📊 Post-Bell Thursday: The Engine Fires Five Times — and the Tape Doesn't Care
Last night's post closed with a single sentence: "Override Day 22 opens tomorrow with a different problem set than any day before it: the macro is louder, the cushion is thinner, and the next leg has actual asymmetry on both sides for the first time since the cycle began." Day 22 picked the upside leg of that asymmetry by close — but the road there ran through the loudest single-session score sequence we've recorded in 22 sessions of override.
The print count alone is the headline. Three weeks of zeros and ones. Today: five. 5.00 at 9:04 AM ET (no QQQ tick yet — pre-market) — the highest pure-signal print of the entire override cycle and the closest the engine has gotten to the 5.05 Cautious line in 22 sessions. Then the open arrived and reality reasserted itself: 4.90 at 9:39 AM with QQQ at $661.19 — a 0.10 fade in 35 minutes. 4.91 at 10:14 AM with QQQ at $659.98 on a brief opening dip. Then the tape unstuck, ran for hours, and the score put together the second half of the day's story: 4.88 at 1:29 PM with QQQ at $666.06 — a fresh cycle low for the 4.90-4.97 band — and 4.89 at 1:54 PM with QQQ at $667.09. Five prints, ~25 cents of QQQ travel between the highest and lowest score reading, and a closing tape that ignored every word of the conversation.
QQQ closed at $667.74, up +$6.17 (+0.93%) from yesterday's $661.57 — a fresh all-time high, taking out Monday's $664.36 by $3.38 and erasing every cent of the FOMC-week chop in a single session. S&P 500 +1.02% to 7,209.01, the first close above 7,200 in history. Nasdaq +0.89% to 24,892.31, also a fresh record. The Caterpillar / Alphabet beat-led rotation absorbed Microsoft's $190B capex sticker shock by midday, and Nvidia's reclaim back above $5T market cap did the heavy lifting on the index. The tape that closed flat on FOMC day decided overnight that the AI capex framing was bullish, not bearish, and proceeded to set new records on the strength of that single re-read.
Final Recommendation: 100% QQQ via EMA override — Day 22, unchanged. Pure score 4.89, raw signal still Extreme Risk-Off (100% SQQQ). Override active because QQQ $667.74 sits comfortably above EMA 70 $618.85. Cushion: $48.89, recovered +$4.18 from yesterday's $44.71 and back to within $2.47 of the $51.36 cycle peak from April 24. EMA 25 stretch zone: neutral, distance +5.4% — pulled back into the +6% bull-stretch trim trigger by today's rally; QQQ is now roughly $3.80 of headline price away from a 50% QQQ / 50% Cash override trim. Bear-stretch flags: both off. Ref still 4.94 from April 11. Delta to ref: −0.05. No rebalance — every print today, including the 4.88 cycle low, sits inside the 0.07 band. Next up trigger: 5.01 on the 0.07 rule (would lift out of Extreme Risk-Off into High Risk). Next down trigger: trend break only (no score-based exit while above EMA 70).
🔔 The 9:04 AM 5.00 Print Is the Most Interesting Tick of the Cycle
For 22 sessions the override cycle has produced exactly one direction of pressure on the score: down. 4.94 → 4.93 → 4.92 → 4.91 → 4.90, occasionally probing 4.97 on chip beats and otherwise grinding the bottom of the band. The model's macro-vs-price arithmetic kept telling the same story: macro deteriorating in slow steady increments, price ignoring it, score holding to its read while the cushion did the work of keeping the portfolio long.
Today's pre-market 9:04 AM tick broke that pattern from the wrong end. 5.00. That's +0.06 from the 4.94 ref, the largest upside delta of the entire cycle, and a print that sits 0.05 away from the 5.05 Cautious line — the boundary that, if crossed, would end the Extreme Risk-Off pure-signal regime that has defined three full weeks of trading. The override would still be active either way, but the underlying engine narrative changes meaningfully if the score lifts into a different zone.
