Why don't you try it and report? I looked at J but didn't find it as practical as OCaml. Jane St Capital (market maker in NY) uses OCaml exclusively to build their infrastracture, with excellent results. I'm starting with a FIX interface to TT and tick storage will logically follow. I'll be reporting my progress at http://groups.google.com/group/algokit.
There is no question that J is "impractical" from the point of view that if you don't come from an APL background, the language syntaxlooks unintelligable. But boy, the prospects of having an entire application fit in a L2 cache is extremely desirable. J programs are easily 10x smaller than comparable programs in other languages, and it is comparable to speed of C. I don't even have the time for the twenty projects I am working on now, ontop of my full time job. nitro
Just to return the discussion back to the topic, I believe KDB appends tick data to flat binary files throughout and incorporates them ... into the database or something (this part I don't remember).
KDB+ appears to have a 32 bit version that is now available for personal use, with some restrictions. http://kx.com/developers/software.php I am evaluating this product to use for a lot of data crunching & stats on commodity spreads. At first glance, this appears to be a good product for this type of work. Has anyone found anything new or interesting in the tick database field? Regards, Eric
It's called kdb personal for a reason. It will quit after an hour, has restrictions on memory used, etc. A full version of kdb can easily set you back 250K or more. Personally, I'd rather put that money into the trading account. Unless of course I had enough of it that I could justify spending 250K on development tools. That said, kdb is da bomb if you can afford it. There's a reason all of Wall St. uses it.
but are they making money directional trading? that's what I thought no fancy stuff here needed, good ol' TradeStation 2000i does the trick Maria
Thanks a lot all for this very interesting discussion! Does anybody have an experience with XDBs and Xenomorph Timescape? How would you evaluate this product in comparison to KX's, for example? Best regards, Fayssal
I thought it might be worthwhile to provide a link to a comparison of current NoSQL databases: http://kkovacs.eu/cassandra-vs-mongodb-vs-couchdb-vs-redis
Interesting link, thanks for that. What are you running on? I am currently looking for a tick database versatile enough to feed Excel, C#.Net 4.0 applications, WPF charts, and the like. Any recommendations? For my backtesting architecture I coded up my own binary data store through which I manage to iterate over ticks at a rate of 6.5 million ticks per second. But I look for something that may be able to feed different applications through APIs, ODBC, and/or other interfaces in order to code up a scanning app in .Net. It may be in combination with a CEP engine because I look to calculate customized indexes and indicators on historical data and store alongside the tick data and also calculate such data in real-time on part historical data, part real-time. Any suggestions regarding suitable databases? Please exclude HDF5 because I target .Net and this database is the worst when it comes to supporting .Net. Someone wrote up a .Net compliant Api but it severely lacks in functionality. Thanks.
Are you guys looking to run a condition through backtesting real-time and then trade instantly or just looking to store a historical tick database for backtest use now & then by a dedicated machine?