S80 ENGINE LAB · RESEARCH BUILD 002
Build the baseline.
Then beat it.
S80 now has a multi-season historical research universe. The goal is not to claim a finished algorithm—it is to establish the consensus information available at the time and measure whether deeper S80 signals improve prediction.
327prospect-season snapshots
176unique players
90outcome-linked snapshots
2017–21archived ranking years
Consensus control group
The model S80
has to outperform.
This baseline uses only contemporaneous FanGraphs rank, FV, age, level and position. It deliberately excludes the deeper contact, swing-decision, production and trajectory signals we want S80 DNA to exploit.
Consensus baseline = FV core + age/level adjustment + small rank adjustment + position contextImportant distinction327 snapshots does not mean 327 fully modeled careers.
Research methodology →The archive layer is now large enough for historical comparison. The deeper outcome/tool/production enrichment is still being expanded. S80 will not turn incomplete archive records into fake precision.