Relative Strength

SAME RULER, SEVENTH PERIOD, AND THE FIRST TIME IT HAS BEATEN ME IN A WHILE. My equal-weight mega-cap benchmark (AAPL, MSFT, NVDA, AMZN, GOOGL, META, JPM, BRK.B) closed 108.56: -2.09% on the week, -0.92% on the month, +9.30% on the quarter. But look at the shape instead of the number — three consecutive up closes off 107.95, and the index is now 0.17% below its 50-day at 108.75 and 1.84% below its 20-day at 110.59. That is a chop, not a trend. And the chop is tightening: across the 32 names I re-marked on the 10/27 close, one-month relative return spans 62.31 points at 14.75 std, from ACN +35.71 to STX -26.60, with 18 of 32 beating the ruler. Last period that was a 66.83 span at 17.73 std. Dispersion compressing means my edge is thinner this week, and the correct response to a thinner edge is a smaller book with harder numbers, not more names. WHAT THE TAPE DID TO ME ON 10/27, SAID PLAINLY. My capex-fade complex got squeezed as a single instrument: WDC +6.67 relative, VRT +4.11, AVGO +3.72, CAT +2.78, NRG +1.49 — five positions, one bid, the day after Meta raised $97.5B for AI infrastructure and Nvidia put $3B into Meta. AVGO closed 342.33, straight through my 335.00 cover. My standing rule was that the first two that tag stops take the whole complex to half size that same week; only one tagged on a close, but when all five move as one, the count is a formality and the correlation is the signal. So AVGO is out and I am not re-shorting a 2.5x-volume squeeze; VRT, CAT and NRG go to half of prior size; WDC I hold at HALF with cover 385.00 and I will take the loss without a debate if it closes there. I am not re-anchoring a single one of those covers. The numbers were written to be hit and they are getting hit. AND MY LONG TAIL BROKE, WHICH IS THE BIGGER PROBLEM. Four of my longs are simply gone. ACN closed 235.16 through my 243.00 stop — and note it was not a wick: -4.88% and it took the one-month line from +43.57 to +35.71 and the quarter from +42.93 to +26.50 in a single week. Still the biggest relative line on the board, still 49 of 50 closes over a 50-day at 198.53, but the deceleration is now two periods deep and the number is the number. Out. Re-entry only on a close back over 243.00, and not a share below it. PSX closed 268.53 through 277.00 on 1.29x with its 10-day already 6.65% below — broken, out, no re-entry until 277.00. MCK took out the 975.00 hard stop I gave it after last period's half-cut; the quarter has rolled to -0.47, so that is a full exit and not another size cut. CAH I am cutting on the line rather than the price: the week is -2.22 relative and the quarter has gone from +0.97 to -1.73, which is exactly the "one-month wonder" I warned about. Out. WHAT SURVIVES. Four FULL positions, and all four earned it on persistence. HAS: +33.21 on the month, +10.13 on the week, sitting on its 20-day high 18.13% above a rising 20-day at 100.17 — parabolic, no adds, stop up to 114.00. ISRG: +16.98 on the quarter, 20 of 20 closes over a 20-day at 424.12, stop 428.00. AMGN is my cleanest healthcare line now that MCK and CAH are gone — +4.14 week, +8.05 month, +6.26 quarter, 20 of 20 over a 20-day at 423.86 — stop up to 435.00. XOM: 50 of 50 closes over a 50-day at 166.64 and +11.62 on the month inside a group running -8.1%. It gave back 2.51% on 1.22x on 10/27 but it is the only survivor in a broken sector; it stays FULL with the stop at 175.00 and no sympathy for the group. THE MIDDLE OF THE BOOK, AND WHY IT IS CAPPED. LLY stays HALF and I want to be honest that this is the third period my own written condition has failed: full size when the quarter turns positive, and the quarter is -3.36. HALF, stop 1245.00. HUM HALF — +7.39 on the month, +4.23 on the quarter, but the week is -0.85 and price 414.85 is barely above a 20-day at 412.15; stop 405.00. IBM HALF, unchanged and un-nudged: rel5 +5.53 is second-best on the board and it closed 250.25, but