NFL DFS DraftKings Divisional Round Quarterback and Stacking Strategy
I’ve switched things up for the Divisional Round. Instead of providing the raw and value probabilities this week, I ran 10,000 simulations, accounting for team and opponent correlations, and found the optimal lineup within each of those simulations.
I’m going to focus on the QB position and post how often each player appeared in the optimal lineup in those 10,000 simulations. We can compare those “true exposures” to market expectations and some gleaned correlation information from the sims to develop +EV roster construction strategies and leverage positions for this weekend’s games. I’ll add in some of the more important non-QB notes throughout the article, where relevant.
Before we get started, I do want to make one point about using the sims information properly. It’s my supposition that the sims naturally are light on the very best plays and high on the worst plays, for reasons I covered in my Week 6 article:
“One point I failed to make in my Week 3 article in regards to the sims has to do with floor. My sims show that the chalk pieces are often overowned based on the percentage of the time they finish outside the Top X at their position.
While this is true, I did fail to flush out a couple of concepts that may still lead you to owning the chalk a little bit higher than, for example, a WR’s Top 4 value and raw total probabilities. Perhaps, Michael Thomas last week had a Top 4 probability of only 18 percent (this is a guess, I don’t have last week’s sims as I was on vacation). We do have to consider, based on a strong regular DailyRoto projection, that some of the 18% of the time he finishes in the Top 4 he’s a must own player (based on how much he outscores the field by). Conversely, we have to consider that some of the 82% of the time he finishes outside of the Top 4, those that beat him do so narrowly or are really low owned targets, and thus that type of finish for Thomas isn’t highly detrimental, especially if you care at least somewhat about min cash changes.
In simpler terms, the sims do a good job of showcasing the chances a player is the Top X value or raw total scorer at their position, but we still need to layer in their base projections as context since one player’s average score when finishing in the Top 4 may be more or less beneficial than another player’s average score when finishing in the Top 4.
You can view this through the lens of a non chalky player as well. Maybe some 1% projected owned WR had a 3% probability of finishing Top 4 at the position, but due to a lower average score/margin of victory when achieving those finishes, that WR’s theoretical leverage wasn’t as high as it seemed by purely looking at the finish probabilities.
This isn’t meant to discourage you from using the sims as a way to find leverage. I want to encourage you to do that. But I should have made it clearer in that Week 3 strategy article that the sims purely on their own aren’t the only information you need to make those decisions, and that’s something that struck me as important following a week like Week 5.”
It’s no surprise to see Jackson and Mahomes leading the pack. It’s a little surprising to see Jackson even with Mahomes given that the DailyRoto projections have both as similar values, which gets Mahomes into optimal lineups at a reduced cost. However, the range of outcomes for Jackson on the high end up are so great, due to his rushing ability, that it makes up for the more expensive price tag.
The tough part about this week is that these exposures likely line up pretty close to what the market is going to play. After Jackson and Mahomes, there’s a cluster of the 4 mid-tier QBs, all getting into lineups at a similar amount – Rodgers, Wilson Garoppolo, and Watson. There’s probably not a ton of leverage to be had on any one of those players, but if you’re MMEing with a spread approach, it makes sense to have ~ 10% exposure to all of them.
As for the very cheap QBs who won’t be owned, the sims have a clear preference of Tannehill over Cousins. Both are big road dogs against Top 5 pass defenses. However, Tannehill’s Titans see a larger percentage of their points scored via TDs. That, along with some rushing equity for Tannehill, leads to him achieving a “good enough” ceiling more often than Cousins, which gets into optimal lineups in the scenarios where multiple expensive skill players separate themselves from the pack. In my personal MME runs, I’m tempted to X out Cousins altogether and shift a bit more ownership to Tannehill, who is unlikely to be owned much higher than 5%.
While it’s valuable to see how often each QB is in the optimal lineup for each sim, the more valuable information comes in the form of roster construction hints. Can you run Lamar naked or do you need to stack him? How comfortable should you be with a game onslaught (4+ skill players in the game plus the QB)? With the sims, we can put some math to it and look at the percentage of time a QB was stacked, double stacked, had a bring back (any opposing skill player), or was part of a game onslaught in the sims where this QB was in the optimal lineup.
Now, the sims aren’t perfect and may be underrating the times that Jackson hits his ceiling via rushing, but he’s so expensive, that in the cases where he scores enough to be in the optimal lineup, it’s likely that at least one skill player is in there alongside him. Also, for as run-heavy as Baltimore is, their pass TD rate does not skew too far from expected relative to Vegas line. We’re projecting a 59-41 pass to rush TD ratio this week.
It also helps that Jackson’s stack candidates are affordable. While it might be difficult to pick the right guy, Mark Andrews ($5,600), Marquise Brown ($4,400), and Snead/Roberts/Boyle/Hurst (sub-$4,000) are all cheap. With the exception of Andrews, the other receiving candidates could potentially get to where they need to be on a single play given their cheap cost and a small slate.
Between the 75% stack rate and 65% bring back rate, you may want to correlate your Jackson lineups more than you’d expect at first glance.