OpenQuant + AmiBroker: my recipe for building robust Automated Trading Systems

thanks dr. fokin for your answer! As far as I know, the quanthouse products are not available for non-institutional users and according to that what I got from many user forums far away from what we would be willing to pay just for the basic routines of automated tradings! However, it would be great if there would be a COMPLETE demo framework for openquant for automated trading for at first to develop it maybe further as an openquant application on an opensource project or at least to have a good starting point for further developments.
Thanks!
 
Quote from ophth1:

... I am still not sure, if it might be the best idea to throw away OQ, amibroker etc, and just use visual studio e.g. c#, which is indeed capable to do an rapid event driven and object orientated software, in combination with the IB interface. That would you give complete control about all events. What are your directions for the future? thanks!
More...
You have diagnosed the sorry state of affairs we are in as retail ATS developers. I have not found the one 'great' integrated retail platform that gives me everything I want in a) exploration and strategy backtesting, b) system development, and c) ATS deployment.

A couple of ideas:
- If your strategies are relatively simple (from a state-transition point of view -- they can be mathematically as complex as you want), you may also be interested in looking at TradeStation. It is the most "integrated" backtesting and automated trading environment. They also provide historical data so you won't need an additional feed. Whether it works for you or not depends entirely on the type of strategies you deploy -- it would be worth checking it out if you haven't already.

- There was an initiative started on the Automated Trading forum to build a low-level open-source framework targeted for programmers (labeled tickzoom): http://elitetrader.com/vb/showthread.php?s=&threadid=147374 which then migrated to http://www.elitetrader.com/vb/showthread.php?s=&threadid=148025
Nothing has been released yet. I also started losing interest when most of the dialogue shifted to licensing models. I will play with it when the alpha comes out and may even contribute to the code. However, I wouldn't suggest building an ATS with it for some time.

- I have seen people refer to Tradelink http://code.google.com/p/tradelink/ in forums, but haven't used it myself. Seems like the same idea as Tickzoom, with a different feature set, but further ahead in development/usage.

So where does that leave me. Well, I continue to use AmiBroker for exploration / strategy development, and OpenQuant for deployment, as I described in my original post. I think that's a good starting point if you want to focus on your strategy logic (making money) without losing time with the underlying bit shuffling.

When I first posted the message, I thought I was finally going to be able to give up custom coding. Alas, things were not to be. I am back to creating custom C# applications written directly to the InteractiveBrokers API for execution. For this, I have been expanding my own framework as I need new functionality (modeled after the OpenQuant/SmartQuant class architecture which makes it easy to move strategy code). But at least, its optional. A custom app is something I now do at the very end, for select strategies, to gain additional flexibility. You may need to create your custom apps at some point, but I suggest you invest your valuable time in your strategy logic instead until you see your strategies making money in live trading.

I plan to keep using all three tools/modes for the foreseeable future.
 
Hi, Dareminator

Have you been playing with static variables in Amibroker ?

http://www.amibroker.org/userkb/2007/04/21/static-variables-in-rt-systems/

Static variable is important than anything else in deploying an ATS in Amibroker. I stumbled on various issues when I was writing my first 1-minute ATS in afl. But when I began to l sort out all things related to improper handling of static variables, things became much easier.



Quote from dareminator:

No I did not try TWSLink. It looks interesting though, I might play with it some time.

My true issue with using AmiBroker to drive automated trading has been the programming model: When my code can only be run from scratch every n seconds, the polling, the error handling, the logic for multiple instuments, and trying to cram multiple strategies into one file (with the help of #includes) is too much.

Also, I am now working on developing a strategy that keeps positions for less than a minute -- and I would really like to respond to an event when it happens, not at the next polling interval.

The event-driven programming model of OpenQuant allows me to simplify the autotrading code. I can keep the program logic and the possible execution paths in my mind when I read the code. For me that translates into robustness and lesser suprises.
More...
 
Hi dareminator! thanks again for all your advices! However, I am a little bit sorry to hear that my "diagnosis" seems to be right instead of hearing: oh, this is the ONE software to go...
Nevertheless, I understand your three approaches and we will checkout both open source projects, too, but both appear to be in a very preliminary stage? Just another question about openquant in terms of speed comparison? If you would like to "scan" in a first step a large part of the market (e.g. 5000 symbols) for some strategy events and then putting those into a higher observation in the second step, I assume that the event driven system of openquant will be suit best the second part, trying to get some strategy events from a few symbols. However, what about the speed of this event driven model in comparison the the array machine amibroker if you want to scan in parallel 5000 symbols with IB as realtime data feed provider? Is or how is this possible to do something like this in openquant or is the overhead of processing time due to this event driven model too extensive? THANKS!!
 
Quote from dareminator:
OpenQuant installs a large number of sample projects. The best way to learn is to download a 30-day trial version of OpenQuant and play with it in a paper-trading account. A piece of code I copied from their "Bollinger Bands with 5-Minute Bars" sample project (http://www.smartquant.com/strategies/strategy_14.htm)
Code: 
	public override void OnBar(Bar bar) 
	{ 
		if (bbl.Contains(bar.DateTime)) 
		{ 
			if (!HasPosition) 
			{    
				// cancel previos buy limit 
				if (buyLimit != null) 
					buyLimit.Cancel();             
             
				// calculate limit price 
				double buyPrice = bbl.Last * (1 - Percent / 100); 
             
				// place new limit orders 
				buyLimit = BuyLimitOrder(Qty, buyPrice, "Entry"); 				
				buyLimit.Send(); 
			} 
			else 
			{ 
				barsFromEntry++; 
          
				// close position at the second bar after entry 
				if (barsFromEntry == 2) 
				{ 
					barsFromEntry = 0;                
                
					sellLimit.Cancel();
					
					Sell(Qty, "Exit (Second Bar After Entry)");
				} 
			} 
		} 
	}
[/B]
More...

Dareminator,

Just a question about this system, I read about it in James Altucher's book, i dont really get how it works...

Going by what is said on the link you provided, it would seem that maximum profit would be 1% and maximum holding time would be 2 bars.

But the stats in the book say that average win is 2.72% and average holding time for losing trades is more than 5 bars??

Have you tried testing this system in real-life? How did it perform?
According to the book, over the 1 year it was backtested, it took 113 trades with hit rate of 92% and average win/loss > 1.

Quite phenomenal.

Keen to hear your thoughts.

Nizar.
 
Quote from thetrendfollowe:

Dareminator,

Just a question about this system, I read about it in James Altucher's book, i dont really get how it works...

Going by what is said on the link you provided, it would seem that maximum profit would be 1% and maximum holding time would be 2 bars.

But the stats in the book say that average win is 2.72% and average holding time for losing trades is more than 5 bars??

Have you tried testing this system in real-life? How did it perform?
According to the book, over the 1 year it was backtested, it took 113 trades with hit rate of 92% and average win/loss > 1.

Quite phenomenal.

Keen to hear your thoughts.

Nizar.
More...

Nizar, I haven't tested this system. I have a few different anti-trend strategies in my ATS arsenal, but this isn't one of them.

I don't have Altucher's book with me, I assume he might have used different parameters than the default settings in the sample code. Good thing with mechanical systems is you can test and see for yourself before committing money.

Good trading.
 
Hi Dareminator,

I am also using AmiBroker for backtesting ideas and trying out OpenQuant trial for the actual automation attempt. Can you comment on which brokers provide paper trading account that OpenQuant can connect to? I would like to see the automation working in real time before I trade with OpenQuant.

Thanks in advance,
 
You must log in or register to reply here.
Back
Top