CSV Data Backtesting

First of all, I'm not sure if something like this is better suited to be web based or a local piece of software. My plan is to sell it as a service so please let me know which you think will be better suited. Description: I need something developed that can look through 100s or even 1000s of csv files based on user selected filters/range of settings to backtest. Each csv file is 1 day worth of data, so a user would select a start and end date for the period he wants to test. They would then select filters based on how they want a lineup of players selected each day for their entire timeframe. Each lineup from each day will be calculated and their total scores averaged for their timeframe. The goal here is to find the best algorithm to automatically predict future lineups with the best projected score based on historical analysis. This analysis would need to run through multiple passes based on the range of filters a user selects. Example: I want to backtest the entire 2014-15 NBA season to see which combination of filters produces the highest average daily score. I want to exclude players that are under $4000 to start, and then increment by $100 up to $5000. So the user would check off this filter they want to use, and enter their start value, increment and end value. With my example, the backtest will have to run through the entire timeframe of data (2014-15 season) worth of CSV files 10 times and output a report for each one calculating what the average score was. The more filters used will increase the required number of passes exponentially. So using a lot of filters will require a lot of computing time and resources. Hence my concern for having this as a web based solution, or local based. Do let me know what you think the best course of action is. I've attached an example csv file of 1 day. A lineup needs to have a certain number of players consisting of certain positions. The bolded 'Actual Score' column is what we are optimizing to find the highest total of each lineup. If you have any questions please ask. I will obvious provide much more detail to the winning bidder with images etc...