Predite integrates Finnhub (financial news + economic calendar) and a custom news pipeline to surface events that move prediction markets. This doc explains the two feeds and how to use them in your trading.
## The News Feed
Source: Finnhub provides real-time finance news from major outlets (Reuters, Bloomberg, WSJ, FT, CNBC, etc.). We supplement with crypto-specific sources for crypto-related markets.
Frequency: Real-time. New headlines appear in the feed within 10-30 seconds of publication.
Filtering: The feed is filtered to news relevant to active prediction markets. A headline about a celebrity divorce won't appear unless there's an active market about it.
Sentiment scoring: Every headline is scored by our sentiment model on a -100 to +100 scale:
- +75 to +100: Strongly positive (large rate cut, surprise upside earnings beat)
- +25 to +74: Moderately positive
- -24 to +24: Neutral
- -25 to -74: Moderately negative
- -75 to -100: Strongly negative (war, major sanctions, disaster)
Direction inference: For each headline, the model also predicts which markets are likely to move and in what direction. Example: "Fed signals June cut more likely" β bullish for "Fed cuts June 2026" YES.
## Using the News Feed for Trading
The feed appears in two places:
Dashboard sidebar widget. Shows the last 10 headlines, color-coded by sentiment. Click any headline to read the full article and see the affected markets.
Dashboard β News page. Full feed with filters by category (politics, economics, crypto, etc.), source, sentiment threshold, and time range. Export as CSV.
Practical workflow:
1. Monitor headlines for events you have a directional view on
2. When a headline aligns with your view, check the affected markets
3. If the market has not yet moved to reflect the news (within 5-30 seconds typically), execute
4. If the market has already moved, evaluate whether the move was over- or under-reaction
The News Reactor bot (Bot Strategies doc) automates this workflow.
## Economic Calendar
Source: Finnhub economic calendar API.
Coverage:
- US: Fed meetings, FOMC minutes, CPI, PCE, employment, GDP, ISM, retail sales, housing starts, jobless claims
- Eurozone: ECB meetings, CPI, GDP
- UK: BoE meetings, inflation
- Japan: BoJ, CPI
- Major emerging markets: India, China, Brazil
Updates: Calendar is refreshed daily. Major event timings (Fed meetings) are usually known 6+ months in advance.
Each event shows:
- Date and time (your local timezone)
- Country / central bank
- Indicator name
- Previous reading (e.g., last CPI was 3.2%)
- Forecast (analyst consensus expectation)
- Actual (filled in when the event publishes)
- Surprise indicator (actual vs forecast)
## Using the Calendar for Trading
Pre-event positioning. Identify markets that depend on the upcoming event. Example: "Will the Fed cut rates in June" depends on Fed meeting outcome. If consensus is "no cut" but you have a view that cuts are likely, position before the meeting.
Post-event reaction. Many markets continue to move for hours after the release as participants digest the news. Watch for over-reactions to fade or under-reactions to ride.
Calendar Bot. The Calendar Bot strategy automates pre-event positioning using historical patterns. Backtest before deploying.
Earnings season: During earnings weeks (Apr, Jul, Oct, Jan), prediction markets exist for many large-cap companies' beat/miss outcomes. The Calendar covers earnings dates for S&P 500 + Nasdaq 100 companies.
## IPO Calendar
A separate tab covers upcoming initial public offerings. For each IPO:
- β’Company name and ticker
- β’Expected listing date (typically firm 1-2 weeks before)
- β’Expected price range
- β’Underwriters
Markets tend to exist for "Will [company] IPO before [date]" β useful for traders following the IPO calendar.
## Setting Up News Alerts
Go to **Dashboard β Alerts β New Alert** and create a news-based rule:
- **Trigger:** "News headline matches keyword [X]" or "Sentiment score above [Y]"
- β’**Scope:** All markets, or specific markets/categories
- β’**Action:** Email, Telegram, WhatsApp, or push notification
- β’**Cooldown:** Prevent the same alert from firing twice within X minutes
Example alert: "Notify me on Telegram whenever a headline contains 'Fed' AND sentiment magnitude > 60". Triggers when Fed-related news drops with strong directional implication.
## Common Mistakes
- **Reacting too late.** By the time you read a headline and execute, the move is usually 80%+ complete. Bots beat humans here.
- β’**Trusting sentiment without context.** A "+85 positive" headline on a market you don't understand can still mislead you. Always read the article.
- β’**Trading every event.** Most economic data prints are already priced in. Focus on truly surprising results (large gap from forecast).
- β’**Ignoring revisions.** A surprise "good" jobs number gets revised down a month later. Markets sometimes overreact then revert.
## Related Docs
- [Reading Signals](/docs/reading-signals)
- β’[Bot Strategy Templates](/docs/bot-strategies)
- β’[Notifications & Alerts](/docs/notifications)