Earlier this year, we released our team expected win totals and playoff odds for the 2021 season. These projections are based on the FanGraphs Depth Charts, which blend Jared Cross’ Steamer projections with those of Dan Szymborski’s ZiPS, with playing time allocated by our own Jason Martinez of RosterResource fame; the playoff odds then simulate the season 20,000 times, taking strength of schedule into account.
We thought it’d be interesting to duplicate that process, instead using the Average Total Cost (ATC) Projections.
The ATC player projections have been available on the pages of FanGraphs since 2017. Similar to the Depth Charts model, ATC is an aggregation of other projections. While most other accumulation models typically apply equal weight to all their underlying data sources, ATC assigns weights based on historical performance. The method is similar to what Nate Silver does with his political forecasting model at FiveThirtyEight.com. You can read more about how ATC works in the introductory article here.
Over the past few years, ATC has consistently been one of the most accurate baseball forecasting models. Indeed, according to FantasyPros, the ATC projections have been the most accurate baseball projections over the past two seasons. The success of ATC stems from minimizing parameter risk across the player pool. You can read more about inter-projection volatility, how to use the ATC projections on the FanGraphs site, and what is new for ATC in 2021 here. Read the rest of this entry »