It didn't last. By 9:39 AM the score was at 4.90 — a 0.10 fade in 35 minutes, with the tape opening at $661.19 and dipping toward $660. That's a clean piece of price-vs-level math: pre-market input was favoring macro inputs (overnight earnings tone, Iran headlines softening, Brent off the highs), and the open's price action immediately re-tilted the calculation toward "QQQ is again pushing into territory where the score takes profits at higher prices." The 5.00 was a real read of the model's thinking when price was below $662. The 4.90 was the same model's read once price was kissing $660 with momentum.
The fact that the score then drifted further down to 4.88 by 1:29 PM as QQQ ripped through $666 confirms the price-weight side of the engine is doing exactly what it's designed to do — every new high prints a smaller score. Five prints, two distinct regimes inside one session: macro-leaning at 9:04 AM ($660-661 territory), price-leaning by 1:30 PM ($666+ territory). The session's lesson: the band hasn't broken upward, but it's no longer one-sided. The engine can now go either direction from here on the next macro shift, which is the first time that's been true since April 11.
📈 April Closes: +15.3% Nasdaq, +10.4% S&P, +7.1% Dow — and the Ref Hasn't Moved Once
Month-end print: Nasdaq +15.3% for April, the index's best month since April 2020. S&P 500 +10.4%, best month since November 2020. Dow +7.1%, best month since November 2024. QQQ specifically: from a March 31 close in the low $580s to today's $667.74 — call it roughly +$83 of QQQ in 22 sessions, give or take the open-to-close drift on March 31. The override flipped on April 1 with a sub-$5 cushion to EMA 70. It exits April with $48.89.
Across that entire move, the Edge Of Markets pure score did this: 4.94 (April 11) → 4.88 (today). Six basis points of decline. Inside the same Extreme Risk-Off zone the entire time. Ref hasn't moved off 4.94 once. The engine's read of the macro got marginally worse over a month where the tape printed its strongest April since the COVID rebound.
That's exactly the divergence the override is engineered to handle. The pure score is a price-discounting macro engine. It does not have a momentum input. When the tape rips $83 in a month on AI capex enthusiasm, the score's job is to flag the gap — not to chase it. The trend filter's job is to keep the portfolio long while the gap exists. Both pieces did their job in April. The portfolio captured the +15% Nasdaq month while the pure score sat at 4.88-4.94.
The honest read going forward: a 22-session, $80+ rally without a single ref update is the kind of stretch that historically ends in one of two ways — either the macro re-rates upward (the score lifts into Cautious or Neutral and the override hands off to a real long signal), or the tape rolls over enough to test the cushion (and the override exits on a trend break, not a score event). Today's pre-market 5.00 was the first plausible flicker of the upward path the cycle has produced. It didn't hold, but it existed.
📐 Cushion Reset: $44.71 → $48.89, Within $2.47 of the Cycle Peak
Yesterday's post tracked the cushion arc as a problem in formation: $51.36 → $49.61 → $41.98 → $44.71, with EMA 70 finally grinding faster than QQQ and the cushion compressing on net. Today snapped that trend in one bar. EMA 70 +$1.42 ($617.43 → $618.85). QQQ +$6.17. Cushion +$4.18 to $48.89. The first time in four sessions the tape outran the trend filter, and by a margin big enough to undo all of Tuesday's compression in a single close.
A $48.89 cushion translates to a −7.32% drawdown requirement from $667.74 to even threaten the EMA 70 floor in a single move. The cycle peak was $51.36 (−7.74%). Today's print is back inside $2.47 of that record. The override has more room than it had at any point this week, and the trend filter is climbing at $1-1.50 a day — fast for an EMA, but not fast enough to overcome a +0.93% QQQ session.
Practical translation: a tape break that meaningfully threatens the override now has to be at minimum a one-day −5% type event combined with a flat-to-up EMA 70, or a multi-session grind with QQQ flat and EMA continuing to climb at $1.50/day (which would compress the cushion ~$30 over 20 sessions of stagnation). Neither is the base case for the next five trading days. The base case is: Apple AMC, jobs report next Friday, and a tape that's still riding the +15% April momentum into May.
🤖 The MSFT Capex Sticker Shock the Tape Decided to Ignore
Last night Microsoft printed a clean revenue beat with Azure +40% YoY (vs. analyst expectations clustered at 38.8-39.3%) — the strongest cloud number in the AMC slot. The follow-up line: full-year capex guide of $190B, citing memory cost inflation. The after-hours tape took the print at face value and dropped MSFT roughly −3.9% in the immediate reaction. Going into Thursday's open, the framing was: AI capex is now eating margin guidance, and the rest of the Mag 7 may be next.
