Drift King

THE ROTATION PAID, AND NOW THE REAL WORK IS NOT CHASING IT. Software ran the table: ZS +31% and PANW and CRWD each +18% over the last three to four weeks while the broad tape ground sideways, and Anthropic's IPO filing on 10/26 (11,900 customers at $100k+ ARR, $16.3B of H1 revenue) gave the whole "software monetizes AI, infrastructure just burns cash to build it" story a second leg. I own that trade. What I will not do is buy it at Tuesday's closing price. ZS is 9.1% above its 10-EMA with RSI 73.2, PANW is 6.7% over its 10-EMA after five straight up days, CRWD is 6.4% over. Those are positions to hold and trim, not to add. SO THE VOLUME TEST DECIDES, BECAUSE A GREEN CANDLE IS NOT A TREND. I re-ran it on the current legs and only one name passes cleanly: ZS advanced 10/21-10/27 on 1.14x average volume against a 0.75x down-leg — that is accumulation, and it's why ZS stays my core even at RSI 73. CRWD is a tie (1.15x up / 1.15x down), but the single heaviest day of the whole flush-and-recovery is its 10/23 +8.25% on 1.55x, and that's the tell I'm paying for: buy against 251.94, T1 286.99, T2 300, half off under 244. PANW FAILS the strict test — 1.10x up versus 1.11x down, and worse, the incremental volume is FADING on the push (1.23x on 10/23, 0.96x on 10/26, 0.90x on 10/27). So PANW is a retest trade, not an extension trade: I want the 376-377 10/20-EMA cluster and I want nothing at 402. That distinction is the whole discipline. WHAT I THREW OUT, AND WHY IT COSTS ME NOTHING. MDB is up 15.6% in twenty sessions on 0.39-0.43x volume — a light-volume pop is an exit, not an entry, so it's off the panel exactly as I said last period. SNOW is grinding higher on 0.54x, same verdict. INTU is still -22.5% from its six-month high with a filled earnings gap and it lost the 10-EMA again on 10/27; dead. HUBS went -5.00% on 10/27 and -2.54% the day before, right at its 10-EMA, into an early-November print — that's distribution in front of an event, so I'm not funding it. SHOP is drifting down into 11/03 (-4.4% in five sessions, below its 10-EMA) and NET is a pre-print coil at 11/05. None of them get a dollar. I WAS WRONG ON ORACLE AND I'M RETIRING THE SHORT. My 10/14 thesis was that a -8.8% gap-down that couldn't reclaim its 10-EMA was the one clean fade in the complex. It failed me on 10/27: ORCL closed 130.53 through the 130.14 10-EMA on 26.0M shares after 19.2M and 23.0M the two sessions before — a bounce on RISING volume is the exact opposite of what my rule calls an exit. The $664B backlog and +121% cloud-infra headlines are biting. Cover, don't flip long, and I don't put a name -47.8% from its high back on the panel until it's above the 50-EMA at 139.75. THE SHORT I DO OWN: ADBE. This is drift cutting the other way, and it's the cleanest failed print in the book. It gapped +6.07% on 10/22 to 270.55 on 1.58x, then went red FOUR sessions running to 248.36 — through the 258.04 gap open, through the 10-EMA at 253.48, and through the 50-EMA at 249.82 on the heaviest day of the decline (4.73M, 1.11x). A post-earnings gap-up that can't hold the gap open is a drift DOWN, and a company that printed a good quarter and still couldn't hold $250 is the weakest relative strength I've got. Press under 248, T1 238.14 (the 10/12 gap close), T2 232.85, and I am flat on a single daily close back over 258.04. No averaging, no adding. MY TWO LAGGARDS GET TIGHT LEASHES. CRM lost the 10-EMA on 10/27 (-2.71% to 240.89) on 1.02x — 7.59M, its heaviest day since 10/22 — while the S&P was green and the Nasdaq +1.18%. Relative weakness on expanding volume is not a base, it's a break: the ~40-session patience trade is over, 239.97 is the exit, and I add nothing. WDAY tested 212.91 for the second time on 10/26 and closed 205.46 the next day, and its volume profile still fails my test (1.00x up-leg vs 1.27x down-leg). I hold the core with a hard trail at 201.42/202.15 and no adds. I DON'T PAY FOR THE EVENT — I GET PAID FOR THE TREND AFTER IT. The most important fact of this period is that 10/28-10/29 is a landmine: Microsoft, Alphabet and Amazon are ~15% of the S&P, and NOW and FTNT print the same day, PLTR on 11/02. Every one of those names is in my book as a POST-PRINT candidate and none of them gets premium from me before the number. The gate is identical for all four: wait until the gap is TWO sessions old, require the gap-day low AND the 10-EMA to hold, then buy monthly calls — MSFT over 529.83 (template: 7/30 gapped +12.13%, closed +15.5% on 3.12x, drifted +17.8% and never filled the gap), FTNT over 179.86, NOW over 151.55, PLTR over 197.44. If instead they gap down and can't reclaim the 10-EMA, the drift pays the other way: MSFT 507.23, NOW 137.96, FTNT 172.94, PLTR back to 184.45. RULES: daily closes only; longs die under their 10-EMA; shorts die over theirs; no adds within 2% of the 10-EMA; no chasing a name more than 10% above its 10-EMA; no averaging down; trim the extended, don't sell the trend; and when a move comes on less volume than the one before it, that's an exit, not an entry.