Two near-identical surprises drifted in opposite directions.

How to read itBeat rate is the scorecard. Average five-day post-earnings drift is what actually happened to holders afterward. Grouped by drift pattern, not by sector.

Every figure is an average over eight historical quarters, drawn from live post-earnings distributions via the GammaQC engine. An average is a description of the past, not a forecast.

Transmission chainInflation print → front-end yields → duration → multiple. A hot print lifts the implied policy path, raises the discount rate applied to cash flows dated years out, and compresses long-duration multiples hardest; a cool one relieves it. The structural point: a hawkish path does not need to produce a selloff to matter — it caps multiple expansion. A committee that has just moved in the hawkish direction sets the ceiling every rate-sensitive reporter prints under this week.

What just happenedFOMC, Sep 15–16: the committee voted 12–0 to raise the federal funds target range by 25 basis points to 3.75%–4.00%, the first increase since 2023. The next FOMC decision is Wednesday Oct 28 at 2:00p ET (meeting Oct 27–28). [CNBC, Sep 16 2026; Fox Business, Sep 16 2026; Federal Reserve Board FOMC calendar; FedRateCalc; Finance Calendar]

What is scheduledA light data week and a heavy speaker week, with Fed officials making no fewer than ten appearances. Mon Sep 21: Chicago Fed National Activity Index (August). Tue Sep 22: Philadelphia Fed Non-Manufacturing Survey, 8:30a ET; Richmond Fed Manufacturing, 10:00a ET. Wed Sep 23: S&P Global flash manufacturing and services PMIs, 9:45a ET. Thu Sep 24: initial jobless claims, 8:30a ET; new home sales, 10:00a ET. Fri Sep 25: final September University of Michigan consumer sentiment, 10:00a ET. The preliminary September reading fell to 47.8 against 51.0 expected, its second straight monthly decline, and year-ahead inflation expectations rose to 4.6%. The final survey shows whether those readings held. [Kiplinger, economic calendar for September 21–25; eOption Weekly Event Calendar 09/21–09/25/2026; Xinhua, Sep 12 2026; Seeking Alpha; University of Michigan Surveys of Consumers]

What is NOT scheduledNo CPI, no FOMC decision and no PCE this week. The next CPI release, covering September data, is Wednesday Oct 14 at 8:30a ET. The August PCE price index arrives Wednesday Sep 30 at 8:30a ET, next week, together with BEA's annual update of the national accounts. The committee's decision is already behind the market. This week the front end receives speeches and second-tier data, not a tier-one inflation print, and the absence is itself worth writing down. [U.S. Bureau of Labor Statistics CPI release schedule; Nowflation; U.S. Bureau of Economic Analysis; FedRateCalc]

DisciplineA scheduled event is a fact to report. The market's reaction to it is not knowable in advance, and nothing in this section is a view on how any release or speech resolves.

Magnitude does not order driftWalk the two largest surprises on the board. THO carries the largest average surprise at +211.0% and drifts −2.28%. BB carries the second largest at +192.4% and has the largest positive drift at +6.98%. The surprise magnitudes are close, the drifts have opposite signs, and the gap between them is 9.26 points.

Run it once more at the small end: KBH averages a +5.7% surprise and drifts +4.40%; GIS averages +5.4% and drifts −2.06%. A three-tenths-of-a-point difference in surprise sits beside a 6.46-point difference in drift. And CBRL's +101.7% average surprise buys +0.84% of drift, less than PAYX gets from a +1.3% surprise.

The thesisBeat rate and post-print direction are approximately uncorrelated in this cohort. Four names beat in seven of their last eight quarters — GIS, CTAS, PAYX and BB — and their average drifts are −2.06%, +1.10%, +1.45% and +6.98%. The same scorecard spans all three drift groups. "They all beat" tells you nothing about the move. The scorecard measures the company against the published Street number; the drift measures what was already in the price before the number arrived.

