What to do...I imagine that as I gain experience and track accurate data on my portfolio this will allow for greater and greater accuracy in capex projections.

Just like a picture is worth a 1000 words, this process is worth a 1000 data points.

After gathering that data, start running a cost analysis that shows the total installed cost of each type of asset.

If you're filtering MojoDial / Propstream's data but does it indicate if its their primary residence? 

Quote from @Kevin Sobilo: ShowMojo wouldn't have the data to evaluate how often a unit turns over.

@Stefano MitranoAlthough Boston can be difficult for great returns, breaks all of the "rules" like the 2% rule, and as a high barrier of entry, there is tons of data available online.

I'm new to all this and I don't know which pieces of data can help best filter my decision.P.S: If anyone knows a good sites with an API that has data to work with, I would much appreciate it.

If you are planning to do this to make money, make the investment to get the data that gives you the best chance of success.

In many cases, they don't go through the purchases if they get them (after they look at the property), and yet, HUD hasn't stopped awarding them winning bids.That's just one piece of empirical data, but it's definitely holding true in that case...

I first filter out the ugly people...that eliminates about 95% of the leads...Then I sort out the idiots and liberals...That leaves me with the moderately attractive conservatives that I can work with..I actually have a VA I've used for years that pulls the MLS data for houses in the zip codes I want.