I have a peculiar trading strategy and I can't seem to be able to find a proper way to measure its performance.
The strategy consists of buying certain stocks and then selling them in short time periods - but for now let's say every long position is terminated after 10 days. Correspondingly, if we run this strategy over a year, the body of data that I need to analyze consists of a multitude of data points such as these: (date purchased, price purchased, stock ticker, % return over 10 days).
First: I need to find a way to somehow standardize the returns. The problem is that returns vary significantly in magnitude over the 10-day period and I found that annualizing them would result in weird numbers like -35000% annual returns for some cases. Plus, I am not entirely sure if annualizing returns realized during just a 10-day holding period is right to begin with.
Second: I need to eliminate the market movements from the picture in order to get net returns provided by the strategy alone.
Third: I need to aggregate these multiple return %s (per each acquisition-disposition) in one metric for the whole investment strategy. The end result after this would be a single time series of standartized (annualized or whatever) daily returns earned above market for the whole strategy.
How would you approach these problems? If there are any other things that I might have missed (apart from standartizing returns and excluding the market), please let me know. Thank you.
TL;DR: What is a proper way to calculate performance return time series for an investment strategy that consists of establishing long positions at various times and then terminating such positions in several days? Such performance metric should exclude the general market performance.