Phil Alexander : This is a great question. With so much analysis out there, it's sometimes tough to decide which numbers actually mean something important. For me, by position:
Quarterbacks - Primarily efficiency stats, including yards per attempt and passing touchdowns per attempt. I also look at how the opposing defense performs in these metrics. Team run/pass splits in the red zone are important for analyzing touchdown ceilings. Vegas lines are probably more important than any of the above. I want a quarterback with a high implied team total (at least 24 points) and the more his team is favored by, the better.
Running backs- Workload rates like snaps, total touches per game, percent of team backfield touches. Targets and receptions take on added importance on DraftKings (full PPR). Overall goal line carries, percent of team goal line carries and percent of goal line carries converted into touchdowns. Same as quarterbacks, Vegas team totals and point spreads inform my decisions. The more a running back's team is implied to score and the greater the spread is in their favor, the more interested I am in the player. Matchup matters a lot here too. Austin Lee’s Normalized Strength of Schedule tool does the heavy lifting there and boils several advanced stats down into easily digestible team rankings.
Wide receivers- Targets and percent of team targets (market share) -- overall, from inside the red zone, and from inside the 10-yard line. If a player who hasn't had much of a chance to contribute is suddenly thrust into a bigger role, I'll consider efficiency stats. Yards per target is more readily available, but if you can get your hands on yards per route run, even better. Wide receiver-corner back matchups matter at the extremes. If it's clear a wide receiver is lining up on the majority of his routes against a corner who has been consistently roasted all year, I'm interested. Likewise, I'd prefer to avoid the Patrick Peterson's of the world. Cornerback stats are harder to come by, but they are out there. Wide receiver air yards (or yards at the catch point) are important because downfield plays allow you to pile up fantasy points quickly.
Tight ends- Mostly the same as wide receivers, though I'm paying a little more attention to matchups. It seems like every year there are a handful of defenses that simply can't cover the position. The 2015 Raiders and 2016 Browns immediately come to mind
Team Defense- My choice of defense is usually more about game script. I love picking big home favorites. There are enough lousy quarterbacks in the NFL today to target a team defense against a QB who has been sacked at an alarming rate and has a low completion percentage (errant passes tend to get intercepted more). I'll also consider sack percentage and yards per pass attempt for the defense, as well as the opposing offense.
Jeff Pasquino: I like a lot of what Phil said here, but at a certain point I think the metrics can lead you astray. While I will definitely look at CB/WR and TE/defense matchups and the like, as well as game scripts for kickers and defenses, overall I like the simpler numbers. Targets and percent of team targets are great numbers to use for receivers, and volume of touches is key for performing well. You need the ball to rack up fantasy points, so touches are key for me. Finding those weak corners can really help, however, when you are looking for the wide receiver that might get two touchdowns and win you a tournament.
David Dodds: This isn't an official stat, but I like to look at target/opportunity distribution after injury. A star player goes down and will miss next week. What happened in this week's game and how will that translate to targets/opportunities for everyone else in the games without the star player. Did the team run more? Did the pass/run ratio stay similar? Who got the red zone looks? In DFS, it's all about predicting what will happen next.
My other advanced stat I like to look at is what I would call similar players. Two weeks in a row the slot WR has torched the Indianapolis defense. Can this week's player do the same? Is he similar to the two players who excelled before him? Can he run the routes that were wide open in previous weeks? The NFL is a copy-cat league. When one coach exploits a weakness, expect the next week's coach to try to as well.
Chris Feery: I’m in line with Jeff’s approach. Simpler is better in my world, as getting bogged down into too many advanced stats can lead to analysis paralysis for me. It really comes down to the approach that works best for you, and zeroing in on that will lead you on the path to success. Some folks need to see an abundance of stats from every angle you can imagine to find their selections for the week. Others can focus on just a few key pieces of information to reach their conclusions. There’s no right or wrong way, it all comes down to what you’re more comfortable with. Those that like a lot of info can feel skittish about their choices when they only have time to look at a few pieces of info. The opposite camp can feel overwhelmed and make bad choices when they go down the rabbit hole of information. Finding your personal comfort level pays dividends over the long run, and it will also make your DFS experience that much more enjoyable.
Justin Howe: I once heard a great quote from the incomparable Bill James that I've mostly forgotten, so I'll paraphrase: If you run across a metric that tells you Willie Mays wasn't very good, you can confidently disregard that metric, regardless of how sexy it is. Don't fall for some obscure stat that tells you to play Dirt-Cheap Quarterback X. Examine it closely. It needs to be remembered that there's a wide gulf between predictive stats and "stats for stats' sake." Don't merely tell me that Matthew Stafford has been great lately and expect me to apply it to this week. Tell me he's dominated against the blitz, and that this week's opponent relies heavily upon the blitz. Don't just tell me Melvin Gordon has been great on the goal line; tell me why you expect him to be in the red zone a lot this week, and why you're confident he'll get a lot of those chances against Opponent X.
The metric also needs to be sticky. There are helpful-looking measures out there like yards-per-target, deep-ball success rate, and several others that don't carry over well from one week to the next. In my experience, the stickier ones are the matchup-based offerings that David mentioned. How often do teams run the ball against Defense X? How well does this defense cover premier slot receivers?
And, as Jeff Pasquino pointed out, look for numbers that are rooted in volume. In both DFS and season-long football, we're all much better and more consistent in projecting volume than efficiency. Yardage and touchdown rates tend to vary wildly, but the top dogs in an offense typically don't. It's much safer to invest in a guy because he projects to a high workload than due to an expected/hoped-for swing in efficiency..