[ChrisG]

Originally Posted by Lukas Berger

The default AI evaluation is 25% for the current year's stats, and even there, the full 25% won't necessarily kick in until players are at pretty close to a full season's sample size.

If you want the AI to react more quickly/strongly to bad seasons, you'd really probably want to boost that by quite a bit.

Now that being said, in both these cases, these lines are actually better than their rl numbers so far this year, and yet Cleveland and Tampa are so far sticking with both guys.

Now, yeah maybe they wouldn't if they go the whole year putting up numbers as bad as they have been. (though in Naylor's case, he's been putting up not dissimilar, though marginally better lines the last two years as well, and Cleveland has stuck with him the whole time).

On a more big-picture scale, if you look at 2025 batters in real-life MLB, there are a LOT of guys who got significant playing time hitting well under .200. This is just the current MLB environment. Guys hitting below .200 but still getting significant playing time is not really unusual.

EDIT: With all the above being said, I do think it's possible some of the lower expected BA guys in the roster set like Mullins and Naylor and Joc Pederson could use a very slight boost to their babip numbers to get them to a touch higher baseline. ZIPS and our algorithm were arguably coming in a little a bit too pessimistic on them (though again, given their rl life-numbers so far, maybe not). So we'll adjust them up very slightly in the future to make them a little less likely to bottom out in their worst seasons.

Makes sense...thx Lukas