Depends how much the data costs IMO, I purchased some NBA season data from a website from the past 6 years but only because it was cheap as chips and was formatted in excel exactly as I wanted.

I'd say data based on previous 3-4 years is getting close to enough so if you could source maybe 07 and 06 on top of that i personally believe that would be enough but others may have a better idea. Also i'd suggest you need more data for the systems that have had fewer selections over the last 3-4 years. Whereas the systems that have had more selections you probably need less additional data.

Not sure about your mentality towards systems but I like the idea of having something in play every 1-2 days even if it means my POT and ROI is lower then someone looking for the perfect winner twice a month or something. It's personal preference but thats just how I am, so that might help point you towards which systems you want to concentrate on and determine how much additional data you need.

The other idea is rather then purchase additional historical data is to actually forward test these 14 systems over the next say 3-4 months (if you have the time up your sleeve and not in a hurry).

Lots of options and paths you can go down, depends a lot on your mentality IMO. If it was me i'd probably start live betting them with small money but others would say thats a rubbish idea because its not proven against data you haven't used to create the system or test it. But thats just the gambler in me, i'd prefer to give them a shot that they work .