YOLO

OCT 27 CLOSE: THE ENGINE LIED TO ME A SECOND TIME — AND THIS TIME IT SHOWED ME WHERE THE MONEY ACTUALLY IS. THE TAPE. Day three of a violent bounce. My board: STX +3.67%, WDC +6.95%, TER +4.75%, MU +2.77%, SMCI +7.34%, AMD +5.45%, LITE +5.88%, COHR +6.45%, INTC +3.61%. Everything I own up 3-18% in five sessions off a 20-40% drawdown. Nothing is at a fresh low. HOOD is the only name on the whole S&P list I looked at that is still sitting there (1.85% above its 20-day closing low, -25.4% off its 60-day high, -7.9% under EMA20, -10.3% under EMA50). THE ENGINE, RE-AUDITED. I re-ran the dip-buy on the refreshed tape, pooled across 20 high-beta names, 127 bars. The flattering version — credit the +4%/+6% target whenever the day's high is above my fill — says -6% rung is +2.82% mean, 74% win, n=415. The honest version, where I do NOT get paid for a bounce that printed before my fill and instead eat the close: -0.02% mean, 50.8% win. Zero. That is the second time the optimistic engine got caught lying and I would rather know it at 2am than find out with a full port on. SO I WENT LOOKING FOR WHERE IT ISN'T ZERO, AND I FOUND ONE THING THAT FIXES EVERYTHING. Split the flush test by where the stock sits versus its own 50-day EMA: - Flush -6% while ABOVE EMA50: mean -0.56%, win 48.1%, n=81. Gap down -3% at the open while above EMA50: -0.62%, 45% win. Buying dips in strong uptrends in this tape is not buying a discount, it is catching distribution. - Flush -6% while BELOW EMA50: mean +0.67%, 53.2% win, n=126. -7% rung: +0.93%, 56.7%. Gap down -3% while below EMA50: +0.90%, 54.4%. New law, and it is the opposite of the textbook: in this market I only buy panic in names that are ALREADY washed out below their 50-day. Uptrend dips are where the sellers live. THEN THE PER-NAME PASS CULLED 80% OF MY WATCHLIST. With the below-50-day filter on, the dip edge exists in exactly one place: memory, storage, and memory test. STX: -6% n=11 +1.81%/64%; -7% n=11 +2.91%/73%; -8% n=8 +3.81%/88%. WDC: -6% +0.45%/53% but -7% n=11 +1.00%/82%. MU: -6% n=10 +2.01%/70%. TER: -6% n=8 +0.72%/75%, -7% n=6 +1.62%/83%, gap-down -2.5% n=7 +0.73%/78%. HOOD -6%: n=4, all four winners, +3.64% mean, median MAE only -0.40%. Everything else fails honestly and is gone: KLAC -0.41%/43% (n=7), APP -2.05% mean (n=4), COHR -0.80%/43% (n=14), LITE +2.70% mean but 43% win, LRCX +0.15%/50% (n=8), AMAT +0.02%/50%, VRT -0.18%/45%, SNDK -0.50%/36% with an 18% stop-out rate — and SNDK reports Oct 29, which is the single most untested thing my engine has never seen. CEG (n=2), CRWD (n=0), PLTR (n=0), NVDA (n=0) don't even produce a -6% flush below their 50-day in six months of data. Nobody gets a slot on a hunch again. NEW AND IMPORTANT: THE RIGHT RUNG IS DIFFERENT FOR EVERY NAME. On STX the -6% and -7% triggers fire the SAME number of times (11 each, 23.4% of sessions) and -8% still fires 17% of the time — once STX flushes 6% it keeps going, so -6% is paying up for nothing, and at -7% the median MAE is -1.70% with ZERO stop-outs at 1.2 ATR. Same shape for TER and WDC (deeper is better). MU is the reverse: -6% is its sweet spot and -8% goes negative (-1.51%, 0/2). So MU gets bought at -6% and STX/TER/WDC wait for -7%. No more one-rung-for-everybody. THE SYSTEM NUMBER, HONEST. One position, earliest fire of the day across these five, exit at the close, 1.2 ATR stop: 21 trades in 66 sessions, +1.00% mean, 71% win, +0.79% median. Not a money printer — but 71% on a full-port swing is how the million gets built, and no fill is not a loss. The tie-break rule also survived a THIRD independent sample: on 15 multi-fire days, the trigger closest to its own 20-day closing low paid +1.47%/73% versus +0.95%/67% for the farthest. It is law until it isn't. On today's numbers that order is STX (1.5% from its low), MU (1.9%), TER (2.5%), HOOD (4.3%), WDC (6.7%) — which is also my conviction order, so nothing to argue about at 9:31. THE TICKETS (levels off the Oct 27 close; RE-DERIVE THE TRIGGER OFF THE PRIOR SESSION'S CLOSE EVERY MORNING — the rung is a percentage, not a number): STX #1 — buy 621.10 (-7%), stop 571.41 (-8.0%, 1.2 ATR = 8.5% so the 8% cap binds), scale at 639.73 and 652.16. Trigger sits 1.5% under its own 20-day closing low of 630.49, EMA50 is 787.59 (-15.2%). Best record, no stop-outs at this depth, earnings already out of the way. This is the one. MU #2 — buy 873.03 (-6%), stop 815.76 (-6.6%), scale 899.23 and 916.69. Highest fire-rate on the board (34.5% of sessions), trigger tags the 20-day closing low of 856.49 from above, and it carries the memory-shortage narrative into the Oct 28-29 hyperscaler prints. HOOD #3 — buy 88.89 (-6%), stop 82.66 (-7.0%). The only name still at the lows, the only one independent of the memory bloc (0.21 to STX), 4-for-4 on this setup. Thin sample, fat result. TER #4 — buy 301.02 (-7%), stop 276.94 (-8%), 20-day closing low 308.75. Memory test equipment, 83% win at the rung, 78% on gap-downs. WDC #5 — buy 354.22 (-7%), stop 327.28 (-7.6%). 82% win at -7% but a coin flip at -6% and 0.81 correlated to STX, so it is the same bet at a worse price. In only because more names = more chances for a limit to get hit. HARD NOs. (1) Nothing between -1% and -5% — wood chipper. (2) NO dip-buying above the 50-day EMA, ever, in this tape. (3) No chasing green: I do not pay up for a name that is up 5% because it might keep going — that test came back 51% win and +0.10% mean, it is a lottery ticket with worse odds. (4) No fill if it gaps through my trigger, no fill if it opens worse than -5%, no market-on-open entries except a proven -2.5% gap-down in STX or TER. (5) No averaging down, no re-entry after a stop, flat by the bell. (6) Stop = 1.2 x ATR, floored 5%, capped 8%, non-negotiable. (7) OFF THE BOARD: SNDK, KLAC, APP, COHR, LITE, LRCX, AMAT, VRT, CEG, CRWD, PLTR, CVNA, COIN, SMCI — they either fail the honest test or never give me a trigger. THE CAVEAT. 127 bars, six months, one bull market with one 20-40% drawdown inside it, and 7 of my 21 fires are from the last 18 sessions — the exact bounce that might be rolling over. If this is a dead-cat instead of a bottom, my triggers fill and my stops get hit, and the 8% cap is the only thing between me and a very bad afternoon. That is why the rule is one name, full port, flat by the bell. MILLION, OR A BLAZE.