We are a small algorithmic trading shop with 6 mathematicions, 2 statisticians, one operator and an administrator. We have been working with NT and IB for a while now and have had reasonable sucess. My team however is constantly frustrated with data discrepencies. For example, we recently ran some comparisons capturing data from IB in NT by using Market Analyzer and then comapred that data to data downloaded via the Historical Data manager via Kineticks.
The discrepencies are not minor.
We are not high frequency traders. We use multitime frame in most strategies. Our entry signals are mostly based on one minute close Price bars and our post entry trade management on 1 second or tick data for calculating exits based on dynamic statistical categorization of expected future values.
This is our most challenging issue. What is the best way to assure that the data used for backtesting, in these intervals, will best replicate the real time scenario? We have tried varios approaches and some have helped, but none that give me the confidence I require to put more capital to work.
We would appreciate any help or recommendations an experienced trader on this platform and data set would be willing to provide. We also would not rule out a reasonable fee for successful solution.