Big Tech's Earnings Almost Never Move Chip Stocks Like This

By Tyler Cheves

Microsoft reported last night and is up 14% this morning. The chip complex came with it: Applied Materials 15%, Micron 13%, Marvell 11%, none of which reported anything themselves. Across the five of seven liquid suppliers with no print of their own this week the median move is 12.9% against 1.4% for a beta-matched group of non-suppliers, a gap of 11.5 points at 10:02 ET. In seventy megacap reaction days since March 2021 the widest that gap has ever been is 7.4.

Every megacap reaction day since March 2021, as the suppliers' median move minus a beta-matched control's. The swarm sits on zero. Today sits off the end of it. Source: ORATS /datav2/hist/dailies and /datav2/hist/earnings, intraday reading from /datav2/live/summaries.

The typical print does nothing

The median reaction day is a coin flip. Suppliers beat the matched controls by -0.14 points and only 46% of the seventy events are positive at all. On a ratio basis the residual is -0.003, the 48th percentile of a placebo drawn on random days. Read that as a failure to detect rather than a proof of zero: with seven suppliers against six controls the interval runs [-0.195, +0.395], so anything under about 0.2 points is invisible.

Microsoft's own record: 4.8% average implied earnings move against 4.4% delivered. The table runs to a pending October column, and last night's print is not ingested yet. Source: ORATS /datav2/hist/earnings and /datav2/hist/dailies.

Two things that make the test honest

The controls are matched on measured beta, inside the suppliers' own 1.49 to 1.81 band, because a chip stock moves more than the market on any busy day. Against an unmatched wide-beta group the suppliers appear to gain 0.08 points, and that gap is the mismatch.

The second is timing. Megacap earnings week is chip earnings week: 81% of these days have a supplier reporting within a session either side, 64% a control. Dropping whole days leaves six events, so instead any name reporting that day comes out of its group. That rule runs on today, which is why Lam Research and KLA sit outside the 12.9% despite moving more than anything inside it: they reported on the 29th and the 28th.

Micron's next print is fifty-five days out, and its 30-day implied matches its ex-earnings implied at 95.25, so nothing in that vol is its own event. The move is entirely someone else's news. Live tick 14:44 UTC, against 13.4% in the 10:02 ET snapshot the panel uses. Source: ORATS /datav2/cores.

What today actually is

The quiet end of the cohort. Broadcom's term structure reads 51.35 at 30 days against 51.98 at a year, flat, with its own print thirty-five days out and nothing inside the 30-day window. It still moved 4.0% on this tick. Source: ORATS /datav2/cores.

I expected the fat right tail to sit in the capital-expenditure guidance quarters. It does not. January and February prints are the worst block of the year, a median of -1.26 points and positive on only 37%. July and August run -0.20 and 47%, everything else +0.02 and 50%. No month tells you a big one is coming.

Today is an outlier with no rule attached, and with the session still open it is an intraday reading, not a close. The premium is not in owning chip options through every megacap print, and not in a calendar. Rebuild the check in the Earnings and News tab: pull a customer's implied-versus-delivered record, then confirm its suppliers are not reporting the same week.

$MSFT $META $LRCX $AMAT $MU $AMD $AVGO $KLAC $MRVL

#Semiconductors #EarningsVolatility #BigTech #EventStudy #OptionsResea