From Narratives to Trades: Our Investment Framework

Arthur: Tony Lyu

9/15/2025

I. Research-Driven: From Storylines to Tradeable Hypotheses

1. AI Value Chain Segmentation

We frame the AI ecosystem through a “multi-stage beneficiary” lens: hardware (compute, memory), infrastructure & cloud, SaaS, and downstream adopters. Each rung moves to its own cycle—hardware with capex, infra with contract lock-ins and compute allocation, SaaS with the durability of its model, and traditional industries with uptake rates and cost discipline.

2. Industry Substitution and Moats

In semis—the classic “picks and shovels trade” of the AI boom —we pit Nvidia’s architecture refreshes against Huawei’s optical interconnect leaps, while tracking Micron vs. SK Hynix in the bandwidth wars. The focal point: where physical interconnect limits collide with optical breakthroughs, and how memory bottlenecks cap model scale.

3. SaaS Divergence

Seat-based names (CRM, Adobe) face structural headwinds from AI substitution, while data infra plays (Snowflake, MongoDB) ride secular on-prem-to-cloud migration. Layer in AI features, and you get stickier customers and heavier query loads—a path toward growth with real durability. The open question: is their “narrative premium”  finally justified by fundamentals?

4. Emerging Payments & Digital Assets

In stablecoins and programmable rails, we map how regulation, monetary traits, and ecosystem embedding shape moats. The big call: whether compliant stablecoins evolve into programmable money and cross-border plumbing—or stall under adoption drag and fractured regimes.

II. Event-Driven: Policy and Catalyst Transmission

1. Policy Catalysts

Washington’s nuclear fast-track—from defense to data centers—shows how “policy is the new QE”，rewiring capital flows into equities. We build a transmission map of “policy → industry → performance” to flag tactical entry points.

2. Corporate Events

Nvidia’s inference launch, Oracle’s hyperscale AI deals, OpenAI’s capex burn—we slot these into competitive positioning, sentiment gaps, and expectation resets to craft trades with teeth.

3. Earnings Reviews

Names like Adobe and CRM: beyond headline beats/misses, we drill into cRPO, net new ARR, and AI feature adoption to test whether the AI narrative is bleeding into fundamentals.

III. Macro and Sentiment: Playing the Cycle, Not Following It

1. Macro Scenario Framework

Payrolls, CPI, wages, PMI—treated not as data prints but as scenario trees: low-unemployment/low-wage vs. high-unemployment/high-CPI. Each branch reframes the Fed’s rate path.

2. Policy Games & Independence

Fed board politics create a double edge: more dovish bias, less independence. That tension informs curve positioning—steepeners vs. flatteners—depending on which edge dominates.

3. Market Breadth & Positioning

We track equity breadth (stocks above 20/50-day MAs), UVIX, and retail vs. institutional positioning. The signal: are we climbing a wall of worry or trading on FOMO? Allocation follows—index beta, equal-weight tilts, or defense.

IV. Methodology: From Narrative to Execution

1. Narrative Deconstruction

We embed hot themes—AI stack, macro trees—into structured frameworks, filtering noise from alpha signals.

2. Trigger Discipline

Every trade is tethered to a “trigger–action–risk” map: policy enactments, contract signings, or earnings prints linked directly to sizing and stop-loss discipline.

3. Dynamic Review

Systematic daily/weekly reviews close the loop—ideas get validated, killed, or doubled-down—keeping research and execution welded together.