Absorbed-Flush Rule, Fourth Breadth Replication on a Name-Disjoint Bybit Panel with a Same-Date Concurrency Cap: buy a Bybit USDT perp at the close of a down day whose range blew out (>=2x its 60-day return std) but closed at >=0.55 of its range, fixed 3-day clock, 10% per leg, max 2 new entries per date and max 4 open (28 names sharing none with the Bybit/HL/spot panels, 1-DAY, long-only, unlevered)

Original Research Lead hypothesis (not seeded). It is a frozen-rule breadth replication of the absorbed-flush rule (lesson L262). That rule was promoted on iteration 1 on 13 Bybit perps (200 trades, PF 1.72, Sharpe 0.82, DD 10.2%) and on 15 Hyperliquid perps (151 trades, PF 1.70, Sharpe 0.73, DD 9.1%). It ended underpowered on a 24-name Binance SPOT panel (PF 1.44, but DD 22.7% and Sharpe 0.48 with up to 8 concurrent legs). Two things are new here. (1) UNIVERSE: 28 Bybit USDT linear perps that share NO names with any earlier panel. The Bybit panel used ADA/ATOM/AVAX/BCH/DOGE/DOT/ETC/FIL/LINK/LTC/NEAR/SOL/TRX/UNI/XLM/XRP; the HL panel used ARB/BLUR/CRV/DYDX/GMX/LDO/OP/RUNE/SEI/SNX/STX/SUI/TIA/WLD; the spot panel used ALGO/BAT/CHZ/COMP/EGLD/ENJ/EOS/GRT/HBAR/IOTA/KAVA/MANA/MKR/NEO/ONT/QTUM/SAND/SUSHI/THETA/VET/XTZ/YFI/ZIL/ZRX. The new names: AAVE, APE, APT, AXS, ENS, GALA, GMT, 1INCH, CAKE, DASH, BAND, CELO, COTI, CRO, CFX, ANKR, FLOW, FET, KSM, LRC, MASK, ZEC, IMX, INJ, ICX, ROSE, STORJ, SKL. The test asks whether the rule depends on the earlier names. (2) EXPOSURE DESIGN per lesson L267: the trigger fires on market-wide down days (529 qualifying name-days fell on only 152 distinct dates in the pre-study, ~3.5 names per date). So at most 2 new entries are taken per date, chosen by the largest range multiple, and at most 4 legs are open at once. Maximum gross long exposure is therefore 40% of equity, and at most 20% is added on any one date. The universe is point in time: a name is eligible only once it has >=90 daily bars on Bybit (universe_at / listed_since); later-listed names (CAKE 2023-11, CFX 2023-02, FET) join when they qualify, and a delisting closes the leg at its last price. Venue choice: BYBIT is 3.4% of the corpus against a >=5% quota, and no instrument here is above ~1% share. The 1-DAY timeframe is over-represented (41%), but this mechanism is a daily-bar event, and the earlier promotions are daily. PRE-STUDY ARITHMETIC (frozen rule, gross, close-to-close 3-day forward return, all 28 names, 2021-05 to 2026-10): 529 events on 152 distinct dates. Mean +2.00%, median +1.03%, hit rate 54%. By year: 2022 +0.61% (43 events), 2023 +2.67% (115), 2024 +2.88% (220), 2025 +2.32% (76), 2026 -1.44% (70). So the recent year is negative, and the year-to-year results are uneven. Capture against cost: about 2.0% gross minus the ~0.11% Bybit taker round trip is ~1.9% per trade, over 15x the cost. Even at half the pre-study mean, the margin is above 8x. Execution: a market order at the signal bar's close. On a 24/7 venue that price is the next bar's open (the registry's use_equivalent for next_bar_open_fill). No stop; risk route (b): a fixed time exit plus capped sizing (10% per leg, unlevered).