Free lesson · Alternative data
A data investment includes coverage, access and ongoing costs
Open interactive lessonPractice calculationsExplore labs
Start with the idea
Paying for a dataset is an investment in a research and operating process, not merely buying a file.
Symbols, units & horizon
- V_incremental: estimated annual strategy value above a matched incumbent in currency, after trading costs
- C_license,C_engineering,C_operations: annual or consistently annualized data-process costs in same currency
- V_net: annual net expected contribution before additional risk-capital charges
When and why to use this
Make a reviewable go/no-go decision on alternative-data procurement and productionization.
Paying for a dataset is an investment in a research and operating process, not merely buying a file.
Compare incremental strategy value with license, engineering, storage, labeling, inference and monitoring costs on the same calendar horizon. Count the cost of maintaining historical snapshots and investigating changed definitions. Rights to use data for research do not necessarily include redistribution or production use.
Evaluate a narrow pilot with a frozen benchmark and a stopping criterion. The provider’s historical demo may use cleaner labels, revised records or a universe you cannot reconstruct. A failed pilot is informative if its discrepancy and cost ledger is preserved.
A data investment includes coverage, access and ongoing costs
- Estimate incremental value against the fixed baseline over the declared annual horizon.
- Put recurring and amortized setup costs on that same horizon without double counting trading costs already included.
- Subtract all data-process costs and stress both benefit and cost assumptions; document access and redistribution rights separately.
Incremental value $60,000, license $20,000, annualized engineering $15,000 and operations $10,000 give net expected contribution $15,000; halving benefit makes it −$15,000.
Apply it in a strategy
- Freeze inputs at the stated decision time and record their units.
- Make a reviewable go/no-go decision on alternative-data procurement and productionization.
- Recompute the example, then change the material assumption and explain the difference.
Research deliverable
A data investment includes coverage, access and ongoing costs: produce the worked calculation, a timestamped input record and a written decision addressing this limitation: Backtest selection, unstable access rights, vendor revisions or underestimated upkeep can invalidate the projected benefit.
These are synthetic mechanics examples, not historical performance or paper replications. Module evidence and research boundaries record the 12 September 2026 review.
Evidence and boundaries · reviewed 12 September 2026
The records below distinguish research status, access depth and data dates. Abstract-only review identifies research questions; it does not establish a replicated empirical claim.
Further reading: Banerjee, Cordova, De Pooter & Grishchenko: Gauging the Sentiment of FOMC Communications through the Eyes of the Financial Press ↗
Federal Reserve Finance and Economics Discussion Series working paper 2025-048. Version: July 2025. Review: 2026-09-12; Abstract on Federal Reserve research index only. Markets: FOMC-related financial press and asset prices. Data dates: Abstract gives May 1999–November 2022. Limitation: The study motivates separating text measurement from event timing. Association with prices does not establish a tradable signal; article availability, revisions and dictionary construction were not inspected.
Further reading: Crane, Karra & Soto: Total Recall? Evaluating the Macroeconomic Knowledge of Large Language Models ↗
Federal Reserve Finance and Economics Discussion Series working paper 2025-044. Version: June 2025; later versions not established. Review: 2026-09-12; Abstract on Federal Reserve 2025 research index only. Markets: Macroeconomic data and release dates. Data dates: Exact series, date ranges and model snapshots not inspected. Limitation: Motivates explicit vintage and availability checks. The lesson does not quantify model error rates or reproduce the study.
Further reading: SEC: EDGAR Application Programming Interfaces ↗
Living primary provider documentation. Version: Reviewed 12 September 2026. Review: 2026-09-12; Documentation page. Markets: US public-company submissions and XBRL facts. Data dates: Filing-specific; no archive downloaded. Limitation: Facts can have amendments, duplicate contexts and differing units. An annual fiscal period is not an information-release time.
Research sources, review dates and limitations
Python implementation
Self-contained teaching example. Python 3.10+; dependencies and input conventions are shown in the code and notation. Run in your own Python environment.
# Python 3.10+; standard library and NumPy only.
# Synthetic teaching inputs; conventions and units are defined in the notation above.
def dataset_value(incremental,license_cost,engineering_cost,operations_cost):
if min(license_cost,engineering_cost,operations_cost)<0: raise ValueError('Nonnegative costs required')
return incremental-license_cost-engineering_cost-operations_cost
assert dataset_value(60000,20000,15000,10000)==15000
assert dataset_value(30000,20000,15000,10000)==-15000
print(dataset_value(60000,20000,15000,10000))Continue learning
Alternative Data: Measurement, Text & Incremental Value — all lessons- From a sampled panel to a population estimate
- A reproducible dictionary score for text
- An event return needs a predeclared benchmark
- Availability delays and signal decay
- Noisy proxies and attenuation
- Measure improvement against a frozen baseline
- From forecast accuracy to a costed decision
- A data investment includes coverage, access and ongoing costs
Quantitative finance and development glossary · Python resources and libraries · Research sources and limitations