@Lemming78 Here's my two cents. Fundamental factors are accounting semi-fictions, but they're the best thing out there. The only way to become familiar with them is to read other people who are also familiar with them. So books and articles that are strong in accounting are going to be good places to dig up factors. There are tons of academic papers published in accounting journals that will be useful. There are textbooks like those by Stephen Penman or Aswath Damodaran. There are free online courses offered by the Corporate Finance Institute and other free online accounting courses from Yale and NYU, I think. Many Seeking Alpha articles discuss accounting practices in a knowledgeable way (and many, as ZGWZ points out, don't). But in general, hunting this stuff down is difficult. Why are balance sheet accruals and cash flow accruals so different, and what sense does that make? Banks use the EBITDA-to-debt ratio to measure solvency, but there are plenty of other debt ratios out there (e.g. current ratio, quick ratio, debt to total assets, debt to free cash flow)--how good or useful are they? Do you have a solid understanding of why you use unlevered free cash flow with enterprise value and levered free cash flow with market cap or why an EBITDA-to-market-cap ratio wouldn't make sense? Do you understand why preferred dividend payments are subtracted from net income when FactSet or Compustat calculates earnings per share? If you're using the most recent quarter's ROA, what do you do about companies that report semiannually? Why are non-debt liabilities excluded from net operating assets? These are all thorny problems when dealing with the data available from Portfolio123.
In my opinion, just using a list of factors and letting a machine tell you which ones work and which ones don't can only take you so far. Even worse would be asking a machine to create new factors. Factors have to make sense in the world of accounting, because that's the only world in which fundamental factors exist. Factor investing is not and cannot be scientific (see the most recent P123 blog post). Too many P123 users try to create investing systems without understanding the accounting rules they're using.
Personally, I think that using a bunch of factors that you don't understand defeats the purpose of factor investing and will get you into hot water. Maybe I'm wrong.
But I am very glad of the question. Just the fact that you're asking these questions shows you're on the right foot.
To get to specifics:
Reading 10Ks and 10Qs is only useful when you know what you're looking for. It's time-consuming and deadly boring. It's necessary when trying to grasp how certain factors work, and it'll be homework if you're taking an accounting course. So it's secondary, not primary. But going through a few is a very good idea.
Screening is incredibly useful when doing factor research. I would use it hand-in-hand with whatever else you're doing.
If you see a factor doesn't work on its own, there's really no good way to know if it could be a good augmenting factor for others. You could plug it into a ranking system and see if there's an improvement. If it's not going to do you any harm, throw it into the mix and see what happens. The factor has to make sense to you, though. Avoid factors that don't.
Lastly, cast a broad net. Trying to focus on a specific industry is going to be very tough because there's simply not enough data to go by.
