ψ Psi Quant Research

Signal Report · US

AMD HOLD

Composite decision report grounded in 1-XH strategy signals and a completed multi-agent debate (agents-3node-v1), generated at 2026-09-03 09:53.

Final Agent Decision

Action
HOLD
Weight
0%
Confidence
55%
Variant
xiaohan_gogo_v2

Reasons

Bull Case (Analyst)

- Technical signal is a clean buy (score 0.70) with multi-timeframe confirmation: 1h through 4h all bottoming, and daily blue fully above yellow — the gogo_detail shows the setup is aligned across horizons, not a single-frame fluke.
- Cross-sectional factor view is supportive: AMD ranks 29/252 (top 11.5%) with composite z = +0.38 in a BULL regime catalog. The dominant lifter is xh_gogo_score (+2.88), which directly corroborates the technical buy signal — the same momentum/mean-reversion factor driving the rank is what triggered the entry.
- Momentum_6m (+1.32) and beta_60d (+1.29) are also positive contributors, meaning AMD retains medium-term trend strength and upside beta — in a BULL regime, high-beta names with intact 6-month momentum tend to outperform as the cross-section rewards risk-taking.
- Fundamentals are not the problem: Q2 2026 was a beat-and-raise with record sales, and Q3 guidance of ~$13B midpoint implies 41% YoY growth — the selloff post-earnings (down 8%+) appears sentiment-driven, not business-driven, creating a potential dislocation for a long entry at $456 vs. insider sales at $514.91.

Bear Case (Analyst)

- The stock is in a clear downtrend: close at $456.2 sits well below ema_high_slow at $471.5, and the 7-day return is -4.08% with a 90-day slide — the technical buy is a counter-trend bounce attempt, not a trend reversal, and the gap between fast and slow EMAs (~15 points) shows persistent overhead supply.
- Cross-sectional draggers are telling: rsi_14 (-1.23), close_to_high_60d (-1.05), and bollinger_b_20d (-0.99) all negative — AMD is overbought on a relative basis, far from its 60-day high, and sitting in the lower Bollinger band. These are momentum-quality and mean-reversion factors that historically precede further downside in this regime.
- The cross-section did NOT place AMD in the production basket — despite rank 29/252, it missed the top decile cutoff. The composite z of +0.38 is modest; the signal is being carried almost entirely by xh_gogo_score (+2.88), which is the same factor already embedded in the technical signal — there is no independent fundamental or valuation factor confirming the long.
- News flow is negative: insider EVP Paul Grasby sold 15,000 shares at $514.91 (a 13% premium to current price), and multiple headlines flag persistent underperformance vs. peers despite the earnings beat. The Seeking Alpha piece explicitly notes AMD "has consistently underperformed broader markets and its semiconductor peers" — a red flag for relative-value longs.
- The post-earnings reaction (down 8%+ on a beat-and-raise) and the Sep 1 open down 3.14% suggest the market is pricing in something beyond Q2 — likely Musk's NVIDIA pledge (headline #7) which reframes the AI capex narrative away from AMD. In a BULL regime, a stock that cannot rally on good news is a short candidate, not a buy.

Devil's Advocate

COUNTER TO BULL:
- The "multi-timeframe confirmation" is meaningless if all timeframes are measuring the same lagging momentum factor — xh_gogo_score (+2.88) is doing 100% of the work in the composite z (+0.38). You are double-counting one signal: the technical score and the factor rank are the same coin, not independent corroboration.
- A rank of 29/252 (top 11.5%) is NOT a top-decile production signal — the basket cutoff is 25.2. You are celebrating a miss. If the factor model itself refuses to put AMD in the book at rank 29, your "supportive" cross-section is actually a rejection.
- The beat-and-raise narrative ignores that the stock fell 8% on the news and another 3% on Sep 1. You call it "sentiment-driven dislocation" — but the market is the only honest forecaster. A 11% two-session drawdown on good news implies the guidance was already priced or the market sees a structural (not transient) problem.
- Insider sale at $514.91 is not a "premium to current price" — it is a trailing reference. The EVP sold BEFORE the drop; that timing suggests he saw the Q3 guide or the Musk/NVIDIA headline coming. You are using stale insider data as a bullish anchor.
- Beta_60d (+1.29) and momentum_6m (+1.32) are regime-dependent tailwinds that flip to accelerants on the downside. In a BULL regime, high-beta names with 6-month momentum that are DOWN 4% in a week and below ema_high_slow are not "risk-on leaders" — they are the first to be dumped when the regime wobbles.

COUNTER TO BEAR:
- The 7-day return of -4.08% and 90-day slide are exactly what a bottoming pattern looks like after a post-earnings flush — you are describing the setup, not refuting it. The technical score (0.70) with 1h-4h bottoming is a leading indicator; your EMA gap is a lagging one.
- RSI_14 (-1.23) and bollinger_b_20d (-0.99) being negative means AMD is OVERSOLD, not "overbought on a relative basis" — you have the sign backwards. Low RSI and lower-band prints are mean-reversion BUY conditions in a BULL regime, not sell signals. Your own factor catalog contradicts your interpretation.
- The "missed top decile" argument cuts both ways: rank 29/252 with a composite z of +0.38 is a near-miss, and the ONLY reason it missed is the negative RSI/close-to-high drag — which are exactly the factors that mean-revert after a sharp selloff. The production basket is a point-in-time snapshot, not a forecast.
- Insider selling at $514.91 is a 13% premium to current price — that means the insider sold BEFORE the drop, and the stock has since de-rated. That is a lagging indicator of sentiment, not a fresh negative. If anything, the selling pressure from that transaction is already in the price.
- The Musk/NVIDIA pledge headline is a narrative, not a fundamental — AMD's Q3 guide of $13B midpoint (41% YoY) is a hard number. If the market is "pricing in something beyond Q2," you have not identified what that is beyond a headline. A stock that drops on good news can be a short — but it can equally be a gift if the drop is a liquidity flush, and your bear case offers no valuation or earnings revision to justify the downside beyond price action itself.

Technical Signal (1-XH)

action
buy
score
0.70
variant
xiaohan_gogo_v2
timestamp
2026-09-03T15:30:00-04:00
close
456.2
ema_low_fast
456.6
ema_high_slow
471.5
gogo_score
0.7
gogo_detail
1h_bottom,2h_bottom,3h_bottom,4h_bottom,daily_blue_fully_above_yellow

📈 Backtest Track Record xiaohan_gogo_v2 · signals simulated over available history (~3-month 30-min window, accumulating daily)

Trades
1
Win rate
0%
Avg / trade
-4.7%
Total (compounded)
-5%
best / worst
-4.7% / -4.7%
avg holding
9.0 days
open position
yes — marked-to-market -4.7%
Entry Exit Entry Exit Return Exit reason
2026-08-25 open 478.87 456.24 -4.7% open_position_mark_to_market

复盘 = each past xiaohan_gogo_v2 buy signal simulated: enter next bar, manage adds / adaptive sells / −25% hard stop. Demonstration only, not investment advice.

News & social (8 items)