Skip to content
All library documents

Choosing Global Equity Data for Corporate Actions and Backtests

Article Quant Q&A · Author: user6564
Read the original

Summary

The document discusses how to source end-of-day global equity prices for historical research and ongoing use. Its central data problem is maintaining consistent security identities through ticker changes and reuse, while correctly accounting for splits, dividends, delistings, and other corporate actions. These details matter when constructing long histories and avoiding errors in strategy results.

The replies suggest two approaches: use a specialist vendor that supplies adjusted data and includes delisted securities, or build a security master and retrieve adjusted or unadjusted histories through Bloomberg. For index-based research, historical constituents can help reduce survivorship bias; maintaining a live database also requires handling identifier changes. The discussion offers practitioner recommendations rather than a systematic vendor comparison. Coverage, cost, licensing, adjustment conventions, and data quality should be checked against a researcher’s market universe and intended use; the replies do not establish that any source is complete or error-free.

Key ideas

  • Corporate actions and reused tickers can make long equity histories difficult to normalize.
  • Delisted securities should be included to reduce survivorship bias in historical research.
  • Historical index constituents can help define a point-in-time equity universe.
  • A production data process needs to maintain security identities as identifiers change.

Tags

EquitiesBacktesting
Full text
# Clean EOD global Equities data provider for backtesting investment strategies


# Clean EOD global Equities data provider for backtesting investment strategies












I'm trying to find a good source for global equities for EOD data (historical and forward basis), currently using Bloomberg's back office data, but it is very hard to normalize it for corporate actions (ticker changes are painful, specially when same tickers are re-used by different companies) - I'm applying all the corporate actions but still it is not the best dataset. There has to be a better source and better way to do it. I know there are lot of sources, but what I'm really looking for is a robust solution that has a very good way to apply corp actions etc.

## Answer by hvedrung (score 1)

https://quant.stackexchange.com/a/18974

You can try premiumdata.net. Among several sources I used this provides most clean data. They account for splits and dividends. Delisted securities are present in the data. This is not free but the price is modest for this quality for my point of view. They provide EOD data for US, Australian and Singapore stock exchanges.

## Answer by ChrisD (score 0)

https://quant.stackexchange.com/a/18470

If you are using Bloomberg then you can pull prices adjusted for corporate actions such as splits, dividends, and other capital adjustments, assuming this meets your needs (i.e. momentum based quant strategies). If you know which global equity indices to track then you can pull the historical constituents to minimise survivorship bias. In Bloomberg, MSCI index constituent lists requires a subscription whereas S&P index members are readily available.

From there you can create a security master list of all the unique tickers in your global universe and pull their adjusted/unadjusted historical data in a specified currency.

Admittedly the tickers can change after your download which requires management if you intend to run a real-time database for production purposes via this process, but it is possible to manage that process.

Otherwise you could look at e.g. Thompson-Reuters, CapitalIQ, Factset, etc. but these can be very expensive. I find Bloomberg data via the terminal to be quite a cost effective solution for global equities, albeit far from perfect. Ideally one would have a loader that auto populates your DB tables each day and manages ticker/ security ID changes.

Hope this is helpful.

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.