Skip to content
Library / Notes & perspectivesOpen access · Freely available
← Notes & perspectives

AMD Pre-Earnings Drift: What the Last 14 Cycles Show

Chart Library · · 5 min read

Dated article. This writing is retained with its original publication date. Product details, third-party prices, and technical guidance may have changed. Chart Library is now free; consult the current API documentation and methodology. This article is not a study verdict.

AMD's Pre-Earnings Drift Profile

Across the last 14 AMD earnings reports, the 5-day pre-earnings drift has been signed correctly (matched the eventual reaction direction) in 9 of 14 cases — a 64% hit rate. That's notably stronger than the large-cap semis average (typically 55-58%), and consistent with AMD's status as a consensus AI name where positioning shifts visibly ahead of prints.

The drift effect is asymmetric across regimes. In the 8 quarters since the AI-capex cycle accelerated (early 2024 onward), the drift hit rate is 6 of 8 — even cleaner. In the 6 quarters before that, it was 3 of 6, essentially noise. The cohort retrieval tends to match AMD's current pre-earnings shape against the post-2024 cluster, which is the right reference.

The Current Setup

Heading into Tuesday's print, AMD's 5-day pre-earnings chart shape pulls a cohort of analogs that includes its own Q3 2024 and Q1 2025 pre-earnings setups, plus cross-ticker matches from NVDA's same-window prints. The cohort's 5-day post-earnings median is +2.1%, IQR [-3.8%, +7.4%], with a 60% directional win rate.

Within the cohort, the strongest separator between winners and losers is hyperscaler capex commentary in the prior 30 days. Cohort members that printed during periods of upbeat hyperscaler capex commentary returned a 5-day median of +4.6%; members during cautious commentary returned -1.1%. The current backdrop maps to upbeat (per the most recent META, MSFT, GOOG calls), which is a tailwind for the bull case but not a guarantee.

  • Cohort n=300, 5-day median +2.1%, win rate 60%
  • Pre-2024 drift hit rate: 3/6 (noise). Post-2024 drift hit rate: 6/8 (75%)
  • Strongest within-cohort separator: hyperscaler capex commentary in prior 30d

What To Watch Tuesday

Three signals get the most weight in the post-call analog retrieval: the MI300 / MI325 data center GPU revenue trajectory, server CPU share commentary (vs Intel), and the gross margin guide. The post-call shape that emerges in the after-hours and Wednesday open is what matches cleanly to the AI-capex-cycle cluster.

Agents retrieving against AMD's earnings should defer until Wednesday morning's price action settles. The pre-open futures and AH price discovery is too thin to produce a clean cohort match.

Wednesday morning, search AMD on chartlibrary.io to see the post-earnings analog cohort and the 80% calibrated forward band.

Related reading

MCD Q1 2026 Earnings: What 12 Years of Reaction Cohorts Say

McDonald's reports Q1 2026 before the open on May 7. Pulled the cohort of MCD earnings analogs back to 2014 — comp-store-sales beat vs. miss is the single feature that flips the 5-day forward distribution from positive to negative.

DDOG Q1 2026 Earnings: The Observability Cohort Going In

Datadog reports Q1 2026 before the open May 7. The cohort of DDOG and adjacent observability/devtools earnings prints shows that net-revenue-retention is the single feature that separates +6% follow-throughs from -8% fades.