the intraday high was 253.18 against my 256.30 full trigger, and the 20-day at 233.82 is STILL 0.01% under the 50-day at 233.85. Neither trigger has fired; I do not get to call it close enough. Stop 246.00. V and MA are one idea and I am treating them as one unit: V beat on 10/27 and lost 0.85 relative on the day, MA has a -1.04 quarter, and both closed within a hair of stop levels set inside the noise. I am re-anchoring both, once, to the 20-day where they actually live — V stop 390.00 against a 20-day at 388.88, MA stop 598.00 against a 20-day at 594.77 — and I am taking both to HALF. Good news producing a bad relative day is the tell I trade on. THE ONE NEW LONG. CSX. A fresh close at 52.27 that is simultaneously the 20-day high and the 50-day high, a +10.80 relative week that is the strongest five-day reading on my entire board, +12.71 on the month, and the breakout leg came on 2.71x volume after a 9.6% EPS beat with revenue up 12%. The quarter is still -6.28 and only 14 of 50 closes are above the 50-day, which is why this is HALF and not FULL: this is a relative line inflecting, not yet a leader. Stop 49.40 under the gap-day low; I add toward FULL on a close over 53.50. FOUR CONDITIONALS, ALL WITH A NUMBER. ACN over 243.00 for HALF. FFIV over 472.03 for HALF — and it is the best structural profile of anything new, +2.67 week / +6.39 month / +9.42 quarter with its 20-day 6.62% above its 50-day and 36 of 50 closes above it, but it beat earnings on 10/27 and closed red on 2.76x volume, and I do not buy churn. DELL over 518.58 for HALF: a +26.50 quarter and 39 of 50 over the 50-day is real, but the month is -6.71 and that 7.47% rip came on 0.99x volume. MU over 972.37 for HALF: the cleanest quarter in the entire memory complex at +16.38, a 10/30 print, and 30 of 50 over its 50-day — but the month is -10.96 and price is 4.48% under the 20-day, so no entry until it reclaims it. WHAT I PASSED ON, AND THE RULE THAT MADE ME. STX just beat, is up 3.67% on 1.64x, and its one-month relative line is -26.60 with the quarter at -21.93, 15 of 50 closes over the 50-day and price 10.57% below its 20-day. That is not leadership, that is a bounce in a broken line with a good story attached. SNDK is the same shape from the other side — quarter +19.15, month -22.90, 3 of 20 over the 20-day — and it reports 10/29; I am not paying up for a catalyst. HOOD beat and has 0 of 20 closes over its 20-day and -17.88 on the month; a beat that cannot lift the line is information. HPE has the single best three-month relative number I measured at +27.61 and 42 of 50 over the 50-day, and I am still passing, because it peaked at 73.46, fell to 55.13, and the last two up days came on 0.49x and 0.58x volume — a trailing-window trophy is not a current line, which is the same sentence I wrote about MU and WDC last period and I am not going to forget it in a week. BXP beat on FFO and closed -2.48 relative with 5 of 50 over its 50-day. RCL is the hardest pass: +22.91 on the month, a close at the 20-day high on 2.14x after a 5.5% earnings gap — and a -17.84 quarter with 15 of 50 over the 50-day, which is CCL's setup from last period verbatim. Passed then, passed now; the moment the quarter crosses zero it becomes a buy. UNH I am dropping from the book entirely rather than watch the same 430.15 breakout refuse me a seventh time. And the sector snapshot still ranks Semis & Infrastructure first at +13.3% one-month on a 600-day window while I have now measured MU at -10.96, WDC at -15.04, STX at -26.60 and SNDK at -22.90 on the month. That ranking is a trophy case, not a race. TWENTY-ONE NAMES, ONE RULER, AND A SMALLER BOOK THAN LAST WEEK ON PURPOSE. Every fade carries a hard stop and every long carries a number; the complex is already halved, and if VRT tags 214.00 or CAT tags 715.00 the whole capex leg comes off and I will not replace it. I own what is winning the race against the index. Nothing else.