📊 Post-Bell Friday: The Tape Printed a New ATH, the Engine Moved the Other Way, and the Override Did All the Work
Yesterday's post closed with the question of whether Thursday's four 4.96/4.97 prints were the start of a real thaw or a one-day acknowledgment that priced would get un-printed the second the tape moved. The tape answered inside two hours of open. Intel ripped +23.94% on the Q1 beat, Nvidia reclaimed a $5 trillion market cap, the semi complex hit its 18th straight green session, and the score — the same engine that walked up to 4.97 on a down day — faded right back to 4.94. Thursday's thaw: gone in one session.
QQQ opened at $658.69, tagged $658.21 on the 10 AM bar as chips lit up, ground higher every hour — $659.26, $661.20, $662.89, $662.79 — popped to an intraday high of $663.97 at 3 PM, and settled at $663.88. Official print: +$12.46 (+1.91%) over Thursday's $651.42. A fresh all-time closing high, blowing through Wednesday's $654.96 and Thursday's intraday $655.75 like they were never there. Range: $658.21 low to $663.97 high — $5.76 travel, one-way, no meaningful intraday retrace.
The broad tape split cleanly along the chip line. Nasdaq Composite +1.63% to 24,836.60 — new record. S&P 500 +0.80% to 7,165.08 — new record, first close above 7,100 on a rally day that actually belongs to the benchmark. Dow −0.2% — no record, no party, classic rotation tell. Semis did the carrying: Intel +23.94% to a fresh all-time high (topping the 2000 peak), AMD +13.9%, TSMC at an ATH, Nvidia +5.2% intraday back over $5T with a record close. SOXX +11% on the week alone. The Intel print was the trigger; the Nvidia reclaim was the confirmation.
Final Recommendation: 100% QQQ via EMA override — day 18, unchanged. Pure signal pulled back from Thursday's 4.97 peak to 4.94 (100% SQQQ on the pre-override read). EMA 70 now $612.52, cushion $51.36 — up +$11.80 from yesterday's $39.56 and an outright cycle record, blowing past Wednesday's $44.40. Ref still 4.94 from April 11; current also 4.94, delta 0.00, no rebalance. Next up trigger: 5.35 (EMA threshold gateway to Momentum) — 0.41 away. Next down trigger: there isn't one on a score basis; the override only exits on a trend break, and QQQ would need to drop $51.36 to lose the EMA 70. Override Day 18 closes the week with the tape at a record, the engine back at the ref, and more daylight between the two than this cycle has ever seen.
💾 The Intel Print That Lit the Whole Complex
Intel reported Thursday after the bell and broke the tape open. EPS $0.29 adjusted vs $0.01 expected — not a beat, a demolition. Revenue $13.58B vs $12.42B expected. Data center revenue +22% YoY to $5.1B, finally catching an AI tailwind on CPU demand for AI-adjacent infrastructure. Q2 guidance: $13.8B-$14.8B revenue and $0.20 EPS vs $13.07B / $0.09 consensus. Not a one-quarter bounce — a guide-up. The stock gapped, opened +~20%, tacked on the rest intraday, and printed its first all-time high since the 2000 peak of $75.81. 26 years later. Year-to-date Intel is up over 100%.
That single print did two things to the tape. First, it re-rated the semi laggards. AMD +13.9% is Intel's direct competitive read-across: if Intel is actually shipping and guiding, the entire x86 server market just got bigger, not zero-sum. Second — and this is the bigger one — it validated the "semis are the cycle" thesis that has been driving the tape since March. SOXX has now printed 18 consecutive green sessions and finished the week +11%. You don't get 18-in-a-row in semis without the market pricing a clean capex cycle, and today's Intel guide gave that pricing a fresh leg.
On top of Intel, Nvidia closed at a record with a ~$5.1T cap — first time back above $5T since the late-October first print. The story is unchanged — hyperscaler spend, Blackwell shipments, sovereign AI deals — but the retake of a round-number cap is its own catalyst at this point. When the heaviest single name in the complex re-takes a trillion-dollar threshold on the same day the laggard (INTC) rips 24%, you get the kind of close the Nasdaq printed: +1.63%, no afternoon fade, no distribution, record high on the tick.
📟 The Score Went the Wrong Way on Purpose. That's the Whole Point of Price Weight.
Thursday the engine walked from 4.91 to 4.97 on a down day. Today the engine walked from 4.91 (9:51 AM, QQQ ~$658) to 4.94 (10:08 AM, QQQ $658.21) — and then went silent for the rest of a +1.91% session that printed a new ATH. Two cash-session prints, both below yesterday's 4.97 peak, both below the 4.95 High Risk threshold, both parked right at the ref. Net read from Thursday's close to Friday's close: −0.03.
This is exactly what the model is built to do, and it's the cleanest textbook example of it this cycle. QQQ ripped through $655, then $660, then $663 — each level above the current signal cohort's calibration. Price-weight subtracts points at each cross. The macro side didn't get materially worse today (claims were fine, HY OAS at 2.86 is tight, mortgage rates drifted down to 6.23%) — the macro side just didn't get better fast enough to offset how much higher the market got. Net on the day: macro flat, price +1.91%, score −0.03. The engine isn't "missing" the rally — it is literally doing the arithmetic of taking profits into strength. It does not care that the tape is ATH. Its job is to say "at $663.88 the risk/reward is tighter than it was at $651.42, full stop."
And that is why the override exists. The override is the system's admission that on a strong trend the pure signal will lag the tape. Since April 11 the pure signal has said "100% SQQQ" every single session. Since April 11 QQQ has gone from $646.79 (April 20 close, closest to ref date) through $654.96, $655.11, $651.42, and now $663.88. That's a +2.64% QQQ print with the pure signal screaming inverse the entire time. The override ate every one of those bars as a win.
