I did this exercise but it's been a few years. My first thought was to rank everybody on a Points Above Replacement per $ (PAR/$) basis, but then I found I'd end up coming up with a lot of the just sub-mediocre starters for each position. This goes against my auction theorem 0.2 that you aren't just going for a team of biggest discounts, but trying to use places where you think you'll get a discount to put together the best team (since after the auction it doesn't matter what the players cost anymore, only their stats they'll put up and their trade market value matters). My thought is that after the auction you aren't likely able to trade two Hunter Pence types for a Carlos Gonzalez, even though 2x the $ cost of Hunter Pence might equal the $ cost of CarGo.

So then I somehow normalized PAR/$ and PAR and created a weighted average. I regressed on how to best weight those two to get some metric that takes into account both the player's value and how his value compares with his price. I believe it ended up being close to weighting PAR/$ about 33% and PAR about 67%. Players who ranked the highest on this scale because targets for me and players that rated the lowest became players to avoid. I'll look back and see if I still have the old excel sheets to find out exactly what I did.

It might not be the best way to do it, but it incorporated the main two goals I have - getting the most top-end talent I could, but do it in a way that also maximizes my entire roster's projected points.

And yes I did just want to use "theorem" and "metric" in an internet post about fantasy baseball.