The Nightcap White-Cap

Wednesday, 06-17-2026

Evening market read · validation & recap

The Milkman
OuroTaurus
Built 2026-06-18 ~00:55 ET · validates the 2026-06-17 Warsh-FOMC regular-session close · anchor: system-reminder 2026-06-18 (PRIMARY) = Bash 2026-06-18 Thu 00:48 EDT; subject_date 2026-06-17 (after-midnight run — prior session) static — regenerate to refresh

01Session Scorecard

The Warsh inaugural FOMC was the binary the whole book was waiting on. It resolved hawkish — the two mirror theses split exactly as designed.

SetupOutcomeEvidenceΔ-ATRActual
LRT SHORT SPY/QQQmm-260617-LRT-SPY-S FIRE Level-rejection short fired on its own mechanic: hawkish dots (2026 cut deleted, 9/12 voters now see a hike) + hawkish Warsh + breadth failure (0 of 11 SPDRs green) + SPY rejected back below the 745.34 cap. Neither kill (dovish_fed / clean-break-hold>745.34) tripped. 0.47 SPY 740.96 (-1.25%) · +0.32R
SBD LONG SPY/QQQ/SMHmm-260617-SBD-SPY-L VOID Wall-of-worry reclaim long required non-hawkish Warsh + hold>745.34 on broad participation. Warsh came hawkish (hawkish_warsh kill), SPY closed below the cap, breadth failed. Never armed — the morning brief itself tagged it a post-event read, not a pre-Fed entry. Discipline win. 0.47 SPY 740.96 (-1.25%) · -0.32R notl.
LRT SHORT SPY/QQQ middaymd-260617-1316-LRT-SPY-S FIRE Path-to-close short: the modeled bearish path (lose 747.85 → 745.34 → 737.76) ran post-2pm on the hawkish dots; SPY closed 740.96, between the cap and the 737.76 next support. No kill. (exec omitted — pre-FOMC 5m ATR understated the post-event stop.) 0.47 SPY 740.96 (-1.25%)
SBD LONG SPY/QQQ/SMH/IWM middaymd-260617-1316-SBD-SPY-L VOID Path-to-close long: two kills tripped post-2pm — hawkish_warsh AND spy_loses_745.34 (closed 740.96). The modeled upside path (vwap 750.36 → 752.15 → 755.44 → 756.68 record) inverted into a hawkish fade. 0.47 SPY 740.96 (-1.25%)

Backfilled — prior-date carries closed on the same Warsh binary (6/15 Sundae + 6/15 & 6/16 Curds)

Setup · sidOutcomeRNote
LRT S sun-260615-LRT-SPY-SFIRE+0.32RWeekly-frame level-rejection short; fired on hawkish dots + breadth-fail.
LRT S mm-260615-LRT-SPY-SFIRE+0.32RMon curd carry; same mechanic.
LRT S mm-260616-LRT-SPY-SFIRE+0.32Rlvl was 757; SPY high 752.15 never tagged the 755-760 supply → fired via the alternate failure-back-below-745.34 path.
SBD L sun-260615-SBD-SPY-LVOID-0.32R notl.Wall-of-worry long; hawkish_warsh kill, never armed. Discipline win.
SBD L mm-260615-SBD-SPY-LVOID-0.32R notl.Mon curd carry; same resolution.
SBD L mm-260616-SBD-SPY-LVOID-0.32R notl.Tue curd carry; same resolution.
Lens A clean, coherent split: all five LRT level-rejection shorts FIRED (+0.32R) and all five SBD wall-of-worry longs VOIDED (never armed) on the same hawkish tape — the print that fired the short kept Scott out of the long into a -1.25% SPY day. One honest caveat for the edge: the only winning side was an index short, adjacent to (not central to) the documented intraday-long / beaten-down-bounce playbook.

02Calibration

0.290Rolling Brier46 eligible · 0.25 = random · lower better
41.3%Hit rate19 FIRE / 46 eligible
46Calibration-eligible48 validated · 2 NO_EVIDENCE excluded

Outcome mix (48 validated): 19 FIRE · 16 VOID · 11 MIXED · 2 NO_EVIDENCE. The headline Brier ticked up 0.283 → 0.290 — the first uptick in nine sessions — but for a constructive reason (see the trend note): the winning side was systematically under-priced.

