cryptolab·nexus

volatility desk

K7.

K7 · short-dated volatility

backtest 2024–26 · paper since Aug 30 '26 · as of Oct 1 '26

systematic short-dated volatility · paper, published as it settles

$100$300$1k$3k$10kBTC +75%2x adaptive +13439%adaptive+1908%PAPER →+12377%May '24Sep '24Jan '25May '25Sep '25Jan '26May '26Sep '26

the live record, up close — one point per session, from $100 (solid conservative · dashed maker)

$100static +42.5%adaptive +21.4%session 0 · Aug 30 '26session 19

this year, head to head — both books restarted at $100 on jan 1 (linear scale)

$100static +52%adaptive +125%2x +171%
backtest return+8658%
max drawdown-31.9%
btc same era+75%
paper (growth)+42.5%
paper sessions19
win rate76%
adaptive (paper)+21.4%

The backtest replays the identical program on actual exchange prints (2024 → launch) at conservative fills, weekly-bucketed; log scale. The live paper segment chains onto its end — solid marks conservative execution, dashed marks maker — and its points are session-indexed, so calendar position is illustrative, as with UV18's live segment. The violet line is the adaptive variant — the same book, sized (and occasionally reversed) each night by two live signals: how rich the night prices against its own recent weekdays, and against realized moves on those weekdays. Its backtest (+1908% at −24.6% max drawdown) trades boom-year upside for resilience when segments reprice; its live segment runs in parallel from the day the rule went live. The dashed rose line is the same adaptive rule at double sizing — illustrative only, not a deployed book: leverage multiplies outcomes and drawdowns together (−35% max drawdown vs the adaptive book's −25%). Drawdowns shown, nothing smoothed. Outcomes are public; the machinery (and the calendar) is not.