market data?

[EliotWoods]

Hi all, I'm pretty new at this (ie, equities) and got a pretty dumb question here (yes, it is dumb... just looking to get some form of confirmation that it is dumb), but here it goes anyway. I realize every firm is different, working with different budgets, different restrictions, different tools, ect.... However, with that said, how is historical market/reference data typically used? I imagine it is usually delivered as a product (ie, Reuters reference data) but is it ever collected/logged/accumulated (to a database) by the firm itself (with each full snapshot update) to get a track record going of a stock for which to base strategies on? Or is this approach far to expensive than reference data packages to ever be used (unless the firm is subscribed / watching the stock very closely anyway, pulling data for that particular stock all the time). Note that I realize in most cases strategies would also need to look much further into the past than an a home grown historical data set like this could allow for. Is it ever actually done this way? Thanks for any help!

 

[Andy Nguyen]

Data providers sell you different tiers of access. You can have access to their server, you can download real time data and store on your own server, you can download their data in xml format and store, etc.
In most case, if you can view the data, you can find a way to store it. And almost everyone stores historical data in one way or another.

 

[Tom Ablewhite]

Hi Eliot,

To your point, It depends on the data vendor.
Thomson Reuters provide real-time feeds that you could store locally.
Or there are reference data options to pull via API or FTP for intra & end of day data.
Or, for example DataScope Tick History http://thomsonreuters.com/products_...roducts/valuation_risk/datascope_tick_history provides millisecond-timestamped tick data going back over eleven years, covering 35 million OTC and exchange-traded instruments worldwide. Basically everything that has passed across their Integrated Data Network.

How is the data typically used?

Typically to refine your algorithms to perform in real-life scenarios, analyze and back-test easily, prove compliance and best execution or
process analytics and search for signals as events happen.

:T