Pattern performance (eligible only)

Pattern · dirnHitBrierRead
LRT · Short (level-reject top)1369%0.400Jumped 50% → 69% on today's 5-for-5. Now the best-hitting pattern — but the worst Brier: it keeps firing at low assigned prob (~0.30). Under-confident, not wrong-direction.
MS · Long (momentum scalp)1136%0.294Unchanged this run. Highest-volume pattern, weakest hit — continuation longs keep dying on event/kill risk.
SRB · Long (energy reclaim)560%0.293Clean two-session-stabilization entries; still a real edge.
SBD · Long (wall-of-worry)50%0.153New this run: 0-for-5, but every one a discipline-win VOID (never armed), priced low — the best Brier on the board. Correctly low-confidence.
SBD · Short (breadth-divergence)40%0.079The old 0-for-4 chain — the chronic mechanism-leak pattern, but priced so low it barely dents the Brier.
Small-sample (n≤2): GFD-L 1/2, LRB-L 1/1, MS-S 1/1; GFU-S / SRT-S / VAB-L / VSR-L 0/1 each.

Calibration trend (rolling Brier by validation date)

05-290.339
06-080.333
06-090.326
06-100.311
06-120.301
06-150.289
06-160.283
06-180.290

Nine sessions of monotonic improvement (0.490 → 0.283) paused here: the 06-18 validation of the FOMC binary nudged Brier to 0.290. Cause is diagnostic, not regression — the 5 LRT shorts FIRED but were priced at just 0.30, so each booked a ~0.49 squared-error even while winning.

Mechanism leak6 right-direction / wrong-mechanism setups, unchangedthis run added none. The SBD-short chain (4x) + the 06-11 MS-SMH-L + the 06-16 MS-CYCL-L remain the standing set; today the LRT short fired on its own mechanic and the SBD long cleanly stood aside.
Lens The lever is now clearly probability calibration, not direction. The LRT level-rejection short is the top-hitting pattern (69%) yet carries the worst Brier because it is chronically under-priced into hawkish-tilted binaries — raise its conviction when the macro setup (a capped tape into a hike-priced event) aligns. The SBD-long voids, by contrast, were priced exactly right. No new mechanism leak this run is the quiet win.

03Tape & Rate Backdrop

RISK-OFF — hawkish-FOMC level rejection SPY rejected the 745.34 cap and closed -1.25% on a hawkish Warsh dot-plot; breadth failed, the dollar bid, gold broke — the bearish fork of the morning binary
IndexCloseDayNote
SPY740.96-1.25%Opened 751.29, ran to 752.15, then faded post-2pm to close below the 745.34 cap but above 737.76 support
QQQ722.51-1.01%Cushioned by a green semis complex; cap-weight tech otherwise sold
IWM289.88-0.75%Small-caps the relative outperformer — still red
RSP (eq-wt)208.99-1.50%Equal-weight underperformed cap-weight SPY — breadth narrowed/failed, the opposite of Tuesday
DIA516.30-0.99%Dow gave back Monday's record
VIX↻ refresh-requiredPremkt 16.2; not entitled on Massive and not printed in the close newsletters — post-FOMC vol read deferred

Dominant driver: the Warsh inaugural FOMC. Held 3.50-3.75% (12-0), but the dots turned hawkish — the 2026 cut was deleted, 9 of 12 voters now see a hike by year-end, 2026 PCE was marked to 3.6% (from 2.7%) and the projected year-end funds rate to 3.8%. Warsh stripped forward guidance, declined to submit his own dot, and cut the statement to ~130 words. "Persistently high prices are a burden for the American people, but the recent past need not be prologue." Rate path: the 2-year soared (front-end repricing higher); the 10y rose on the day (FRED last confirmed 4.43% on 06-16; the 06-17 close is ↻ refresh-required — the 4.60 kill line was not confirmed breached). TLT closed +0.16% (long-end ~flat) = a hawkish bear-flattener. Realized regime: the morning RANGE-into-binary read resolved to its bearish fork — a clean hawkish-event level-rejection, not the wall-of-worry reclaim.

Lens This was the hawkish tape the SKEW had been hedging for two sessions. Dollar up, gold down, front-end up, breadth failing — every cross-current pointed the same way for once, and the index respected the 745.34 cap precisely.

04Cross-Asset

AssetCloseDayRead
UUP (USD)28.18+0.90%Dollar bid hard — the cleanest hawkish-FOMC confirm
GLD388.60-2.27%Gold broke on higher real-rate expectations — ratifies the hawkish repricing
TLT (20y+)86.33+0.16%Long-end ~flat while the 2y soared = bear-flattener; not a duration-led risk-off
HYG (HY credit)79.73-0.38%Credit softened modestly — orderly, no stress (HY OAS still ~271bp)
USO (WTI)114.23-1.07%Spiked to 120 intraday then collapsed on the Hormuz reopening / Iran de-escalation — the disinflation offset to the hawkish Fed
Lens A 180 from Tuesday: cross-asset ratified the hawkish FOMC rather than arguing disinflation. Dollar up + gold down + front-end up is the textbook signature. Oil's spike-and-collapse is the one offsetting disinflationary thread, but it didn't rescue equities.

