I've seen and read about both but am interested in hearing about personal experience with either of these or any other algo applications that offers similar capabilities. Thanks.
I would like to talk with you about your interest in customer usage of StreamBase for Algo applications. Ray
StreamBase as in the streams processing DB? Stonebraker is a great mind in the DB community and I know of at least one bulge bracket firm quite happy with it for things as simple as VWAP to much more complex stuff. However, I would think that unless you need [extended] SQL interoperability between StreamBase and a historical data DB, it's probably better to go with a simpler solution. That's the true appeal... test your algo on historical data and then throw it at the real time stream, w/o risk of introducing any errors as you're running the same SQL. Curious as to know your use case for StreamBase.
Im not a tech guy which is why I'm here for advice. I need to run about 120 algos on a simulator simultaneously and 5 or so live, alot of which are cross-market strategies. Obviously low-latency is paramount along with the ability to create (or recreate in my case) new algos with ease in a time-efficient manner. If you know of any other options I sure would like to hear about them rather than parting with the $100,000+ that these platforms cost.
I'll defer to others on this one. I've seen StreamBase used at a bulge bracket bank and the sheer amount of data they run it on... well, that kinda volume doesn't come cheap, much more than a 6 digit # if you include the hardware involved. I am less familiar in this area with small solutions.
I think that Streambase offers a free developer edition... where you might be able to test your ideas very inexpensively either thru your own coding and setup or by outsourcing to a vetted professional developer... rather than having to buy it first...anyway here is a link to their dev page on their site... hope it helps some... http://www.streambase.com/developers/ cj... HAVE STOP <img src="http://www.enflow.com/p.gif"> WILL TRADE
Complex event replay ("backtesting") isn't simply querying a SQL store and iterating through rows of input data. It's recreating temporal relations between captured external events AND query chains exactly as they occured in real-time. The problem with SQL is that you end up with a backlog of events, and everything is thrown out of sync. Based on my experience with Esper and Coral8, any object database (events as tuples) will maintain constant time.
How competent is your tech. guy then? Who will do the "customization" for you? Regardless of which type of CEP/ESP package you use, you'll end up needing to hard code your algorithm in the core software. That said, Apama. Though, any 3rd party vendor package is pretty much like a Tradestation or Wealth-Lab for CEP/ESP.