collecting tick data where and at what price?

Discussion in 'Data Sets and Feeds' started by drBearMrBull, Jul 11, 2010.

  1. So the hot topic is to build your own database of tick or bars or trades to optimize your automated strategies. Many places I know of buy tick data from vendors. Problem with this is that the data you buy from exchange re-seller will give you back testing results that are too optimistic. The time at which you can trade is not the stamped time of the exchange but the time at which you at home or at work would have known of that piece of info a time that the exchange doesnt sell and it depends on your infra/network/where you are etc.

    Given the above I am strongly opinioned that the proper way is to collect the data on your own. Here comes my question, I'd like to hear from you with regard to your experience with data feed, which ones you use, how expensive they are, what solution you suggest for collection. Obviously you can tell me to get direct exchange feeds with your own dedicated lines but I am more interested in coming up with a decent enough solution a bit DIY way that is less expensive.

    EU equities quotes and trades.
    US equities quotes and trades.
    US equity options
    Asian equities

    My 2 cents experience is with Activ for US equity data and Opra data. Not bad but highly overpriced and you can be certain that it breaks at peak times; I would never buy it for my own small operation. On EU data Activ is an overpriced lost cause.

    many thanks
  2. rosy2


    i know for CME futures you can buy the actual FIX messages from the exchange with the relevant timestamps.
  3. Are you saying you are trying to record data with a timestamp as it arrives to you? Essentially recording on a delay and then i would venture a guess that you are going to try to test for an edge based on delayed data?
  4. yes, you are only able to act on what you know when you know it.
    Even for those who are colocated they can only hope to be able to act on the exact times that the exchange publishes in its fix logs. If your data center is miles away your network latency is stochastic and you can trust that it will be at its worst when you most need it (when markets go nuts)

    There is no point at having backtested a Sharpe 10 strategy that relies on exchange timestamps as there is poor evidence that strategies backtested on close prices will work. It aint easy to trade at the official close or at the time the "official" fix logs tell you to...

    any takers on price feed and costs for the asset above?
  5. why not just buy data (like from esignal) and record forward as you get it?