Can anyone recommend a source of historical data WITHOUT survivor bias?

Discussion in 'Data Sets and Feeds' started by cashonly, Dec 29, 2010.

  1. cashonly

    cashonly Bright Trading, LLC

    I'm looking for a source of historical data WITHOUT survivor bias.

    You know, when a company is bought, or goes into bankruptcy, or is delisted, or some other such thing, their symbol basically goes away from all historical data sources that I can find. So if you're doing backtesting it's impossible to have the results run on these stocks and you have "survivor bias"

    Looking for a data source that has the data for these symbols as well.
     
  2. rwk

    rwk

    The CRSP point-in-time (PIT) database is the gold standard for backtesting, but it's very expensive and not generally available.

    While their database not specifically design to exclude survivorship bias, csidata.com told me they never delete stocks. I'm not sure how they handle symbol changes (e.g. "C" was Chrysler, now Citigroup).
     
  3. flip

    flip

    Do you need daily or intraday data?

    For daily data I can recommend CSI (csidata.com, as mentioned by rwk). Symbol changes are handled by having an internal symbol number. This way different stocks can have the same ticker symbol.

    The delisted stocks package is not part of the usual subscription, so if you need any more infos or samples of a specific stock just let me know.
     
  4. http://www.compustat.com/ is a popular vendor of research-quality data, and some find it convenient to access it via http://www.qaisoftware.com/.
     
  5. Does Bright have a Bloomberg? ServerAPI or B-Pipe?

    EDIT: what data do you need? Close? Open? Tick? etc...

    EDIT #2: How far back?
     
  6. rwk

    rwk

    Correction: It's Compustat that has the PIT database. I have not used it because it's so expensive.

    You might also look at portfolio123.com. They say their data does not have survivorship bias, though it looks like you can only access it via the Internet.
     
  7. I had your back -- I'd already sent the correct link.
     
  8. rwk

    rwk

    The OP didn't say what he wants the data for, but I would think a coach/trainer for Bright would be working mostly with high momentum stock strategies. I don't think survivorship or lookahead biases are all that important for high momentum, but they're absolutely critical for anything like value investing.
     
  9. crsp is also survivor-bias free. crsp has prices, returns, volumes for all NYSE, AMEX and NASDAQ stocks, monthly back to 1926 and daily pretty far back. ASFASIK, compustat PIT is only financial statement data.
     
  10. Compustat supplies prices, returns, volumes. What leads you to believe otherwise?
     
    #10     Dec 30, 2010