Messages posted by: ytalmor

Currently we provide high level architecture document, and additional details for each component (see links in the same page). We also provide Javadocs.

In the future we might release additional design documents and a more detailed architecture document. I will post a link in the forum when we do.

Thanks,
Yoram

We are using QuickFix/J in our platform (including ORS/MTS).

We do mention Quickfix C++ in our documentations for a very specific and limited use case of compiling QuickFix/Ruby bindings when compiling libraries used by Tradebase.

Thanks,
Yoram

This summary is correct. Note that both the exchange simulator, market data interface and the random market data are not "real" market data - it is simulated data. The main difference is that the market data interface on port 7003 reflects the "book" for the symbols being "traded" by users of the exchange simulator while the random market data doesn't - it is just random market data (usually higher volume of updates).

Thanks,
Yoram

Hi,

We posted couple of sample strategies here. Those are intended for demo purposes only. Hopefully these examples will help you understand how headless Strategyagent and Photon can be used with such strategies. The samples do not make use of all our APIs, but they do show how to use strategy parameters and they implement a simple "real life" strategy (Order Slicer and VWAP approximation).

As always - your feedback is appreciated.

Thanks,
Yoram

Please check my response in this thread - specifically the requirement to have a unique senderCompID when connecting to our Exchange simulator.

Thanks,
Yoram

Another potential reason for this problem is that you need to have a unique senderCompID. When you check out the source code we have a default senderCompID, however each connection to our simulator requires a unique senderCompID. You can configure it in srctest/cmd_exec/conf/user.properties

For additional info, please check these instructions , specifically search for "Broker Disconnect" in that page.

Thanks,
Yoram

Please see my response in this thread

Thanks,
Yoram

This means that the order you created was sent to MTS (ORS), however MTS can't forward your order to a Broker. Either your MTS broker configurations are incorrect, or ORS can't connect to the configured broker. If you are using our "out of the box" configurations then MTS is configured to connect to our Exchange simulator. In that case make sure your system clock is accurate (if not on NTP) as the exchange simulator will disconnect you when your server time is skewed.

You should also check MTS (ORS) logs in the ors/logs directory, file is called ors.log. You should see this error message and potentially additional information in that file.

Thanks,
Yoram

Udi,

The strategies don't need to be "compiled" before they are executed - it is done by the strategy module in the Strategy Agent on the fly.

Thanks,
Yoram

The reason I mentioned client API is because I wasn't sure what are the set of problems you are trying to address. Strategyagent is our recommended way of deploying strategies. We have a few very basic examples packaged with Strategyagent that you can check out and the above example posted by Colin should help you as well. Beyond that it is just going through our strategy API Javadocs or here and writing your own strategy.

Hopefully that will help. Please let us know if you have specific questions/examples you are looking for.

Thanks,
Yoram

Also check this short documentation that provides an overview of how to use our client API:

http://www.marketcetera.org/confluence/display/MDN/ORS+Client+-+Connecting+Applications+to+Marketcetera

As the documentation mentions, OrderLoader is a simple client you can use as a reference/example of how to send orders to MTS.

Note that the client APIs are meant for basic interaction with MTS and are "lower level API" compared to our strategy API. Also, the client API are not stable yet so it might change between minor releases (though we do try to maintain backward compatibility).

Thanks,
Yoram

We will add more samples soon. I will either post them here or a link to the documentation.

In the meantime, if you have specific questions regarding our strategy API documentation, either in our wiki Strategy APIs, our Java strategy Javadocs, or our Ruby strategy Javadocs - please let us know.

Thanks,
Yoram

There are multiple ways of how to approach backtesting. Currently we don't offer it as a built-in feature of the platform though it is on our roadmap.

You can use our csv module to read in historical data (OHLC) and stream it into your strategy, but that is different than simulating all the trades/bids/asks, etc that happened throughout the day - so it depends on your needs, how you implemented your strategy, and will require you to add custom code to the modules in our modules framework.

Thanks,
Yoram

Please see my response to your other post at this thread.

Thanks,
Yoram

Did you follow the instructions described here:

http://www.marketcetera.org/confluence/display/MDN/Building+Marketcetera+Platform

If you follow those - you should be able to compile and run ORS from source. Make sure you are using the correct version of 3rd party tools, especially maven. I recommend you clean your env (mvn clean:clean) or check out the code again and follow those instructions.

Thanks,
Yoram

Footer Links