Why This Exists

The user flagged that Donald Trump has become a market-moving policy and stock-narrative catalyst. Because he is the sitting president, the right model is not "random FinTwit pump" and not "automatic truth." It is a small event-study lane:

• capture direct @realDonaldTrump posts as primary-source event tape;
• separate direct posts from secondary claims and resurfaced clips;
• map mentioned companies, sectors, or policy beneficiaries to tickers;
• compare 1D / 3D / 5D tape against obvious confounds like earnings, analyst upgrades, and sector beta;
• promote only if the signal becomes repeatable or decision-shaped.

Current State

As of the first 2026-06-01 pass, the direct last-month Trump scrape produced two current May posts, neither a clean ticker-specific primary receipt. The strongest ticker evidence is secondary or resurfaced:

• $PLTR: secondary claim that Trump explicitly mentioned one ticker and owns it.
• $DELL: secondary claim that Trump repeatedly told people to buy Dell; heavily confounded by Dell earnings.
• $IBM: Sherwood reported a Barclays initiation plus a resurfaced 2025 Trump clip saying the stock would go up.
• $INTC: secondary "U.S. government / Trump ownership" framing.
• $NOK: useful negative control. The stock rerated hard, but the direct Trump-pump Nokia receipt did not verify.

Operating Rules

1. Direct post > resurfaced clip > secondary source. Label the tier every time.
2. Media-only direct posts are incomplete until transcript/OCR lands. Do not infer the event from engagement alone.
3. Price truth comes from market-engine OHLC, not the tweet. Every candidate gets same-day, next-day, 3-day, and latest-window returns.
4. Confounds are mandatory. Earnings, analyst actions, 13F/government ownership, and sector beta must be named before calling a move "Trump-driven."
5. NOK stays the guardrail. A stock can rip while the Trump story fails. That is evidence against overfitting the pattern.

Graduation Criteria

Promote from thread to scan/perspective only if at least one of these happens:

• direct @realDonaldTrump post names a ticker/company and the stock moves materially within 1-3 sessions;
• three or more direct policy posts create a repeatable sector basket response;
• a secondary/resurfaced Trump clip demonstrably moves a stock after controlling for obvious same-day news;
• the event-study produces enough examples to build a recurring scan rule.

Next Actions

• On the next capture sweep, run the desk's own tooling.
• For any media-only Trump post with market relevance, fetch transcript/OCR before routing the event.
• Append new candidates to an event-study table with 1D / 3D / 5D moves and source tier.
• If a clean direct-post example appears, emit an investigation or scan note rather than burying it in capture.