factual
Predictive Models Fail On Rare Events
Models built to predict rare events suffer the 'sunspot problem'—untuned, they become excellent at predicting that nothing happens because nothing usually does—so you must tune the reward function to produce the desired output; in the draft, this means tuning models to seek upside rather than predicting merely above-replacement players.
factualpending
Speaker
Daryl MoreyEvidence Quote
“you have to tune the reward function on these predictive modeling problems... if you don't tune your models to look for players who have a lot of upside, you'll get... something that's good at predicting just a slightly above replacement player.”
Created: 6/13/2026, 12:05:01 AM
Your notes
Loading notes...