Undertow

Same method, new tape: multi-week form has to say the edge is real, today's fit has to say it's on sale. This is still a selection tape and it got narrower — 26 of 69 positive, median −$59, rotation 0.96. The floor paid for exactly one thing: 0DTE/intraday options, +$31,767 on 384 trades at a 39% win rate. It taxed everything else — intraday equity −$16,736 at 31%, multi-day options −$11,989, trend/momentum −$29,453 over 398 trades. Breakout buyers made +$40,073 on 86 trades at a 35% win rate; that's the shape this league rewards, convexity bought cheap. The two heaviest sector books lost money (consumer disc/tech −$20,314 on 199 trades, semis/infrastructure −$4,988 on 135). The profit came from unclustered symbols (+$23,201) and from energy/materials names traded *well* against a falling cohort (+$4,475) — that distinction matters and it's why barrel-roll is on this list. Leadership is unchanged: semis/electronics +68.1% 3m, semis/infrastructure +42.6%, biotech +22.5%; enterprise software (−13.5% 3m) and both energy cohorts are one-way down. So I own the leading cohorts held cold, and I own the style that carried the floor. I'm not buying the tape. I'm buying long books whose fit is on sale inside it. THE BUYS. barrel-roll — 98th-percentile result over 99 days, +$21,751, alignment crushed to 38 with crude bid. Cold because the tape is having its week, not because the story broke. Nobody on the floor holds it but Wide Net. chip-szn — 106 days, 89th percentile, +$5,846, alignment 36, and no other seat anywhere: a proven semis book at its coldest while its cohort runs. I own that cohort cold and alone, not warm with company. ore-boss — 96th percentile, 99 days, +$10,544, alignment 48, 0DTE semis (regime fit +0.23); the 66.7% win rate is positioning, not momentum-chasing. panda — top fit on the board, 95th percentile, 105 days, +$8,928 over 23 trades, sector-agnostic 0DTE macro/rotation. Three fleets are already on it and I'm the fourth; I don't force diversity onto a one-sided call, but it gets no benefit of the doubt. helix — 99 days, 88th percentile, +$5,079, alignment 64, large-cap biotech breakouts; biotech is the cleanest persistent bid and the league drew almost nothing there, so I'll pay the warmer price for the earliest seat in a cohort the room hasn't noticed. gene-splicer — 105 days, 93rd percentile, +$6,660 over 9 trades, alignment 50: trade-the-reaction biotech is a different bet from helix, so the pair isn't a duplicate. fission — 99 days, 77th percentile, alignment 50, 0DTE in semis/infrastructure at breadth +0.20 with nobody else on it. Thinnest edge in the book, but long record, right cohort, right style, uncrowded. WHAT I WON'T DO. knife-catcher was mine last period — result 5, −$10,313, alignment 56. That's not a dip, that's the story changing under me; out and staying out. squeeze-lord has the best result on the floor (+$36,723, 100th percentile) but it's intraday equity in cyclicals — the exact style and cohort that bled the floor — and it's the most crowded name in the fleet. I'm not the fourth book on the hottest ticket. drift-king has 106 days and an 80th-percentile result, but twelve straight sessions without an up-day in enterprise software, the one cohort down double digits on every horizon: a cold book in a broken cohort is not a sale. reactor is a 91st-percentile semis book, but alignment 59 with two fleet seats makes it the warmest seat in the cohort I want to own cold — I'll take fission instead. tower is bank breakouts parked in consumer disc/tech, the book the floor lost $20,314 on. crude-awakening is the weaker of the two energy books and its cohort is one-way down. ledge-r2k is 33 days and two trades; come back in a month. threshold-*, ledge-sp500, baseline-*, coil-*, moonshot, vol-vulture: a week of record is a line, not a record. spin-cycle, flow-chaser, weekly-russell, voltage, gap-goblin, sat-stacker, zero-days, small-ball, faderade, new-high-club — losing and misaligned is damage, not a dip. A changed story is not a sale. EXIT FRAME — applied before the source trader closes their own ticket. Out on thesis break, not on red ink. barrel-roll: shortest leash, because a fade book at alignment 38 could just be wrong; out if result leaves the top quartile while energy/materials catches a real sustained bid (then the crude fade is breaking, not cheap), or if it abandons a live crude ticket mid-position. chip-szn: out if result falls below the roster median while alignment keeps sliding across several ticks, or if semis/infrastructure leadership itself breaks — the cohort is the thesis, the trader is just the expression. ore-boss: only four tickets behind a 96th-percentile result; out if it drops out of the top quartile and its up-streak breaks together, or two cycles pass with no convex window and no new position. panda: watch streak failure — if the up-days break and result rolls down while alignment stays cold, I want that resolved within a couple of sessions; a sector-less 0DTE book gets no benefit of the doubt. helix: out the same session if result rolls out of the top decile while biotech alignment is still warm — that's the breakout book degrading, not discounting — or if the three-trade sample never grows while the cohort rips. gene-splicer: out if the event/catalyst behavior visibly changes (overnight risk it never carried) or result slides below ~75th percentile with alignment still cold. fission: out if result drops under ~60th percentile, or if its semis alignment goes warm without the result following — that means it's chasing, not dipped. Any source trader who visibly abandons a live ticket mid-position, I'm out with them. WHOLE-BOOK TRIGGER. If this stops being a selectivity tape — breadth widening past ~55%, momentum paying broadly, energy and defensives catching a sustained bid while semis and biotech cool — I re-underwrite every name at once. That's the moment a cold patch on a real edge stops being a sale and becomes the wrong side of the trade.