Recent content by maxdama

I've used AlgoDeal a little bit. Their backtester is free and has a lot of data available. It's easy to use but also pretty powerful.

I haven't looked into it much. You'll probably just end up doing more work. I recommend joining a prop shop. I'm going to stop watching this thread fyi.

Only one I know of is Jeff Ryan's IBrokers package for R: http://www.google.com/search?q=R+API+IB The rest are mostly derivatives of the Java API and have the same design. It's not just the implementation, the documentation is also lacking and lags the current release.

Yeah agreed. Lime's API is very nice. Seems like an obvious business opportunity since IB's API has been "mature" for half a decade.

IB seems to be the only choice.

Data is included, fees may apply to some feeds. No futures, they recently got into options but traditionally have been exclusively equities. I know for a fact the minimum is 100k. I spoke to someone there two weeks ago, and a year ago. You may have signaled you were going to be an...

Commission + fees. I think GS provides margin and gets the interest on it. The only fee I can think of is an optional 400/month in their managed colo facility. The volume requirement is 20,000 shares per day across all stocks.

Minimum is 100k if you generate very high volume (eg HFT strategy).

I've written a couple of recent tutorials on trading through Matlab, in addition to the older ones that have been circulated - Getting historical data Design patterns So I think it goes without saying I'm open to collaboration... feel free to pm me.

Here are some sources I've kept track of: Book data for NASDAQ and NYSE. Only a few days of data, but it's 9GB each and for high frequency you could get 10 trades in a day in each of 100 symbols, and suddenly it's not such a small number of data points for a backtest. 5-minute bars from...

Stox, The original #, ex 9, is determined by how flexible/curvy the object you're trying to identify is. For example you need fewer points to fit a linear regression model than a neural network model. The neural network is more flexible/curvy in general. You're right that it depends on...

I just wrote a note in response to this on my blog maxdama.com. Here it is copied and pasted: Here are my opinions on the API's I've worked with. I recently sent them to someone and thought they'd be worth sharing with anyone who wants the opinion of a programmer and trader who's used them...

To build a more general automated trading system with Python that connects through Interactive Brokers rather than that prop firm, here's a tutorial that I made as I went through it. It works well. Just thought it might be useful. Regards, Max maxdama.com

Thanks imbetbm. It's Max Dama, not Max Mama. Maybe I should change it though... Regards, Max

Corey, I'd like to know more about how you use EC2. My only problem was that Matlab required some ridiculous licensing to run on multiple computers in the cloud. I'm sure it would be good using a different language. And I just so happen to have been credited $2100 toward AWS by Amazon...