05Sector & Breadth (realized)

XLKTech-0.34%
XLFFin-0.55%
XLEEnergy-1.25%
XLIIndu-0.14%
XLYConsCyc-2.51%
XLPConsStp-2.23%
XLVHealth-1.46%
XLUUtil-1.33%
XLBMat-1.33%
XLRERE-2.51%
XLCComm-2.78%

0 of 11 green — a total breadth failure, the mirror image of Tuesday's 8/11. Equal-weight RSP (-1.50%) underperformed cap-weight SPY (-1.25%) and even QQQ (-1.01%): the selling was broad, not a megacap-only drag. The hardest hit were rate-sensitive and defensive — comm-services (-2.78), consumer-cyclical & real-estate (-2.51), staples (-2.23). The curiosity: XLK was the least-red sector (-0.34%) only because the big semis (AVGO +4.3, MU +2.2) propped it — tech "outperformed" on a down day purely on the AI-infra bid. S5FI/S5TH/$ADRN/$TICK/$TRIN not pulled this run (no scored leg gated on a numeric breadth trigger post-event); morning carry had S5FI ~62 — that healthy reading clearly rolled over into the close.

Lens Breadth is the cleanest tell that this was a genuine hawkish de-risk, not a rotation: when equal-weight leads the index down and 0/11 sectors hold green, there is nowhere to hide but the one idiosyncratic AI-infra pocket. That directly satisfied the LRT short's breadth-narrowing condition.

06Single-Name Movers

Illustrative only — close % confirmed (Massive) where shown; narrative from the Stocktwits Daily Rip close / Axios Closer (plaintext). Never trade plans.

Lens The hawkish dots drew a bright line: AI-infrastructure / memory (real-demand, balance-sheet-funded) was bought, while rate-sensitive credit, software multiples and the SpaceX-adjacent momentum cohort were sold. A textbook "higher-for-longer" rotation under a red tape.

07Morning Lens vs Reality

The Wednesday Early Bird Curd framed a RANGE-BOUND tape pinned into the Warsh binary and did the right thing: it surfaced both forks — a wall-of-worry reclaim LONG gated on a non-hawkish Warsh, and a level-rejection SHORT gated on hawkish dots + breadth failure. The binary resolved hawkish, so the short fork fired and the long fork voided — exactly as gated. The brief's own honest caveat ("only the bearish-sentiment leg is at a true extreme → a post-event read, not a pre-Fed entry") was vindicated: the long correctly never armed. Loop closed cleanly; the only thing the read under-did was conviction on the short, which deserved more than 0.30 into a hike-priced event.

Lens Discipline watch — tomorrow. Thursday 6/18 is a quad-witch expiration into a 3-day Juneteenth weekend (Fri 6/19 closed), with the Iran formal signing scheduled Friday over a closed market. That is the modified form of Scott's #1 documented leak (Friday-into-weekend-gap): flat-or-defined-risk into the Thursday close, no naked exposure carried over a 3-day, headline-exposed gap. The book did its job today by waiting on the binary — the same discipline applies to not over-staying the quad-witch.

08Execution Debrief

Realized execution on closed, scored setups — not a forward trade plan. Eight of the ten scored setups carried a gate-built exec block (the two midday legs omitted it — pre-FOMC 5m ATR understated the post-event stop). The two theses share identical swing parameters, so they are shown once each (×4 carries).

Setup (×4 carries)ProfileEntryStopTargetRealized R
LRT SHORT SPY — FIREDswing · 1d745.34759.23717.56+0.32R
SBD LONG SPY — VOID (never armed)swing · 1d745.34731.45757.00-0.32R notional

The short side, entered on the post-2pm failure back below 745.34 and marked at the 740.96 close, returned +0.32R per leg (stop 759.23 = the swing 1.5×ATR band, never threatened; target 717.56 still open). The long side never triggered — its -0.32R is notional, the loss the gate's hawkish_warsh kill avoided by keeping the book out of a reclaim long into a hawkish print.

Lens Realized execution on closed setups, not a forward plan: the level-rejection short, honored with the swing 1.5×ATR stop, paid +0.32R into the close while every wall-of-worry long was correctly skipped. The gate's discipline — short the rejection, don't buy the reclaim into a hawkish Fed — was the day's edge. The open question is whether the short should have carried more size, given how cleanly the binary was telegraphed.