Event Catalyst
Catalyst detection and surfacing — the system that attaches a why to every meaningful price move.
What this is
Most price moves don’t come out of nowhere — they ride catalysts. Event Catalyst is TRUE’s system for detecting those events early, classifying them, and explaining them simply. It’s what lets the chat say “SOL just broke a level because the FTX estate just unstaked X” instead of just the first half.
Pipeline
Event taxonomy
Every classified event carries one of six type tags and a magnitude (1–5). Magnitude is computed from a weighted blend of source authority, asset market cap, social velocity, and on-chain anomaly score. The tags:
- Listing — a CEX/DEX adds or removes a market.
- Regulatory — enforcement, framework, ETF, jurisdictional change.
- Hack — exploit, exit scam, bridge incident, validator failure.
- Macro — Fed action, CPI, FX shock, geopolitical event.
- Partnership — protocol integration, distribution deal.
- Unlock — token unlock, vesting cliff, staking unlock window.
TRUE Highlights
Highlights are the user-facing form of catalyst events: short, illustrated cards with a CTA, multilingual, time-limited. They’re authored partly by the news-agent and curated by the editorial team. Each Highlight carries a tag taxonomy that lets the home feed mix breaking events with evergreen explainers.
Why this matters
A price chart without context is noise. Event Catalyst attaches a why to every meaningful move so users can decide whether the move reflects new information they care about or noise they should ignore. The signal-agent uses the same attribution to compute the cleanliness axis on signals — a signal generated into a fresh catalyst is automatically lower-cleanliness.
Highlights appear on the home feed, sorted by relevance to your watchlist and tags you follow. You can mute a tag entirely (no more macro Highlights, for instance) or pin one for the top of the feed. Time-limited Highlights expire automatically — you’ll never see a “what just happened” card three days after the fact.
Subscribe to catalyst feeds
curl -X POST https://app.truefinance.ai/api/v1/webhooks \
-H "Authorization: Bearer $TRUE_TOKEN" \
-H "Content-Type: application/json" \
-d '{
"url": "https://example.com/hooks/catalysts",
"events": ["catalyst.detected", "catalyst.updated"],
"filter": { "min_magnitude": 3, "types": ["regulatory", "hack", "macro"] }
}'Webhook payload
{
"id": "evt_01JX9...",
"type": "regulatory",
"magnitude": 4,
"sentiment": { "direction": "negative", "confidence": 0.81 },
"headline": "...",
"summary": "...",
"assets": ["BTC", "ETH"],
"sources": [
{ "url": "https://...", "publisher": "...", "fetched_at": "2026-04-29T12:34:56Z" }
],
"first_seen_at": "2026-04-29T12:34:50Z",
"last_updated_at": "2026-04-29T12:35:14Z",
"dedup_key": "reg-20260429-..."
}Replay endpoint
For catastrophic events (hacks, regulatory shocks), historical detail is available via the replay endpoint. Use it to backfill a missed window during downstream downtime.
curl -G "https://app.truefinance.ai/api/v1/catalysts" \
--data-urlencode "from=2026-04-29T00:00:00Z" \
--data-urlencode "to=2026-04-29T23:59:59Z" \
-H "Authorization: Bearer $TRUE_TOKEN" Sentiment is approximate, not predictive. Direction labels reflect a snapshot read of newsroom tone and on-chain reaction in the first minutes — they can flip as more information surfaces. Never trade on a single headline. The asset-graph attribution can over-attribute when projects share contributors or share a parent foundation; a Highlight tagged on Asset A may also move Asset B without TRUE having attributed it.
Safety, limits, failure modes
- Source latency. Some publishers update slowly; on-chain anomalies are typically faster than newsroom coverage.
- Cross-source dedup. Events from multiple sources collapse to a single Highlight with a list of source URLs. Dedup keys are stable for replay.
- Sentiment flip. Re-evaluation runs on a 5-minute cadence in the first hour, then 30-minute cadence. Webhooks emit
catalyst.updatedon every flip. - Attribution drift. The asset graph is curated; if you spot a wrong attribution, please report — the editorial loop is open.