Anton is there market data timestamps precision limits? can I import timestamps in windows ticks (1/10000th ms)?
DateTime is stored in the historical database as a number of ticks (i.e. long), so I think the answer is yes. You don't see milliseconds or fractions of milliseconds in OpenQuant GUI windows such as the data manager, but you can print trade.DateTime.Ticks from a strategy code and check it yourself. Cheers, Anton
But don't forget this http://blogs.msdn.com/b/ericlippert/archive/2010/04/08/precision-and-accuracy-of-datetime.aspx when you are talking about fraction of millisecond accuracy
windows time problems arent that grave. otherwise there wont be soft realtime windows trading system and there are. at least one.
Basically people run time critical HFT systems under Linux / c++. Not just due to DateTime issues but due to the fact that you have no control of what .NET is doing right this moment (sending an order or doing Just in Time compilation of the portion of code that should send this order )
Just curious why do you need nanosecond precision in .NET DateTime. It's a good thing that microsoft people use ticks to store datetime anyway since it can be utilized in the future under better hardware / OS. We are currently working with other platforms where standard datetime objects have millisecond precision. I am trying to understand if we should use a better time resolution for trades/quotes timestamps right from the start or if this is ok to have it just till one ms.
No problem. I am interested in precision down to 0.01 ms at the moment which is far away from ns. I personally achieved timings down to 0.05 ms (tick to order) on W2008R2 barring Eric's bullsh1t without much of optimisation. However the precision I asked in the beginning is needed to store and process data in OQ running it under Windows, right? I never said the data is originated on a Windows server. Obviously no unfiltered ticks data provider (in his right mind) will use Windows for it. Hence the timestamps have better precision - all 1/10000.
Hi Anton are you the same Dr Anton behind OQ? I had a close look at OQ. I've also looked at RightEdge, TradeLink etc. I really really Like OQ for it's long range of adaptors for data/execution connectivity etc. but I just want to understand why do you not spend some time on the documentation? I really don't get it? yours is the worse I've seen for any software? here you have an amazing, capable product that probably is better than the rest, yet the document is just useless (dated 2007?) I'm convinced you're a very smart guy, who knows how to code, build software, but you seem to have completely missed an important aspect. you're methods take a unique approach which is different to the others use, e.g. creating cross overs etc. almost you're personal way of programming (which is ok) but even for someone who can code in c# it takes a long time grasping your way of thinking and you've left it for others to figure out. this isn't about c# or python, it's the the thinking behind the methods provided. it took me around 20min in rightedge to build a strategy i had in mind to test and see how it works, it gives you a whole good list of performance output, maybe 10+ panels. As for OQ? this is the 2nd day I've spent on OQ and I'm still confused, and looking API help doc is just painful. right now when I run something with the demo strategies provided I get a SINGLE chart showing the equity curve? is this real? I'v spent 10min trying to see if there is more output analysis breaking down the performance e.g. sharp ratio etc and I can't find it and I thought i really must find you and speak to you. here is an institutional product with big fails in the most basic areas. I'm either a complete idiot if this is proving painful or that some others agree with me. I wish someone could take the method thinking of rightedge or tradelink and include the long list of FIX adaptors hot point etc that's included in OQ supporting multiple strategies etc.