Hey Tom, For DTN, tick - 8 days minute intervals - 120 days daily - several years For esignal, I think the tick goes back 10 days. Stephen
Stephen, >tick - 8 days excellent. BTW - what did you code your interface in - if you don't mind me asking. Thanks, Tom
If all you need is tick data, not streaming real-time data, check out CME. You can get the latest tick data there directly.
I used Python. There are libraries for everything you can think of. It is interpreted and people say it isn't as fast as C/Java, but simple optimizations make it ten times faster. for example, when receiving your packets, don't concatenate strings by going: data = data + packet Instead, store packet in array and then go: data = string.join(packet_list) . Since the string lib is in C, it goes full speed. I also use scipy / Numeric to for analyzing and storing data in memory. You can also use psycho, a JIT-compiler for python, amazingly simple to use and get almost as fast as C some code.
Thanks for posting Schan... I have been looking at DTN API for developers for awhile - i would like to do the following in real time do you think it is possible... For all sp500 stocks i would like to grab only the actual printed trades that have a tradeVolsize of 10,000 shares or greater in real time and drop them in an MSDE or AccessDB... do you think i could do this... i looked at their api docs and it says you can filter for trades only - but can i also filter or only grab trades by certain sizes... for this many stocks... 500?... do you think this is possible without choking my CPU - willing to go to dual opterons if needed... hope i was clear as to specs... cj
Hi cj, I don't think DTN has a block size filter. Even so, I think you can do it. My 50 watch symbols peak at 2000 updates/sec, 25 KB/sec bandwidth, processing 250 KB/sec text with unoptimized python. My PC is a 2.2 Gig Althon handles it ok with charts and IB java app running also. Your sp500 might be 10X this. If 10K blocks are 2% of all trades, 400 seperate inserts/sec might break the database unless you batch them up. I'm waiting to get a AMD dual-core Good luck, Stephen
I started a project just for futures tick collection. Take a look at http://wagerlabs.com/uptick. I'm doing it in Lisp, though. I will post performance figures once I have them.
There is no function to filter the quotes in the API, but you can definately do this in your program. Watching 500 symbols shouldn't be an issue (I have watched thousands of symbols on a P4 1.8ghz laptop). Just have your program look for trade volume >10,000 and write that to the database. The relative number of trades with volume > 10,000 is pretty low, so even with 500 symbols you should be able to write that to a database without any issue. Jay
Thanks Jay... helps me better understand my options ... cj... _________________ HAVE STOP - WILL TRADE If You Have The Vision We Have The Code