trump tariff tone
What it checks
When Trump posts heavily about tariffs or China, we test a one-day-later move on every ticker and let the harness discover which ones actually react.
Mechanism
Trump tariff- and china-classified posts on Twitter (2017-21) and Truth Social (2022-) drive next-day moves in importers, defense primes, and domestic manufacturers. Aggregate count of policy-classified posts per day is z-scored over a 60-day window; on z-spike days we broadcast a long signal to every ticker and let DSR / PBO / multitest gates surface the genuine cross-section.
Signal rule
Daily count of Trump posts classified as tariff or china by political_nlp, z-scored over 60d, fires at z >= 1.0/1.5/2.0, holds 1/3/5/10d, broadcast.
Data dependencies
daily_pricesAdjusted-close OHLCV for every US-listed ticker; primary price feed.
political_postsWorker data table, see services/worker schema.
political_posts_nlpWorker data table, see services/worker schema.
Expected edge
- Paper window
- 2017-present (TT+TS stitch)
Cross-section effect; substrate gates surface 5-50 tickers with real attribution.
Related families
trump post volumePolitical / SpeechPure post-volume attention signal, no content classification. Spikes in daily Trump post count z-score correlate with tape-wide vol expansion + breadth contraction. Broadcasting lets the harness separate ticker-specific reactivity from generic noise.
trump company mentionPolitical / SpeechWhen Trump names a specific ticker (cashtag or company name), the post is a directional event. FinBERT sentiment on the post text drives a long (positive) or short (negative) 1-3-5d hold. Attribution is unambiguous; the substrate-gate Bonferroni penalty for this family is much lighter than the broadcast variants.
Explore trump tariff tone on alphactor.ai
See which tickers this family is currently firing on, with live signals and rankings.