. The cohort definition and its conditioning materially change the interpretation.
- Descriptive, not predictive — execution evidence contextualizes active decisions without implying outcome forecasts. 'Rare among historical winners' is not 'likely to lose.'
- Trust-gated — path metrics require sufficient candle coverage (≥80%) to be reliable. Untrusted path data is excluded or flagged.
- Family-differentiated — different strategy families exhibit distinct execution-path profiles by design. DCA strategies have deep MAE by thesis; scalping strategies have shallow MAE. No universal 'good' or 'bad' profile exists.
Why It Matters
Most evaluation systems describe how a decision finished. Execution Evidence describes how it behaved while arriving there. This distinction matters because two strategies with similar headline performance can impose fundamentally different capital experiences — and those experiences determine whether a strategy is sustainable, allocable, and trustworthy under stress.
Boundary
What this concept is not
Execution Evidence is not a prediction system. It does not forecast trade outcomes from path similarity. It is not a single-number health score — composite normality scores risk false precision. It is not independent of its comparison cohort — the same drawdown can be 'typical' among all trades and 'rare' among winners.
Examples
Conditional MAE distributions: among 104 historical winning DOT momentum trades on the 4-hour timeframe, the median MAE was −0.70% — most winning trades were never more than 0.70% underwater. Among 33 losing trades in the same cohort, the average MAE was −14.63%.
Execution Context: comparing an active position's current drawdown against the historical winning-trade MAE distribution for the same strategy version, producing rarity labels (typical, uncommon, rare but observed, outside historical range) with cohort disclosure.
Loser path profile: losing trades in the observed cohort often experienced greater favorable excursion than winners before ultimately closing at a loss — average MFE +8.66% for losers vs +2.40% for winners. A candidate behavioral signature suggesting profit-first-then-collapse failure patterns.
Conceptual Neighbors
Requires
- Evidence Systems
- Exit Instrumentation
Enables
- Behavioral Signatures
- Execution Context
- Decision Quality assessment
Produces
- Path-oriented evidence
- Conditional distributions
- Rarity context
Opposes
- Aggregate-only evaluation
- Outcome-as-prediction
Current Research Questions
- ?Are behavioral signatures derived from path evidence stable over time within a single strategy version?
- ?Does live execution-path behavior diverge from paper trading path behavior in systematic ways?
- ?Can execution-path features predict trade outcomes, or are they purely descriptive?
- ?What minimum sample size produces reliable percentile estimates for contextual statements?
- ?How should the comparison cohort fallback hierarchy balance specificity against sample size?
Related Frameworks
Related Research
Related Papers
This concept is actively implemented in Orqis, Warren Labs' first decision infrastructure platform.