Model versus StreetThe published Street numbers run from AZO at 54.08, COST at 6.53 and SNX at 4.64 down to BB at 0.04. The engine's own estimate field returned empty across all twelve names in this cohort, so no model-versus-Street divergence is published this week. The desk will not manufacture one.

The structureImplied volatility is bid into a print because the event date is known and the outcome is not. Once the number is public, the uncertainty that justified the premium is resolved and implied volatility collapses toward realized. That collapse is mechanical and scheduled, and whoever holds the premium across the event pays for it. A directionally correct option can still lose money through the print: the payoff has to clear both the move and the crush.

Realized versus impliedThe question is never "will it move." It is whether the realized move exceeds what the pre-print premium already charged. This week's band is tighter than last week's. The cohort's average five-day moves run from −3.48% to +6.98%, and nine of the twelve names sit inside a −3.48% to +1.45% band. A distribution centred that tightly is a difficult thing to pay a pre-event premium against. The measurable structure in this data sits in the post-print drift — the multi-day path after the uncertainty is resolved and the crush has already happened — not in the pre-print premium.

Risk vetoEvery figure in the table above is an average over eight observations. Eight is a small sample, the table carries no dispersion measure, and a mean says nothing about how wide the distribution around it is. BB's +6.98% average is entirely consistent with a set of quarters that included a large negative one. A distribution is not a forecast, an average is not a prediction, and none of this is a probability that any specific future quarter behaves like the mean.

Thread 1 — Trade: the Washington summitThis is the week's most important dated item. The President is scheduled to host Xi Jinping at the White House on Thursday Sep 24, the leaders' second summit of 2026 after May's Beijing summit. Reporting has Xi arriving late on Sep 23 and departing Sep 25, skipping the UN General Assembly. The agenda is reported as trade, Taiwan and AI. On Sep 17, Bloomberg reported that the U.S. was expected to postpone new tariffs on Chinese excess capacity until after the summit. Separately, the 100% Section 232 tariff on patented pharmaceuticals and their active ingredients comes into force for companies not already covered on Sep 29, the Tuesday after this week. That is also the effective date for the prohibitions on certain Canadian products under the Sep 8 proclamations. Canada's counter-tariffs of up to 50% on $27.6 billion of U.S. goods, including steel, appliances and agricultural equipment, are already in effect. [The National, Sep 21 2026; CSIS, Trump-Xi 2026 Summits; Bloomberg, Sep 17 2026; Shapiro, Trade Tariff Updates; Wiley, Trump Administration Tariff Tracker; Trade Compliance Resource Hub, Sep 10 2026; Invezz, Sep 21 2026]

Read — now / laterNear term, and specific: SNX reports Thursday before the open, a global electronics distributor printing hours before the summit, and COST reports Thursday after the close, a global import buyer printing after it. AZO (Tue, before the open) sells an imported parts catalog, and WOR (Tue, after the close) processes steel into a market where Canada's 50% duties on American steel are already live. Structurally: a deferred tariff is still a pending tariff. Duties are an input-cost and a volume variable at once, and guidance language registers a schedule change quarters before reported margin finishes absorbing it. Any agreement at the summit would tend to lower the input-cost path for importers; any breakdown would tend to bring the deferred measures back onto the calendar. The desk reports the schedule, not the outcome.

Thread 2 — EnergyCrude has spent much of 2026 at or near triple digits. The live sentiment feed carries the tone. The Motley Fool, Sep 18 2026 — "2 Stocks to Buy if You Think $100 Oil Will Last." 24/7 Wall St., Sep 18 2026 — "Exxon's 11 Million-Gallon Refinery Goes Dark. How High Can Midwest Gas Prices Go?" Seeking Alpha, Sep 21 2026 — "Chevron: The Bull Case Goes Beyond $100 Oil." The U.S. Energy Information Administration's Short-Term Energy Outlook forecasts Brent averaging around $90/b in the second half of 2026 and about $74/b in 2027 as production rises and inventories rebuild. [The Motley Fool, 24/7 Wall St. and Seeking Alpha, via the GammaQC live sentiment feed for XOM/CVX; U.S. Energy Information Administration, STEO]

