ARTICLEFANTASY
RoboScout’s Top 100 Fantasy Baseball Prospects For 2024
February 13, 2024
February 14, 2024
Dylan White
0 Comments
Orioles third base prospect Coby Mayo during a March spring training at-bat.
Orioles third base prospect Coby Mayo during a March spring training at-bat.
Image credit: Coby Mayo (Photo by Tom DiPace)
When we put together our Dynasty 700 and the subsequent Top 100 Fantasy Prospects, we mined from our scouting reports and fantasy expertise to put it together. We also had an unsung contribution from RoboScout.
A huge part of fantasy baseball involves projections. This makes sense, of course. If you can estimate, as accurately as possible, the potential contribution of a player to your fantasy roster, you can make better informed decisions on how you want to draft your team, build your roster and even make trades. Projections are a necessary ingredient for fantasy success in redraft leagues.
Although the cumulative accuracy of projections dissipates as you move further and further into the future, these same principles hold true in a dynasty league. If you can estimate, as accurately as possible, the future (yearly) contributions of a player to your dynasty fantasy roster, you will have an advantage over other leaguemates who are less rigorous in their approach.
What is RoboScout?
The high-level basis for how projections are created boils down to one simple truism: past performance—despite what the legal disclaimers on your 401(k) may say— are related to future returns. By looking, for example, at the average paired-year performances of hitters and pitchers historically, weighting by sample size, adjusting for survivor bias, one can generate expected age curves with reasonable accuracy of various statistics, such as walk rate (of both hitters and pitchers), strikeout rate (of both hitters and pitchers), ground ball rate (of both hitters and pitchers) and home runs per plate appearance and OPS (of hitters). Given a hitter’s OPS, walk percentage, and strikeout percentage, one can reasonably infer what their batting average is, and so forth.
We can apply this same approach to the minor leagues. Take paired-”level” performances of hitters and pitchers historically, one can estimate what a pitcher’s strikeout rate would be in Double-A given that he had, say, a 12% strikeout percentage in High-A. By understanding the expected equivalent performance at a higher level—including MLB—we can thus generate an “expected” major league performance based on a minor leaguer’s performance (after additionally adjusting the statistical performance to the league’s run environment and also from Matt Eddy’s park factors. Now add in the “age curve” calculations from the previous paragraphs to this expected major league projection, and you can estimate what the hitter’s projection would be in his prime performance years.
Depending on how deep you wanted to go—for example, incorporating platoon splits, quality-of-competition or deriving independent age curves for different “phylums” of similar hitter archetypes—more granular adjustments can be made.
Hitting a home run on opening day does not imply a player will likely finish the season with 162 home runs. Likewise, we also apply regression to more accurately reflect expected season-long performance based on performance from small sample sizes.
The final piece to the recipe is minor league Statcast data. Supplementing the performance inputs used in the “projections” for hitters are barrel rate, exit velocity, contact percentage and other metrics that are shown to be correlate to future wRC+. For the pitchers, RoboScout folds in the pitch-level metrics (movement, velocity, etc.) that are inputs into traditional Stuff+ models.