[R-SIG-Finance] [R-sig-finance] yahoo historical data quality

[gug]

As an additional comment: you will see data errors even with paid data
services.  I have seen errors with historic prices from Reuters and also
with Thomson (prior to them being the same company).  With both of these
data providers, the data was from paid (professional-level) services.

Incidentally the only service that I have seen that has historical market
cap data (i.e. able to give a time series of historic # shares as a separate
data item) is Datastream, also now a Thomson product.


Andy Zhu wrote:
> 
> James:
> 
> MarketCap is spot data from public domain; no way to have historicals
> compared to the price series. However, comparison between google and yahoo
> for ATML is the same result for market cap. Actually it is hard to tell
> the quality regarding market cap, even some commercial data sources
> oftentime don't agree with each other, and sometimes the difference is non
> trivial. So I just live with it. A bit improvement could be that if you
> have access to different sources regarding MarketCap, take average or
> median on market cap (based on either outstanding or floating, but don't
> mix).
> 
> Another look on the historical data quality is from trading volumn. One
> can see the differences between yahoo and google. 
> 
> The further extension on MarketCap is the spinoff. We have no way directly
> from the available public domains to capture spinoff (opposed to
> dividend/split). although spinoff is a rare corp event, yet it has far
> major impact on stock metrics including MarketCap.
> 
> Erin:
> 
> Thanks for answering my email. The problem I identified in my post is
> independent of which R package one uses. If underlying data source is from
> Yahoo/Finance, one will inherit the quality problem because its root lies
> in yahoo/Finance, not in R packages/functions. The missing data example I
> gave below is obvious no matter which package you run, unless Yahoo fixes
> it or one goes to a quality (paid) data service. (By the way, I do use
> get.hist.quote).
> 
> Thanks
> 
> 

-- 
View this message in context: http://www.nabble.com/yahoo-historical-data-quality-tp22990415p22997700.html
Sent from the Rmetrics mailing list archive at Nabble.com.