AMZ.US (AMZN) — OVERWEIGHT, Conviction 3/5 as of 2026-05-06 | Current price: $273.55 | Entry zone: $248–$258 | DO NOT BUY AT CURRENT PRICE

---

Executive Summary

• Rating is OVERWEIGHT at conviction 3/5, sized at 3.5% NAV — but the entry has not opened yet. AMZN closed at an all-time high of $273.55 on 2026-05-05. Both the bull and the bear researcher converged on the same operational conclusion: buying here is the wrong trade. The rating applies to a staged limit-order entry into a $248–$258 pullback, not to the current price. If the pullback never arrives, the orders cancel and the trade is skipped.
• The thesis rests on three pillars that survived adversarial debate intact: (1) a documented mid-cycle margin expansion still far below peers (operating margin 2.4% → 11.5% over three years, vs MSFT at 46% and GOOGL at 36%); (2) AWS re-accelerating to +28.43% YoY with CEO Andy Jassy describing demand as supply-constrained, not demand-limited; (3) an uncrowded sentiment backdrop — all five crowded-trade danger signals are absent despite an all-time high print.
• The binary test is Q2 2026 earnings on 2026-07-31. The position deliberately carries through that date because the AWS print is the thesis test. AWS must deliver at or above 22% YoY growth for the thesis to hold. Below 22%, exit on the 2026-08-01 open regardless of price.
• Trade geometry: blended entry ~$251, hard stop $234 (daily close), targets $264 / $279 / $295–$310. Blended R/R of approximately 1.51:1. NAV-at-risk on stop is 0.24% — well inside the 1.0% per-trade ceiling.
• The risk committee did not agree on size. The aggressive voice wanted 6.5% NAV; the conservative voice wanted 3.0% NAV; the neutral voice recommended 3.5% NAV. The portfolio-manager accepted the neutral recommendation as the binding default per the risk=neutral user parameter.
• This is the first AMZ.US decision in this system. No calibrated hit-rate exists for this ticker. Conviction is appropriately capped at 3, not 4, reflecting genuine uncertainty on entry timing even as the directional case is solid.
• The probability that either entry leg fills within the 8-week order window is estimated at 35–45%. A price that never pulls back is a missed trade, not a failure — the plan does nothing rather than chase.

---

Trade Parameters

---

Investment Thesis

Pillar 1 — Margin runway is real and not yet priced

Amazon's operating margin expanded from 2.4% in 2022 to 11.5% TTM — a 9 percentage-point improvement in three years (per fundamentals.md). MSFT runs at 46% operating margin; GOOGL at 36%. At EV/EBITDA of 19.2x versus a peer median of 19.0x, the market is pricing AMZN at parity despite this gap, as though the margin expansion has already played out. It has not: if AWS and Advertising reach a 35% blended operating margin over the next five years while retail stabilises, consolidated operating income roughly doubles from current levels. The advertising segment alone — at a ~$57B annual run rate growing 22% YoY with near-zero incremental cost — is embedded in what the market prices as a retailer. The FCF picture is genuinely negative (-$2.5B TTM) but is entirely a capex deployment story: operating cash flow is $148.5B and growing 28% YoY. The AR/Revenue flag is yellow, not red, graded consistent with B2B invoicing mix shift rather than a collection problem (per fundamentals.md, Red Flag Scan).

Pillar 2 — Supply-constrained AWS re-acceleration

AWS grew revenue +28.43% YoY in Q1 2026, re-accelerating after a 2024–2025 plateau (per news.md). On the earnings call, CEO Andy Jassy described cloud demand as "outpacing our ability to supply" — a supply constraint, not a demand concern. This framing is a forward earnings revision signal: new data-centre capacity coming online in H2 2026 is a direct revenue tailwind that consensus is likely understating. Jassy's $10M open-market share purchase on 2026-04-20 (Form 4, per news.md) is non-routine — he does not buy frequently, and the timing ahead of the Q1 print makes it more meaningful. The crowd's concern that Google Cloud grew 63% YoY is legitimate but partially a base-effect comparison: AWS runs a $37.6B quarterly revenue base against Google Cloud's $20B, and AWS's own supply constraint is suppressing the reported growth rate, not structural share loss.

