Historical futures data for many contracts & exchanges?

Discussion in 'Data Sets and Feeds' started by prophet, Aug 11, 2013.

  1. prophet

    prophet

    What vendors sell historical futures data covering a large variety of worldwide contracts and exchanges?

    Ideal frequency is 5 or 15 minute price bars with volume. 30 or 60 minute bars might be OK. T&S or 1 minute bars are not necessary.

    10+ years historical.

    My primary interest is energy, agricultural, metals, minerals and manufactured products.

    Secondary interest is forex, interest rate and everything else.

    I want contracts of any liquidity. The more contracts the better.

    I need most or all delivery months, not continuous contract.

    It would be nice if time zone data is done right.
     
  2. Nanex has that. Expect to PAY though. Around 250 per month. And then spend weeks to write the extraction - they come with a C/C++ API. OTOH you get all exchanges on tick level - bid, ask - and can do with the data what you want. That is all instruments. TONS of information.
     
  3. prophet

    prophet

    Thanks! Just contacted Nanex.

    Weeks to write extraction code? That seems odd.

    Have you used Nanex data? How does it compare to other vendors?

    $250 per month is reasonable if Nanex logged the data from multiple exchanges consistently and with quality timestamps. It still might be overkill for my strategies since I only need 5 or 15 minute bars.

    Any other vendors?
     
  4. Weeks until you ahve it all right,. There are issues in bid/ask -nanex reports what the exchange sends and man, some of the data you sometiems get is crap.

    YOu should haev a first version out within 2-3 days though. But this is a LOT of data - nanex does not filter by symbol, so the CME tapes alone are about 1.8gb per market day average. 350.000 symbols (all options are symbols). Filtering that fast requires some smart programming.
     
  5. prophet

    prophet

    Thanks. Looks like nanex does a good job logging and extracting their data. It may work well for me.