# Longer-history structural comparison

Public summary of a checked-in historical research report. Not live P&L, not an audit, and not trading approval. No new research was run for this registry.

## Provenance

Reference source revision: eb1ce1e7977e7911425769d1099946e37b1ec65a

Report: round2-long-history-2026-10-20.md

Original report SHA-256: 9d81e2929b4ec768d2735a7a90152adb68024d711b5571b5bd8000a43686fddb

This is a curated public summary, not the raw internal report or an exact reproduction. Internal implementation and operational details are deliberately omitted.

## Period and universe

2024-10-20 → 2026-10-20 UTC end-exclusive; 730 daily bars

BTC, ETH, SOL, XRP, DOGE, BNB, AVAX, LINK

## Cost assumptions

$10,000 initial capital; 11.5 bps per side (4.5 taker + 5 builder + 2 slippage); 1.0× gross baseline. 14/18 bps stress and 0.5× gross sensitivity.

Funding was loaded as exact contiguous hourly observations for funding-dependent candidates; original Velocity Ranker screens book no funding by design. No maker-fill assumption.

## Reported result and verdict

A failed training (−10.59%); B and C failed validation (−22.19% and −23.23%). No candidate selected. Reserved final 2026-04-20 → 2026-10-20 and final stress scenarios/gates were NOT EVALUATED.

Verdict: **NO PASS**.

## Evidence limits

Corrected signed-quantity/cash accounting; long-only keeps cash. Reserved final overlaps previously studied dates and would be reused evidence. Same-close fills are optimistic. Pre-order gross drift reached 1.2159×; the paid close-observed cap is not a continuous intraday guarantee.

All figures are historical report receipts. Code tests do not validate strategy performance. No strategy is described as profitable or Live-ready.