Like you said before, working combinations is good.
I stuffed up on the Knights/Storm game the other day, because I had Ponga and Best across my teams but always individually never actually put them in a combo, which is a poor oversight from my part.
Making sure your teams follow a game script is generally how its phrased. So if Player A benefits from player B doing well, get them both, and reconsider tackle bot hookers if you are expecting their team to win big.
Normally I work on at least 3 teams. One for each team winning, and a 3rd which is based on a close game or just ignoring matchups and just going on who might score highly.
Whilst I avoid POD talk in overall comps, its crucial in DFS
If you think everyone is going to be on a certain player, have at least one team without them, so that if they fail you can get a big jump. the more teams you enter the more viable this strategy is.
eg for Roosters Dogs it was a calculated risk to have a team without Teddy. In this case no teddy meant you lost, but if he'd had a quiet game, a team without him suddenly has a massive jump on the majority. if you only enter 1-2 teams probably best to keep players like that in.
Also if you think you have a good POD, go for it. I got 3rd in the Cowboys/Warriors game the other day by getting 10% owned Peta Hiku's hattrick. What cost me top spot was ditching Molo when he went from starting to the bench - low ownership and killed it.
Note you can't see ownership when picking a team so its a bit of gut feel as to who will be popular.