Equity melt-up versus building recession risk
What changed
The headline signal — how confidence moved from the previous snapshot, and why.
Moved from 2.1.9 0.71 ± 0.07 to 0.72 ± 0.07 - mean step of +0.01 on the risk-on reopen. S&P 500 futures +0.78%, Nasdaq-100 +1.14%, Dow futures +371 points led into the post-holiday cash session on US-Iran deal optimism, extending the Friday Dow-record rebound. The step is held to +0.01 because these are futures, not a close, the
fresh US strikes on Iran are a two-sided risk, and neither invalidation leg moved (no vol-expansion break; VIX, plan-restricted on the feed, nowhere near the < 15 melt-up-confirmation leg), with
breadth still concentrated in mega-cap AI a fragility tell and the 84%-beat-rate / 13.4%-margin earnings backdrop unchanged. Width held at 0.07. Beta(34, 13.5) ~48 effective observations. The Tuesday cash close, Warsh's reform agenda, and May 28 PCE are the next tells.
The thesis
The claim and where confidence stands now.
The Monday cash session held up FAR better than the Monday-morning bearish framing implied. SPY closed 738.65 (-0.07% from Friday's 739.17), QQQ 705.88, Dow -0.34%, Nasdaq Composite -0.07%, VIX actually FELL to 18.43. The Tuesday cash session then closed lower for a THIRD consecutive day: SPX 7,353.61 (-0.67%),
Nasdaq Composite 25,870.71 (-0.84%),
SPY 733.73, QQQ 701.53;
VIX closed 18.06 (+1.35%) - a volatility bid, reversing the morning's softening read.
Wednesday pre-market futures have since recovered (ES +0.4%, NQ +0.8%, Dow +75 pts) ahead of the binary cluster, integrating the down days rather than cascading. The Friday tape (SPX 7,408.50 -1.24%, NVDA -4.4%, AMD -5.7%, INTC -8%) has not been re-tested. CME FedWatch hike-by-December holds the ~56% level. Q1 2026 earnings season closed at 84% beat rate, 27.7% blended EPS growth, blended net margin 13.4% (highest since FactSet began tracking in 2009). The invalidation grammar requires either a vol-expansion break (VIX > 25 with SPY breaking 50d MA for 5 trading days) or unimpeded melt-up confirmation (SPX > 7,300 with VIX < 15 for 5 trading days) - VIX at 18.06 sits squarely in the mid-zone (SPX 7,353 is above 7,300, but VIX is far from the <15 the melt-up-confirmation leg needs), neither has fired. Three binary tells land in the next ~30 hours Wed May 20 - Thu May 21: FOMC minutes (2 PM ET Wed), NVDA Q1 FY2027 (5 PM ET Wed), Walmart Q1 FY2027 (pre-open Thu, tariff-passthrough lens). Update May 20: the session rallied broadly -
Nasdaq +1.54%, S&P +1.08%, Dow +1.31%,
SPY 741.25 (+1.02%) - reversing the three down days, though the
hawkish FOMC minutes and
muted NVDA after-hours reaction temper the read into Thursday's Walmart print. Update May 21: the rally faded -
Nasdaq 100 -0.6%, S&P 500 -0.3% into Thursday afternoon as yields rebounded and oil topped $100, the
NVDA beat-and-raise was shrugged off, and
Walmart fell ~7.6% on consumer-distress caution;
the S&P is nearing overbought with breadth concentrated in mega-cap AI. Neither invalidation leg moved closer. Update May 22: the fade reversed - the Thursday cash session actually closed green (
SPY 742.72 +0.20%,
QQQ 714.51) and Friday
stocks rebounded broadly to a fresh Dow record (S&P +0.55%, Nasdaq +0.53%, Russell 2000 +0.93%) on revived US-Iran peace hopes and
easing yields; neither invalidation leg moved. Update May 25: the Friday cash closes confirmed the rebound -
SPY 745.64 (+0.39%),
QQQ 717.54 (+0.42%) - and over the holiday weekend (no Monday session) commentary framed the tape as a continued melt-up while flagging mega-cap-AI breadth concentration. Neither invalidation leg is closer: VIX is plan-restricted on the feed and nowhere near the <15 melt-up-confirmation leg, and there is no vol-expansion break. Update May 26: risk-on into the post-holiday reopen -
S&P 500 futures +0.78%, Nasdaq-100 +1.14%, Dow futures +371 points on US-Iran deal optimism, extending the Friday Dow-record rebound - though the
fresh US strikes on Iran are a two-sided risk; these are futures, not a close, and neither invalidation leg moved.
Drivers
The underlying macro forces this thesis expresses - the loading mean is how much each force drives the thesis, the stddev our confidence in the mapping.
Supporting evidence
Typed, citation-backed observations across time, grouped by strength. Hover a point for the claim.
What would invalidate this
The machine-evaluable conditions that would falsify the thesis.