Alternative Data for Investing: Satellite, Web Scraping & Unconventional Sources

What Is Alternative Data?

Alternative data = Non-traditional data used for investment insights.

Traditional data: Financial statements, economic releases, prices

Alternative data: Everything else that signals economic activity

Market size: ~$7B and growing rapidly.

Categories of Alternative Data

Satellite & Geolocation

Satellite Imagery:

• Parking lot car counts (retail traffic)
• Oil storage tank shadows (inventory)
• Crop health (agriculture)
• Construction activity
• Shipping/port activity

Providers: Orbital Insight, Planet, Descartes Labs, SpaceKnow

Geolocation Data:

• Foot traffic (SafeGraph, Placer.ai)
• Store visits
• Trade area analysis
• Competitor monitoring

Web & App Data

Web Traffic:

• Site visits (SimilarWeb)
• Search trends (Google Trends)
• App downloads/usage (Sensor Tower, App Annie)
• E-commerce activity

Web Scraping:

• Pricing data
• Job postings
• Product availability
• Review sentiment

Transaction Data

Credit/Debit Card Data:

• Consumer spending (anonymized, aggregated)
• Sector trends
• Geographic patterns

Providers: Bloomberg Second Measure, Earnest Research, Affinity Solutions

Point-of-Sale Data:

• Retailer sales
• Item-level data
• Real-time indicators

Social & Sentiment

Social Media:

• Twitter/X sentiment
• Reddit discussions
• StockTwits
• Influencer tracking

News Sentiment:

• News volume
• Tone analysis
• Topic extraction

Providers: Dataminr, Sprinklr, RavenPack

Expert & Survey

Expert Networks:

• GLG, AlphaSights, Third Bridge
• Industry expert calls
• Primary research

Proprietary Surveys:

• Consumer surveys
• Business surveys
• Custom panels

Other Alternative Data

Patent Filings: Innovation tracking

Regulatory Filings: SEC, FDA, EPA

Government Data: Permits, licenses

Weather: Agricultural, energy impact

Shipping/Logistics: Container data, AIS

How Alternative Data Is Used

Company-Level Signals

Revenue Nowcasting:

• Credit card data → quarterly sales estimate
• Web traffic → customer trends
• App usage → engagement metrics

Example: Track Target foot traffic to estimate same-store sales.

Sector/Industry Signals

Consumer Health:

• Aggregate spending patterns
• Restaurant visits
• Travel activity

Example: Airline bookings data for travel sector.

Macro Signals

Real-Time Economic Activity:

• Aggregate credit card spend → consumption
• Job postings → labor demand
• Shipping data → trade activity

Example: Satellite parking lot counts for retail sales estimate.

Evaluating Alternative Data

Key Questions

• Coverage: Does it represent the target adequately?

• History: Enough backtest data?

• Frequency: How timely?

• Accuracy: Validated against ground truth?

• Alpha decay: How long before signal commoditized?

• Compliance: Legal and ethical sourcing?

Common Pitfalls

Survivorship bias: Only current companies in sample

Backtesting issues: Data not available in real-time

Overfitting: Too many variables, too little history

Correlation ≠ Causation: Spurious relationships

Regime changes: Pandemic disrupted everything

Data Quality Considerations

Data Sourcing Issues

• Panel representativeness
• Geographic coverage
• Demographic coverage
• Opt-in bias

Processing Challenges

• Noise vs signal
• Seasonality adjustment
• Normalization
• Missing data

Compliance & Ethics

• Privacy regulations (GDPR, CCPA)
• Consent and disclosure
• Material non-public information (MNPI)

Alternative Data Vendors

Data Aggregators

• Quandl/Nasdaq Data Link: Multiple alt data sources
• Eagle Alpha: Alt data marketplace
• Neudata: Alt data scout and reviews

Sector-Specific

Consumer:

• Earnest Research (credit cards)
• Placer.ai (foot traffic)
• SimilarWeb (web traffic)

Energy:

• Orbital Insight (tank levels)
• Kpler (shipping)
• Kayrros (satellite)

Agriculture:

• Gro Intelligence
• aWhere
• Descartes Labs

DIY Alternative Data

Free sources:

• Google Trends (search interest)
• Reddit API (sentiment)
• Government data (unconventional uses)
• OpenStreetMap (location data)

Web scraping (carefully):

• Job postings
• Pricing data
• Product availability
• Review aggregation

Integration with Traditional Data

Blending Approaches

• Alternative confirms traditional: Higher conviction

• Alternative contradicts: Early warning?

• Alternative leads: Nowcasting advantage

Weighting Considerations

• Alternative data often higher frequency
• Traditional data more reliable
• Optimal blend depends on use case

Building an Alt Data Capability

Starting Points

• Google Trends: Free, easy to start

• Indeed/LinkedIn: Job posting analysis

• Web traffic tools: Free tiers available

• Satellite (free): Sentinel data available

Scaling Up

• Identify specific use case

• Evaluate vendors (trials usually available)

• Build data pipeline

• Backtest rigorously

• Monitor live performance

Team Capabilities

• Data engineering
• Statistical skills
• Domain expertise
• Compliance awareness

Pro Tips

• Start with thesis: What are you trying to predict?

• Beware overfitting: Alternative data = lots of variables

• Real-time challenge: Ensure data available when claimed

• Decay is real: Good signals get arbitraged

• Cost-benefit: Some data very expensive

• Combine sources: Multiple alt data > single source

Explore unconventional data →