Ref math, for the record: 4.94 on April 11 at 8:49 AM. Current 4.94. Delta 0.00. The 0.07 rule requires a ≥0.07 move and a range change; we have neither. Upside trigger: 5.05 Cautious entry requires both conditions at once (4.94 + 0.11 = 5.05), so even a +0.07 print here to 5.01 would not flip allocation. The first level that would change the pure signal is still 5.05. The first level that exits the override entirely is the Momentum gateway at 5.35. Nothing on the downside is in play short of a trend break.
📏 $51.36 Is the New Number to Beat
The EMA cushion has become the single most useful risk metric of this override cycle, and it just printed its biggest daily jump in weeks. Thursday close: $39.56. Friday close: $51.36. One-day delta +$11.80, +29.8%. Previous cycle high was Wednesday's $44.40. The old record got cleared by seven bucks on a single session. To put the number in perspective: at current EMA 70 $612.52, QQQ would need to fall to $612.52 — a −7.7% move from Friday's close — before the override rule would even consider flipping off. That's not a normal-week event. That's a correction.
What that cushion does to position sizing decisions: it mathematically de-risks the override. When cushion was $5-10 back in mid-April, a two-day 2% pullback could put the trade on watch. At $51, a full-week rout wouldn't crack it. The combination of "pure signal pinned at 4.94" + "tape at ATH" + "cushion at cycle record" is the most one-sided the override trade has looked since it started. If anything, this is the moment where complacency is the real risk — not the override itself.
🏛️ DOJ Drops the Powell Probe. Warsh Path Clears. The Fed Story Changes Shape.
Mid-session, DC headline: US Attorney Jeanine Pirro walked back the criminal investigation of Jerome Powell, the one tied to the $2.5B Fed building renovation cost overruns. The probe gets punted to the Fed's inspector general — which is exactly where it started, exactly where it should have stayed, and exactly what a federal judge in March had already said when he called the DOJ investigation "an unjustified act of intimidation" aimed at pressuring the Fed into rate cuts.
Why it matters for the tape: Senator Thom Tillis had an effective hold on Kevin Warsh's Fed chair confirmation until the Powell probe was gone. That hold just evaporated. Warsh is now materially more likely to get confirmed, which the market has been pricing as a modestly more dovish chair than the alternative set. The tape didn't rip on this — the Intel story was doing the heavy lifting — but the headline removed a lingering tail risk around Fed independence that was flickering in bond-market chatter all month. HY OAS at 2.86 isn't tight because credit is perfect; it's tight because the institutional story of "Fed gets captured, long-end blows out, risk assets re-price" just got one step less probable.
Not a story I'd lean on for the score — the engine doesn't trade DC rumor — but worth noting as part of the "why did the S&P also print a record today, not just QQQ" question. Part of the answer is Intel. The other part is that the risk premium on Fed-adjacent tail scenarios got marked down on a Friday afternoon.
🎯 My Take: The Override Has Never Worked This Hard and This Clean
Eighteen sessions in. Pure signal has said sell/short every single day. The override has said long every single day. QQQ has gone from ref-day levels near $646 to $663.88. That's a full cycle of "the tape is right, the macro is right, the score is right, and all three of them say different things." The only reason that's not a paradox is the EMA filter — it's an explicit admission that on a trending market the 4.94 pure read is going to be wrong until it isn't. Today is the cleanest proof of that framework this cycle has produced.
Where I disagree with the natural read: the score going 4.97 → 4.94 is not the engine "giving up" or "retreating." It's not even a bearish signal shift. The move from 4.97 to 4.94 is 0.03 — three basis points in score space. That's noise at the macro level and a single price-level cross on the price-weight side. The engine is saying "today was too clean — take back yesterday's small bullish lean until the macro catches up." It's a 0.03 adjustment, not a thesis change. Readers who see 4.94 and think "the model just went more bearish" are mis-reading it. The model is pinned at 4.94. It has been since April 11. Whether it prints 4.91, 4.94 or 4.97 tomorrow is inside the margin of error of the ref.
The trade I actually care about: what does the engine do on the first real pullback? A −1% QQQ day from here ($663.88 → ~$657) is the first test. If the engine sags further — 4.90, 4.88 — I'd call that confirmation that 4.97 was a short-lived price-weight blip and the macro is still soft under the hood. If the engine holds 4.94+ on a red day, that's the first real bullish tell from the model in a month, because a hold of the ref on a pullback means the macro is actually drifting up quietly and only the price weight has been keeping the score suppressed. That's the asymmetry to watch next week. Everything above $660 is just the override grinding. Everything below $660 starts to teach us what the engine actually knows.
💡 Bottom Line: Day 18 Closes With a New ATH, a Cycle-Record Cushion, and the Engine Back at the Ref
QQQ $663.88, new all-time closing high. Nasdaq and S&P both at records. Semis +11% on the week, 18 straight green sessions. Intel printed its first ATH since 2000. Nvidia back over $5T. Pure signal 4.94 — back at the ref. Override active — 100% QQQ, day 18. Cushion to EMA 70: $51.36, cycle record by a margin of $7.
Next triggers: Upside, 5.35 — 0.41 score points away, the Momentum gateway that would flip override off and allocate 60% QQQ / 40% TQQQ. Downside, no score trigger — the override only exits if QQQ loses the 70 EMA at $612.52, which is a 7.7% drawdown from here.
One side says "pure signal has been wrong every day for eighteen sessions." The other side says "the tape just printed a new ATH and the trade is up from ref-week prices." Both are true. The override is why the second one wins. See you Monday.