I need to do cointegration analysis on large sets of financial time series for testing and hopefully to then generate csv files that will be used by an automated trading platform. Ideally the platform would do the calculations but first there is a lot of analysis to do. I need someone who can not only program but has a solid mathematical background in cointegration to deal with issues such as analyzing two series one a monthly and one a daily.
The cointegration analysis needs to be robust and able to the following at a minimum.
determine if the series are cointegrated long term but also based on a rolling window where I can vary the lookback period.
I need to also determine the lead-lag relationship or causality which changes over time. At times the Series A will lead Series B, and vice versa and at times they are coincident or in a some range of the equilibrium long -term relationship.
It also must be capable of analyzing lags and changing the lags based on the result in the rolling windows.
Finally I need to determine the coefficients of the VECM that tells me how much each series typically corrects deviations from the long-term relationship.
Other issues include the total breakdown in the relationship for certain periods of time and determining optimal lookback windows.
Maybe cointegration is not the right technique for the information I'm trying to get from this analysis? Maybe I should be doing correlation analysis, granger casuality, an Ornstein-Uhlenbeck model etc. The program should be robust enough plug in some additional techniques since the data inputs will be almost exactly the same.
I don't need graphing functions etc just a csv file output with the results. This could be easily done with Matlab and some other statistical packages but all I have is Eviews and R so if you can use one of those the functions will already be built in and you can focus on the GUI for specifying certain parameters, managing input data files and outputting csv files for further analysis.
I need someone who is not only a good programmer but also understands the equations and work with me to make sure I'm getting what I need.
I would like to find someone good that is interested in doing this and other quantitative financial time series analysis and programming.