I'm looking for software and a data feed that will allow me to calculate and chart the price correlation between two securities. Currently I use Interactive Brokers, but I cannot do this w/ their Trader Workstation. I'd like to do this w/ as high frequency data as is possible on a solo day trader's budget. Generally, I'm trying to find the best combination of broker, data feed, software and API that provides fast execution, fast data, can calculate and display charts of price correlations, and will allow me to write my own algorithms for processing streaming data and charting these custom algorithms' output. I'd like to be able to get very detailed w/ custom algorithms.. For example, I'd like to experiment w/ custom algorithms to calculate probabilities from streaming data along w/ historical data, using both price and volume as inputs. I'm hoping to find the best, most full featured system(s) now, so that I don't spend my time learning a system that I will outgrow, then have to learn an entirely different system in order to progress further. Thanks, any info is much appreciated -Max PS- I mainly trade the major ETFs: SPY, EEM, GLD, TLT, commodity ETFs, and eur.usd
I did this a while ago using eSignal data feed and a Ninja Trader indicator which I coded. It would have two charts in it and do the correlation studies from there. I since abandoned it and am using my own platform coded from scratch with either eSignal or Interactive Brokers data feed via the TWS API. My conclusion with regard to correlation studies was that they are not worthy signals. I looked at the same ones you listed and there were no arbitrage opportunities there. Nothing lags in this market, and nothing is oversold/overbought based on another correlated underlying, if that's what you're looking for. It's also not useful for creating buy/sell indicators for one underlying based on another. Statistically it came out pretty random, and I did lots of studies. Either way, any data feed with Ninja Trader would be easiest, then just make your indicators if you know how. However NT will get very slow if you compare lots of them via the grid (I ran about 60 and it started getting unusable, the code was efficient though, NT is slow).