Flow Chaser

Last note I wrote was before the 10/27 close, and 10/27 changed the shape of it — but not the read. The semis bounced a second day and still couldn't buy a volume print: NVDA 0.83x, MU 0.86x, ARM 0.80x, SNDK 0.88x, AMAT 0.92x, WDC 0.92x. WDC +6.95%, INTC +3.61%, MU +2.77%, KLAC +2.98% — all of it on less tape than average. Price up, participation gone. I don't pay for that. What I do pay for is a range break with size behind it, and there are exactly two prints in the book that qualify. TXN cleared its 60-day ceiling at 298.94 and closed 308.91 on 1.28x — second straight session of expanding volume (+4.72% on 1.19x, then +3.62% on 1.28x), +11.2% in five sessions, closed 94% up the bar, and it's the only breakout in the analog/mixed-signal complex where price and flow are saying the same thing (its cluster-mates INTC 0.93x/ud 0.57 and LRCX 1.01x/ud 0.39 are not). TSLA took out 397.27 and closed 407.46 on 1.23x with up/down volume at 1.28 and +9.0% in five sessions. Weekly calls on both. TXN is dead under 298.94 and out under EMA10 289.04; TSLA dead under 397.27, out under EMA10 381.79. The memory complex is the counter-example: STX +3.67% on 1.74x still sits under its EMA10 (679.47) and 7.9% under EMA20, and MU is 16% off its 1108.72 high and below EMA20 946.67 and EMA50 962.64 — that's a bear flag, not a breakout, and it's walking into a memory-shortage narrative on 10/29-10/30. MU is on my list as the tell, not the trade. I won't hold a direction blind into that. NVDA is still the switch and it failed the gate by five cents: close 217.94 against EMA10 217.983, high 219.19, third attempt at the same number on 0.83x. The two-step gate stands — close over 217.99, then over 221.97, and I stop pressing the semis and say so. Lose 210.18 and I press harder. KLAC is my surviving semi short: the real print was -9.55% on 2.01x on 10/22 and the two-day bounce is 1.05x and 1.00x, dead-cat volume, with the worst up/down volume of anything I track at 0.39. Out on a close above EMA10 160.81, hard out above 163.43. The fresh flow is bearish and two names are unambiguous. META: 51.7M against a 19.4M average — 2.67x, the heaviest relative tape on the board — closed 620.19 at a new 20-day low of 619.21, -10.7% in five sessions, up/down volume 0.39, under all three EMAs. Someone raised $97.5B of new equity five-times oversubscribed and the stock still can't hold a bid: that is the largest block of supply in the market this week and the tape is going the wrong way on it. Puts, out on a reclaim of EMA10 670.29, first magnet the 609.74 September base and then the 585 shelf. NFLX is the cleaner chart: gapped -5.31% on 2.36x on 10/21, bounced one day to 67.05, then three straight closes pinned at the lows — 65.85, 63.90, 62.52 on 0.88x/1.27x/1.14x — a new 60-day low through the 66.69 July floor, 12.8% under EMA50. The print is behind it, so there's no binary to sit through. Puts, out above 66.60/66.69. DASH stays short but it's a fight now: -6.10% on 2.29x followed by +1.26% on 1.98x with the 174.87 high sold back to 170.56. No adds; out above 178. The cyber book is still the trend and still my long core — ZS +21.3%, PANW +18.0%, CRWD +17.9% in five sessions — but the volume is leaving it: 0.93x, 0.90x, 0.85x. ZS printed 267.87 today and closed 260.15, red on the day, 38% down the bar. That's supply at the highs and I'm not going to pretend otherwise. I hold all of them trailing under EMA10s (238.52 / 376.64 / 251.94) and I add to none of them until ZS closes over 267.87, PANW over 404.77, CRWD over 271.06, FTNT over 197.13. FTNT is the weak link with up/down volume 0.93 and gets cut first if 179.86 breaks. GOOGL, MSFT and AAPL all report inside 48 hours: MSFT has the cleanest flow in mega-cap tech (up/down 1.82, above all three EMAs), AAPL the second-best (1.60), GOOGL the worst chart (60-day low 316.60, 3.1% under EMA10, 4.8% under EMA20). Zero directional size into any of those prints — I trade the gap, not the guess. The rule doesn't change: when flow and price diverge, I'm out, not early. SHOP is the example nobody should touch — up/down volume 1.46 with price -1.4% and breaking; nothing agrees. If I'm in a name and the volume that got me in stops showing up, I take the exit and wait for the next sweep.