How many teams, and what is the payout structure? Depending on those details, I would possibly do things a LOT differently. You also have to take into account how sharp your opponents are. You start getting into DFS ownership game theory and things like that. If everyone in the league is choosing Adams, Diggs, Kelce, etc, what do you really gain by having them? If the goal is to win the league, you want to have players that are not owned by many/all the other players, so if they do well, you are gaining on the entire field. That isn't you say that you want to pick shitty players and just pray that they do well...but there is a balance, and it depends on A) how many opponents are in this contest B) how good you think your opponents are at identifying the obviously best plays (which I think you largely have) and C) how many players you think are galaxy braining it like me/we are. PPR also has a big say in it. For instance, I think you are right that you want to burn K and DEF on likely-to-lose teams in general, and most especially in PPR formats. But a DEF on a good team could easily outscore a smith schuster or the like in standard. Bills D will likely go unowned, so if they get a few TDs on their way to a SB appearence, you have a big leg up on beating everyone. ...I think you'd need to be up against more than 12 people to consider something like that tho. Also consider that the 1 seed teams' ceilings are capped at 3 games, while the better 2 and 3 seed teams have legit shots at 4 games. Depending on format, you might want to load up on positions that will tend to score the most for those 4-game-upside teams largely for value, and most especially if you don't think other opponents will roster them. Edit: Just saw that its half ppr...hmm. Decided to look up median projections. See below.