Pillar 3 — Fundamentals-driven, not crowd-driven

The sentiment regime is NEUTRAL — a score of +0.03 is statistically indistinguishable from zero (per sentiment.md). All five crowded-trade danger signals are absent: no euphoria, no capitulation, no main-character volume spike, no double-euphoria overlay, no Seeking Alpha echo chamber (SA bullish rate is 40%, versus the 85% danger threshold). The crowd debate is substantive — AWS relative growth versus peers, capex risk, geopolitical exposure — not emotional. An uncrowded setup on a large-cap with three surviving thesis pillars and a constructive insider-buy creates a specific asymmetry: if AWS Q2 validates supply-constrained re-acceleration, the rotation-from-AMZN thesis fails and the stock outperforms from an under-positioned field.

---

Risk Committee Summary

Three voices reviewed the trader's 4.0% NAV base plan. All three affirmed the stop at $234 and the target structure unchanged. They split only on size.

Aggressive voice — recommends 6.5% NAV. At 4% NAV, the 1-day 99% VaR is 0.16% of portfolio — less than one-sixth of the 1.0% per-trade ceiling. The stop at 3.76 standard deviations from entry means random noise will not trigger it. The three thesis pillars survived adversarial debate; quarter-Kelly (4%) is the floor, not the ceiling. At 6.5% NAV the holding-period 99% VaR reaches 2.36% — just above the 2% threshold at which this voice stops pushing higher — which is why 6.5% is the ceiling rather than 7% or more. The worst-case earnings gap at 6.5% NAV is -0.91% NAV, inside every hard limit in the framework (per risk-aggressive.md).

Neutral voice — recommends 3.5% NAV; accepted as binding by portfolio-manager. 4% NAV places the 1-day 99% Expected Shortfall at 0.97% using the conservative HV252 (31% annualised) — approaching the 1.0% soft ceiling. The Q2 2026 earnings binary is a structural tail not captured by parametric VaR. At 3.5% NAV, the 95% VaR, 99% VaR, and 99% ES all fall inside the neutral band simultaneously, and 0.5% NAV reload capacity is preserved post-Q2 if the binary resolves constructively (per risk-neutral.md).

Conservative voice — recommends 3.0% NAV (or 4.0% with a $240-strike put expiring 2026-08-15). The hard stop at $234 is a daily-close stop and provides zero gap protection on an earnings event. A -12% earnings miss from a $250 entry opens near $220, producing a $30-per-share loss versus the modelled $16 stop-loss — approximately double the stated NAV-at-risk. At 4% NAV, a -12% gap generates 0.48% NAV loss, marginally exceeding the average projected win of 0.43% NAV. At 3.0% NAV the same gap produces 0.36% NAV, inside the average win. The non-fill probability (55–65%) is also underweighted in the base plan (per risk-conservative.md).

Portfolio-manager reconciliation — 3.5% NAV confirmed. The risk=neutral parameter makes the neutral voice's recommendation the binding default. The conservative voice's gap-risk arithmetic is specific and valid between 4.0% and 3.5% NAV, but does not bind below 3.5%: at that size the worst-plausible -12% gap loss (0.42% NAV) fits inside the blended projected win (0.36% NAV), and the 99% ES falls to 0.85% NAV. The $240-strike put hedge was acknowledged and declined — alpha drag is not offset by tail protection that 3.5% sizing already delivers. The aggressive voice's unused-headroom argument is internally consistent but is asking the system to run at its upper size band on its first US-equity call with zero ticker-specific calibration data, which is not the right moment (per portfolio-decision.md).

---

Bull/Bear Debate Summary

The research-manager's verdict: bull won on direction; bear won on entry mechanics. Both wins are reflected.

The bull's strongest surviving point was the margin-runway-at-parity valuation argument: Amazon is in a documented mid-cycle margin expansion, its AWS just re-accelerated to +28% YoY with supply-constraint framing, and it trades at peer-parity EV/EBITDA (19.2x vs 19.0x) as though the inflection has already fully played out. The bull also successfully defended the OBV trend (+150.85% over 50 sessions per technical.md) as confirmation of institutional accumulation, not distribution, and the NEUTRAL sentiment regime as evidence that the setup is fundamentals-driven rather than crowd-momentum-driven.

