A Profitable Price Excursion Is Different from a Profitable Trade

In our historical fleet, 43.2% of closed positions reached at least +300 basis points of maximum favorable excursion within 24 hours. Among those positions, 49.3% still closed with a negative trade return. Price moving favorably after entry and an agent keeping that move are distinct findings, and the distinction directs our research toward exits.

The population was 6,400 closed positions, with 2,765 meeting the favorable-excursion threshold. The continuous record describes the pre-alpha fleet through August 15, 2026. Its predominantly paper fills and simplified margin assumptions apply to these results. They provide a historical diagnostic rather than proof of today’s DXAP performance.

Opportunity comes before capture

Maximum favorable excursion records the best favorable move observed within the specified horizon. It is an upper path statistic, not an executable trade receipt. Captured return records what the position actually retained under its observed exit path. Comparing them asks whether an exit mechanism preserved a favorable move, subject to the timing and fill assumptions used.

Only 13.3% of the qualifying positions kept half their excursion. Median capture where upside existed was 2.0%. A ratio can become unstable when its denominator is tiny, which is why a declared eligibility threshold is useful. Here, the +300-basis-point threshold gives the reported cohort a meaningful favorable move before discussing how much was retained.

The 2,765 qualifying positions also leave 3,635 below the threshold, by subtraction from the stated population. An exit study restricted to the qualifying group would miss the effect of an intervention on those remaining positions. We therefore need both the conditional capture result and an all-position return comparison.

Separate two questions in the replay

Our proposed analysis asks first how often a favorable excursion exists, then how an exit rule performs across the entire entry population. The first is descriptive. The second can compare paired paths using the same entries, timestamps and available price data.

A stop and target replay should state whether intrabar ordering is known, what price triggers the exit and what fill assumptions apply. A favorable peak that occurs before an order could have reached the venue supplies weak evidence of attainable return. An exit theory earns more credibility when the replay preserves the temporal sequence that a real order would face.

The record’s fixed 2%/4% stop/target replay is relevant, but it is a separate intervention result with its own uncertainty. It should stay separate from a claim that the model predicted the favorable movement. The paper reports a directional-edge null across the measured fleet groups.

An exit problem worth building around

The first controls paper gives the trace discipline needed to connect proposed exits, policy approval and settlement. DXAP publicly describes recorded decisions and external policy checks. Those capabilities make exit behavior a concrete question for product evaluation, rather than a reason to infer superior returns from architecture alone. We would use the capture gap to prioritize an executable protection study, then require deployed fill evidence before claiming that the gap has narrowed.