4

I am searching for finance datasets that has the following format

year,quarter,month,Company,S&P index,P/E ration,feature x,feature y, feature z....to test a machine learning prediction algorithm. Does anyone know where to find a good dataset online?

Actually I want year,quarter,month,Company,S&P index,P/E ration,feature x,feature y, feature z for a historical time period (i.e 2005-2009).

Apparently yahoo finance, Quandl seems to have these but only for the current year. I have been searching high and wide but didn't find a API, dataset that has these features for a certain historical time period. Like for example

For Google and Microsoft, for years 2005-2009, I want (Sector Price Dividend Yield Price/Earnings Earnings/Share Book Value 52 week low 52 week high Market Cap EBITDA Price/Sales Price/Book).

Nicolas Raoul
  • 8,426
  • 5
  • 28
  • 61
mathopt
  • 141
  • 1

3 Answers3

1

Here's a link to datasets (Excel spreadsheets) from a NYU professor whom has been keeping corporate finance data on major corporations in US, Canada, UK and Australia for 20 years.

http://pages.stern.nyu.edu/~adamodar/New_Home_Page/data.html

Andrew - OpenGeoCode
  • 8,657
  • 17
  • 28
  • That's a lot of data, but from a quick review, none of it is at the per-company level. If anyone can point directly to that, it would improve this answer. – Joe Germuska Mar 10 '15 at 14:17
1

The Markit On Demand API has an endpoint called Interactive Display which includes historical data as a JSON response.

Example URL:

http://dev.markitondemand.com/Api/v2/InteractiveChart/json?parameters={%22Normalized%22%3Afalse%2C%22NumberOfDays%22%3A365%2C%22DataPeriod%22%3A%22Day%22%2C%22Elements%22%3A[{%22Symbol%22%3A%22AAPL%22%2C%22Type%22%3A%22price%22%2C%22Params%22%3A[%22c%22]}]}

The URL-decoded JSON packet looks like this:

{
   "Normalized":false,
   "NumberOfDays":365,
   "DataPeriod":"Day",
   "Elements":[
      {
         "Symbol":"AAPL",
         "Type":"price",
         "Params":[
            "c"
         ]
      }
   ]
}

(hint - replace Number of Days from 365 to 3650+).

You can expand the parameters to include price, volume, and simple moving average. I don't think it includes derived values like P/E.

enter image description here

philshem
  • 17,647
  • 7
  • 68
  • 170
1

http://www.quandl.com best public listed firms, API works with http://www.thinknum.com & http://www.spearian.com.

For paid vendors like; http://www.privco.com or http://www.bvresources.com or http://www.cbinsights.com or http://www.duedil.com are 90% Western info.

Open & non-western repositories you instead have to go to compliance sites like; http://cdm.unfccc.int/Projects/projsearch.html

Want more? Think "activist" vs "compliance" and your success rate goes higher, ie OpenCorporates = entity ONLY vs financials, but that = "seed list" to append financial KPI elsewhere, ie "units of production" from UNFCCC environment modeling Vs compliance actuals.

Andrew
  • 73
  • 5