Alternative Data Explained

Alternative data is information about companies, industries or economies that comes from outside traditional sources such as financial statements, market prices and official statistics. Examples include aggregated credit card transactions, website traffic, app downloads, satellite images of parking lots, shipping records and job postings. Hedge funds and quantitative firms use alternative data to estimate results before companies report them or to track economic trends in real time. The industry has grown rapidly, but so have the costs, legal questions and competition.

Main categories#

Why investors use it#

• Earlier information: estimate a retailer's quarterly sales weeks before it reports.
• Higher frequency: daily or weekly data instead of quarterly reports.
• Independent verification: check management claims against outside evidence.
• Macro nowcasting: track the economy faster than official statistics. See GDP.

The alternative data workflow#

1. Source and evaluate datasets: coverage, history, frequency, cost.
2. Legal and compliance review: privacy, licensing and material non public information.
3. Clean and map: link data to companies and tickers, handle gaps and bias. See Cleaning Market Data.
4. Test predictive power: does the data predict reported results or returns, out of sample? See Signal Discovery and In-Sample vs Out-of-Sample Testing.
5. Integrate into models or research processes.
6. Monitor decay as more firms buy the same data. See Signal and Alpha Decay.

Costs#

Datasets can cost from a few thousand to over a million dollars a year. Add the costs of data engineers, storage and analysts. Large funds can afford many datasets; smaller traders typically use cheaper or free sources such as search trends, public filings and app rankings.

Legal and ethical issues#

Regulators, including the SEC, have scrutinised how funds obtain and use alternative data.

Limits and risks#

• Bias: panels may not represent the whole population.
• Short histories: many datasets cover only a few years, limiting testing.
• Overfitting: testing many datasets increases false discoveries. See P-Hacking and Multiple Testing.
• Crowding: once many funds use a dataset, its edge shrinks.
• Changes in the data source: a vendor's methodology or panel can shift suddenly.

Frequently asked questions#

What is alternative data?#

Information from non traditional sources, such as card transactions, web traffic, satellite images or job postings, used to gain insight into companies and economies.

How do hedge funds use alternative data?#

To estimate company results before they are reported, track trends in real time and verify management claims.

Is alternative data legal?#

Generally yes, if it is properly sourced, anonymised and does not involve material non public information obtained in breach of a duty.

Next, look at spending data in Credit Card and Consumer Spending Data.