Skip to content
All library documents

Choosing Historical Market Data Vendors for Large Multi-Asset Research Sets

Article Quant Q&A · Author: Forgottenscience

Summary

The document asks about obtaining long historical daily price series across thousands of instruments and several asset classes. The response says that paid vendors do offer historical datasets, but that assembling prices is only part of the task. Data quality and ongoing maintenance matter: corporate actions and other adjustments may be missing or inconsistent, leaving researchers to clean and validate the data before using it.

The answer frames vendor selection as dependent on the asset classes and the cleanliness of the resulting histories. It alludes to a vendor list and other source recommendations, but those details are absent from the supplied text. It provides no comparison of providers, coverage statistics, pricing, or validation methods, so it is useful mainly as a caution about data preparation and the limits of raw historical price collections.

Key ideas

  • Paid vendors may provide large historical market datasets across asset classes.
  • A price history may still require substantial cleaning before it is suitable for research.
  • Corporate actions and other adjustments are important sources of data-quality problems.
  • Vendor suitability depends on asset-class coverage and the maintenance of the series.

Tags

Full text
# Acquiring large sets of price series


# Acquiring large sets of price series












Selling and delivering real-time data seems to be the focus of practically all large data vendors, but I am more interested in acquiring large sets of historical daily data covering, say, 5.000 instruments (equity, rates, fx, commodities, etc) about 15-25 years back. It does not seem to be a priority for any vendor to deliver such a service - is this true? It would seem like an interesting set to mine.

Please note that I am not asking for a free service - I am aware of the typical free sources, but calling them repeatedly seems like a hassle and besides, they often have greatly varying time series lengths depending on the instrument.

## Answer by Eli (score 1, accepted)

https://quant.stackexchange.com/a/18497

It's not entirely true. The challenge is to obtain a clean and well maintained historical set, not just a prices set.

Typically, there are many vendors that provide historical data sets, but fail to accommodate corporate actions or other adjustments, so in order to use the set in a research, you'd still need to clean the data thoroughly.

Depending on what asset classes you are looking for, here is a short list of vendors with relatively clean historical data:







There are more sources indicated here.

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.