SportGrid Radio SportsGrid
NFL NFL Strategy Articles Premium

NFL DFS DraftKings Super Bowl Showdown Simulations

NFL DFS DraftKings Super Bowl Showdown Simulations
Print Friendly, PDF & Email

NFL DFS DraftKings Super Bowl Showdown Simulations

Throughout the season, I’ve used our Range of Outcomes (ROO) methodology to provide simulations and probabilities for players in different formats – sometimes a player’s chances at being a top positional value or raw scorer, sometimes a player’s chances of finding themselves into the optimal lineup. This week, with a huge prize pool for the DraftKings Showdown slate, I simulated the slate 20,000 times, while using a correlation matrix in conjunction with simulated percentile outcomes, to glean the following information below:

-Overall Exposure Percentages

-Captain Exposure Percentages

-Where Correlation Is/Isn’t Important

-Leverage Strategies

If you’re unfamiliar about our ROO methodology (ie floor/ceiling projections), you can read about them here:

In a nutshell, we’ve been able to leverage our historical projected baselines matched to actual performance to find out which baselines are most fragile and most important in a linear model when projecting percentile outcomes.

As a quick example, we’ve found catch rate has a higher relative weight for a wide receiver’s floor outcomes versus their ceiling outcomes, while yards per target (YPT) sees its importance meaningfully rise as percentiles increase. For a quarterback, projected pass attempts matter much more floor than ceiling, while team total plays a much larger impact at the high end of the ceiling projections.

One aspect that our ROO methodology does not account for is historical team or player specific variance. You can read why in the link above, but it’s likely that may result in lower ceiling projections than there should be in some situations: Tyreek Hill, Mecole Hardman, and the SF RBs all come to mind.

A final caveat before we dig into the specific exposures: Throughout the season I had *felt* that the high-end DST outcomes were populating too much in my sims, and as a result, I artificially deflated their top-end percentile outcomes. As a result, it’s possible their exposures below lower than they should be.

Overall Exposure Percentages

Below are the exposure levels of each player in the 20,000 optimal lineups generated from each of the simulated projection sets. I made a couple of assumptions in terms of controlling the player pool that you should be aware of, excluding players like Anthony Sherman, Deon Yelder, and Richie James but including players like Ross Dwelley and Darwin Thompson. If you don’t see a player listed below, it means I didn’t include them in my simulations; it does not mean they didn’t make any of the optimal lineups. It’s difficult to decide which of the 0 – 1 projected point players are worth considering or not. The other major assumption made (taken directly from the current DailyRoto projected baselines) is that Tevin Coleman will play and split carries with Raheem Mostert (30% rush share expectation was used in the sims below, 47% for Mostert).

This content is for members only.

NFL NFL Strategy Articles Premium