The bear's strongest surviving point was that the technical analyst's own timing tag is EXTENDED_CHASE_RISK_HIGH with a suggested entry of $258, while current price is $273.55. RSI at 80.51, 20-day ROC of +27.96%, and the technical analyst's explicit warning of "8–15% historical mean-reversion pullback" on such moves make a market-order entry today the wrong trade even if the direction is right. The bear's two weakest moves — framing AWS +28% YoY as deceleration (the analyst pack defines the deceleration threshold at below 15%) and treating the yellow-flag AR expansion as a primary thesis pillar — did not survive contact with the evidence.

By round 2, both sides had converged on the same entry zone with different conviction labels: the bull called it "LONG via staged entry into $248–$258"; the bear called it "the correct trade if $256 holds." The research-manager committed to that zone, awarded direction to the bull, and cut conviction from the bull's claimed 4 to 3 on the entry-mechanics concession (per research-verdict.md).

---

Analyst Pack Overview

---

Invalidating Assumptions

1. AWS Q2 2026 growth below 22% YoY — announced 2026-07-31 This is the bull's stated must-hold assumption. Below 22% flips the verdict to NO-TRADE and requires a full exit on the 2026-08-01 open, regardless of where price trades. Below 20% flips the verdict to SHORT and triggers the news analyst's projected 10–15% immediate repricing (per research-verdict.md §6, news.md). This binary sits on day 61 of the 84-day horizon and is the single most load-bearing assumption in the pack.

2. Daily close below $256.16 on greater than 1.2x average volume within the next 4–6 weeks This invalidates the high-tight flag structure that the entry zone depends on (technical.md, Pattern; research-verdict.md §6). The LVN gap from $248 to $256 means price moves rapidly through that zone — a high-volume close below $256.16 before entries fill cancels the unfilled orders; after entries fill, it is an event stop requiring exit on the next open even if $234 has not been reached. A move below $240 breaks the flag outright; below $227.41 (SMA50) the strong-uptrend regime flips to topping.

3. Third consecutive quarter of AR/Revenue ratio expansion in Q2 2026, with ratio above 41.6% The fundamentals analyst grades the current AR/Revenue expansion (34.8% → 41.6% YoY) as yellow — consistent with B2B invoicing mix shift but requiring confirmation as non-recurring. Two consecutive yellow quarters is a flashing warning. Three consecutive quarters converts the flag to red and reopens the revenue-quality question, which would flip the fundamentals verdict to NEUTRAL and invalidate Pillar 1 (per fundamentals.md, "What Would Change My Mind").

---

Catalyst Calendar

The June secondary events are tape risk to manage through. The 2026-07-31 Q2 print is the only event that determines whether the thesis survives.

---

Data Quality Notes

• All news and fundamental figures are knowledge-base synthesis, not live API pulls. The Q1 2026 figures (AWS +28.43% YoY, EPS $2.78, revenue $181.5B) triangulate consistently across news.md, sentiment.md, and fundamentals.md, but a live filing pull would sharpen confidence (per research-verdict.md, Methodology Notes).
• Sentiment coverage is partial. StockTwits was blocked; the bull/bear ratio is null. Twitter yielded only 3 investment-relevant results from 49 queries (95%+ affiliate spam). Fear & Greed (61) and Google Trends (+0.9z) are estimates. The NEUTRAL regime conclusion is robust — lexicon and LLM methods converged independently — but raw scores carry wider uncertainty than normal (per sentiment.md, Caveats).
• Beta estimates differ across the three risk reports (aggressive: 1.25; neutral: 1.35; conservative: 1.30). The neutral voice's 1.35 is the binding input per the PM-selected voice. Scenario P&L outputs vary modestly as a result.
• Options data is absent. No IV percentile, put/call ratio, or skew data was pulled. The earnings gap risk is sized off historical move distributions (-8% to -14% range), not current options pricing. A live options pull would sharpen the gap-risk arithmetic before entry.
• No prior AMZ.US decisions exist in this system (per memory.md). The 52% conviction-3 win-rate prior is the system default, not a ticker-specific calibration. All R/R and expectancy calculations should be read accordingly.

---

This is research, not trading advice. Reports: data/reports/AMZ.US/2026-05-06/