2016 Fantasy Baseball Sleepers Cont’d and Ending the Position Scarcity Argument

If you don’t have an issue with position scarcity adjustments, then I already furnished 2016 Fantasy Baseball Rankings with Steamer Projections and highlighted sleepers last week.

Naturally, position scarcity remains an argument, but there is one solid way around the argument: find the best value (projections relative to average draft position) within the scarcer positions and prebuild a roster with options heading into the draft; then you can focus on best available value and not position scarcity. I will explain and provide options below.

Below, is an embedded file of updated (1.5.2016) NESN NFBC Average Draft Position and Rotochamp Composite projections (Rotochamp & Steamer). This time, I did not adjust the rankings for position scarcity. Everyone is ranked simply by their relative value (hitters to all other hitters and pitchers to all other pitchers). That value is then compared to ADP. I also included the sorting capability so that you can manipulate the file:

The first column provides the Rotochamp composite rankings. The second column provides the ADP rankings. The third column depicts the differential. Again, it’s shaded by elevating shades of Green (value > ADP) or Red (value < ADP).

If you scroll all the way down, you will find Rotochamp projections without associated ADP (Kenta Maeda for example) and ADP without associated Rotochamp rankings (Rich Hill, Jose Berrios, Aaron Hicks as examples). They either didn’t show up in the top 300 hitters or pitchers for Rotochamp or the top 750+ picks in the ADP data.

Updated Sleepers/File Sorting:

It’s currently sorted by ADP rankings, but if you sort column three (“Diff”) descending followed by column four (“Pos”) ascending, then you will see the largest value within each position. Feel free to download and re-highlight some of the guys you prefer. I did so in the embedded file already. For example:

Again, these rankings now exclude position adjustments. Peralta/Semien/Escobar are not being valued as Shortstops. Schwarber/Posey/Norris are not valued as Catchers. If they were, their rankings would jump and the ADP differential would become that much more impressive.

Prebuilding a Roster:

This approach allows you to ignore position scarcity. Instead, you can look within the scarce positions and find the greatest value. NFBC leagues (23 roster spots; 15 teams) as the format, let’s do this for Catchers, Middle Infielders, Third Baseman and Relievers because I don’t like drafting them too early:

I thought David Ortiz (7) at U and Hanley Ramirez (9) at CI were also too good to pass up.

The latest RP value I went with was Glen Perkins (13), Will Smith (18) and Jason Grilli (23)

You can find this framework depicted in the last tab within the embedded file, titled “Prebuilt Roster.” You might have to click on the “…” tab toward the bottom right of the file.

Posey/Schwarber, Norris, Kang, Escobar, Peralta/Walker, Ortiz and Hanley left us with an uber good framework of counting stats. Now I can draft two aces in rounds 3-6 and another one or two in rounds 10-11 while having four slots to draft the best possible early value. I’m thinking Rizzo/Miggy in round one; two of Felix/Carrasco/Kluber in rounds 3-4 and 20/20 options in rounds 5-6 (Kemp/Puig/Ellsbury/Polanco) or power (Fielder/Freeman/Agone/Pujols) if you went with a balanced combo in round one over Rizzo/Miggy.

Watcha think? I fart in position scarcity’s general direction.





Daniel Schwartz contributes for RotoGraphs when he's not selling industry leading thermal packaging. You can follow him on twitter @RotoBanter

16 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
CB
8 years ago

curious what you think about aggregating projection systems/rankings then using a Gaussian mixture model to spit out tiers?

kind of the best of both worlds, I don’t think you can dismiss the idea that top tier talent is scarcer at some positions more than others, but establishing some nuances to that (tiers) instead of just rolling with a black/white hypothesis of scarce/not scarce is the way to go, in my opinion.