The Thursday open did not validate that framing. Alphabet's prior-night beat held its bid, Caterpillar's industrials read came in clean, and the chip rotation that has been running since the start of the month re-engaged. Nvidia reclaimed the $5T cap line intraday and led the Nasdaq to a 24,892.31 close. The MSFT capex story did not generalize — the market read the print as "Microsoft specifically has memory cost issues" rather than "AI infrastructure spend is now uneconomic across the cohort." That's a meaningfully different conclusion, and it kept the day in the green.
For the score, the MSFT capex line is exactly the kind of macro input that takes time to filter through. Capital expenditure detail doesn't move the engine the way an Iran escalation or an oil print or an inflation report does — it shows up indirectly through credit spread effects, earnings expectations, and eventually consumer sentiment. A single −3.9% MSFT print will not reset the 4.94 ref. A re-rating of cloud margin expectations across MSFT/GOOGL/AMZN might, over weeks. None of that is in today's tape. Today's tape is: rip on the chip names, hold on the index, ignore the capex headline.
🎯 My Take: Five Prints in One Day Is the Engine Saying "I Am Watching, I Just Don't Have a Reason Yet"
For three weeks the criticism of this engine has been easy to articulate: nothing is moving, the band is dead, the model is asleep. Today refuted the lazy version of that critique cleanly. Five prints. Two distinct macro reads inside one session. A fresh cycle high (5.00) and a fresh cycle low (4.88) inside the same eight-hour window. The engine is not asleep. It is doing exactly what a properly-calibrated price-discounting macro model should do: registering every meaningful shift in the inputs, refusing to commit to a new regime until the inputs cumulatively earn it.
What killed the upside path today was the 35-minute round-trip from 5.00 to 4.90. That's a model saying: "Whatever priced me at 5.00 in pre-market just got fully reversed by the open's price action." If the ref was going to crack upward, it needed the 5.00 print to survive the open — to be confirmed by an opening tape that didn't immediately push price into the territory where price-weight starts subtracting points. Instead the open did the exact opposite: QQQ pushed into the territory where price-weight cancels macro lift, and the score behaved accordingly.
Reading forward: tomorrow's Apple AMC print is the only remaining single-stock binary of the gauntlet. If AAPL prints clean and the tape gaps up Friday, the EMA 25 stretch math becomes very interesting very quickly — the bull-stretch trim trigger at +6% is currently sitting roughly $3.80 of QQQ away. A real green Friday could activate the first allocation modifier of the override cycle. If AAPL whiffs and the tape fades, the 4.88 cycle low almost certainly extends — but probably not far enough to break 0.07 from ref absent another catalyst stacked on top. The asymmetry of the setup is now real on both sides for the first time since this whole sequence began.
💡 Bottom Line: Override Day 22 Closes With the Cushion Reset, the Engine Awake, and the Trim Trigger Inside Striking Distance
Recommendation holds: 100% QQQ via EMA override. Pure score 4.89 (Extreme Risk-Off — raw signal would be 100% SQQQ). Ref 4.94, delta −0.05, no rebalance. Cushion $48.89 — back inside $2.47 of the cycle peak. EMA 25 distance +5.4% — within striking distance of the +6% bull-stretch trim. Stretch zone still neutral. Bear flags off.
April books as a top-five Nasdaq month of the past 25 years and a top-five S&P month of the same window — and the override caught the entire move long while the pure score never updated its ref off the April 11 reading. That's the trade. Looking into May: Apple AMC tonight is the last earnings binary on the calendar, the EMA 25 stretch trim is now within one good Friday of activating for the first time, and the engine demonstrated today that it has both directions of asymmetry available again. Override Day 22 closes intact, awake, and structurally healthier than it has been since Tuesday.
Three weeks of ones and zeros. One day of fives. The engine talked today. The tape didn't listen. Whether that's a feature or a setup depends on whether the band breaks or holds in the next five sessions — and right now the only catalyst on the calendar with the structural weight to break it is sitting on the AAPL print clock.