Hi, The minimum data format would be (time, bid, bid size, ask, ask size, price, volume). The price and volume might be empty if no sales on that second. I searched the forum and came out couple of different services, such as http://www.elitetrader.com/vb/showthread.php?s=&threadid=248841. I don't need data for symbols for all market for every trading seconds. I only need 3+ years of 5 minutes of open (9:30-9:35AM) and 5 minutes of close (3:55-4:00PM) with about 50 symbols so far. 1. NASDAQ-On-Demand 2. Active Tick 3. IQFeed 4. IB. I have account with them but they has only tick data up to 1 year. My planning is to download the needed ticks to my local database and do the back tests. So I will not keep the subscription for very long period. I will use their services on demand. Any suggestions on which one I should go with? Thanks Daniel
FYI, you may not get the complete picture unless you get the order imbalance feeds directly from the exchanges (Nasdaq is one of the big ones for this, as is NYSE). Which is unfortunate in my view, as it adds to the heterogeneity of market data. Independent of that, this seems to be an extremely competitive space now (from the perspective of trading, that is) -- significantly more so than it was 3 years ago. So my take is that traders will have to wallow through a big mess of ever-more heterogeneous data for an ever shrinking profit margin, all while bearing the same level of risk as before.
QuantQuote might be the best bet for you, they are the only provider I know of that actually provides second resolution data. For other providers, you have to pay extra for tick and then make your own 1 second bars. See here: https://quantquote.com/products_second-data.php You can get it in CSV or their proprietary compression format. They don't do monthly accounts for download so you can't sign up for a month, download everything, then cancel. You'll have to buy it outright. Most of the other services you mentioned probably have safeguards in place to avoid you doing the same thing. Usually, they do things like limit the history that is available for subscribers.
It's a matter of perspective. I prefer my data top quality, free of errors, and easy to use. Bad data can suck away massive amounts of time trying to debug/fix/replace. If you buy bad data, you're out of luck and have to spend even more money buying good data later. Not to mention the huge amount of time wasted. Then there are the data bugs that you don't notice, but can be quietly screwing up your backtests which in turn screws up your trading. At the end of the day, spending more from the get go for good data more than pays for itself later on through time saved.
Do you have evidence that "trading is significantly more competitive than 3 years ago"? Or is it just another baseless uninformed opinion we see on here every day?
Thanks for your info. The cost is in scope if not the whole market data is needed. Data in seconds for 3 years (01/2010 - 04/2013) with 30 symbols costs ~ $300. I think I will go with QuantQuote for simplicity and convenience.
his comment wasn't baseless. opens/closes are definitely tighter. not being aware of that is pretty ironic given your second question (as snotty and rhetorical as it may be).
Apologies, reading back I see he was talking about Open/Close trading, thought he meant trading in General.
What do you mean 'tighter'? It means smaller bid/ask spread on opening and closing? I do see some strategies used to work not working any more.