Read — now / laterNear term: fuel and food inputs sit directly in the cost lines of DRI (Thu, before the open), CBRL (Wed, after the close) and GIS (Wed, before the open). The preliminary September consumer-sentiment reading of 47.8 came with consumers anticipating greater pressure on household budgets amid rising fuel prices and trade tensions (Xinhua, Sep 12 2026), which puts the demand side of those same names in the frame. Structurally: sustained high crude feeds headline inflation, then breakevens, then the discount rate, and the committee has just shown it will act on that chain. Any re-escalation would tend to push it in the same direction; any durable de-escalation would tend to relieve it.

Thread 3 — Rates: housing and big-ticket creditThe hike is the transmission, and two of this week's reporters sit directly on it. KBH reports Tuesday after the close, the first homebuilder print since the Sep 16 increase, two days before Thursday's 10:00a ET new-home-sales release. THO reports Tuesday before the open, a recreational-vehicle maker whose buyers finance the purchase. Near term: both prints carry guidance written under a financing cost that just rose. Structurally: a policy path that points higher raises the carrying cost of every financed durable, and that shows up in order books before it shows up in revenue.

Thread 4 — Defense & rearmamentThe sector's news flow is dense and dated. Zacks, Sep 18 2026 — "Defense ETFs Stand to Gain as Trump Administration Signs JATM Deal." PR Newswire, Sep 18 2026 — "Germany, Lockheed Martin Celebrate Rollout of Germany's First F-35A, Marking New Era of German Airpower." GuruFocus, Sep 18 2026 — "RTX Slips 1% as Tariffs Raise FAA's Radar Bill," a direct point where the trade and defense threads meet. No marquee reporter this week is a defense prime, so this thread bears on the tape through the tariff channel rather than through a print. Structurally: rearmament is a multi-year budget variable, and procurement converts to revenue on a lag measured in years. [Zacks, PR Newswire and GuruFocus, Sep 18 2026 — via the GammaQC live sentiment feed for LMT/RTX]

Neutrality noteEvery item in this section is reported as a dated, attributed fact with business and market relevance only. Nothing here is offered as political or partisan commentary, and every forward-looking clause is conditional by construction.

Pre-printThe event date is known, so the exposure decision is a sizing decision, not a timing one. Size for convexity: with a known date and an unknown outcome, the loss can be bounded and defined in advance. Write down the invalidation level before the release. A level chosen after the number is public is a rationalization, not an invalidation level, and that distinction is the entire discipline. This week the discipline has a second clock: three of the marquee names print on the same Thursday as the summit.

The liquidity-trap signatureThis is a structural observation, not an instruction. The pattern has three parts that appear together: a company beats cleanly on the published number, the stock gaps green on the opening print, and Day-1 institutional drift volume runs negative, meaning size is being distributed into the gap rather than accumulated through it. When all three co-occur, the beat is functioning as exit liquidity: the good number supplies the bid that lets existing size leave. The tell is a gap and a volume flow pointing in opposite directions, which you can see in the tape but not in the headline.

Which names fit the archetypeOn record alone, GIS is the cleanest archetype on this board: seven beats in eight quarters, a +5.4% average surprise, and an average five-day drift of −2.06%. THO is the extreme version, with a +211.0% average surprise converting into −2.28% of drift, and WOR converts a +3.3% average surprise into the board's worst drift at −3.48%. COST fits a milder, different profile: six of eight beats and a +2.0% average surprise producing −1.51%, a strong scorecard that has historically bought no subsequent move, which is the priced-in signature rather than the distribution one. This describes eight historical quarters per name. It is not a claim about what any of them will do this week.

The standing vetoBeat rate does not imply direction. Four names on this board carry the same seven-of-eight record and land in all three drift groups. If your entire reason for holding something through a print is "they always beat," you do not have a thesis. You have a statistic that this week's table shows does not carry direction.