OOTP Developments Forums - View Single Post

LansdowneSt · 05-03-2022, 02:40 PM

Yup, agree with all that, sprague.

After things settle with 23's release, I'd like to learn more from the development team on if it treats players that have "MLB" stats but still reside on the MiLB csv file differently when creating the ratings. I think we all agree the NeL real stats are in desperate need of upgrade, but even I'll put that to the side and look at... say, Heliodoro Diaz.

In my recent sim (recalc every year, 5-year span, retire to history), Helio won the 1928 Cy Young for the Red Sox going 24-8, 3.04 ERA, 302 IP, 132K, 3.5 BB/9 and 3.9 K/9 but that same year's OOTP "real life stats" have him at 3-14, 4.56 ERA, 140 IP, 100K, 3.5 BB/9 and 6.4 K/9. His surrounding years aren't spectacular either. That was league average for that NeL season.

I'm not sure if I used bRef's Stathead and looked up a comparable pitcher in the AL or NL for those real stats, I'd get someone that ought to be a CY Young winner. It's a sim, so there's a chance I would. But even anecdotally, I think something isn't being considered by the AI. By that I don't mean a league quality dampener for the NeL. I mean, if Helio pitched that line for the Browns or the Sox in real life and the AI was converting those legit MLB stats to game performance... I'm not sure he'd be a CY Young winner. So, I'm inclined to rack my brain thinking of a technical AI translation problem before moving on to it being an NeL problem. For example, the same issue could arise if we took the Pacific Coast League and changed those stats to major league. Would the AI treat the PCL stat line X differently than the MLB .csv stat line Y... even if X = Y. Know what I mean?

05-03-2022, 02:40 PM	#14
LansdowneSt Global Moderator Join Date: Feb 2004 Location: From Duxbury, Mass residing Baltimore Posts: 7,226	Yup, agree with all that, sprague. After things settle with 23's release, I'd like to learn more from the development team on if it treats players that have "MLB" stats but still reside on the MiLB csv file differently when creating the ratings. I think we all agree the NeL real stats are in desperate need of upgrade, but even I'll put that to the side and look at... say, Heliodoro Diaz. In my recent sim (recalc every year, 5-year span, retire to history), Helio won the 1928 Cy Young for the Red Sox going 24-8, 3.04 ERA, 302 IP, 132K, 3.5 BB/9 and 3.9 K/9 but that same year's OOTP "real life stats" have him at 3-14, 4.56 ERA, 140 IP, 100K, 3.5 BB/9 and 6.4 K/9. His surrounding years aren't spectacular either. That was league average for that NeL season. I'm not sure if I used bRef's Stathead and looked up a comparable pitcher in the AL or NL for those real stats, I'd get someone that ought to be a CY Young winner. It's a sim, so there's a chance I would. But even anecdotally, I think something isn't being considered by the AI. By that I don't mean a league quality dampener for the NeL. I mean, if Helio pitched that line for the Browns or the Sox in real life and the AI was converting those legit MLB stats to game performance... I'm not sure he'd be a CY Young winner. So, I'm inclined to rack my brain thinking of a technical AI translation problem before moving on to it being an NeL problem. For example, the same issue could arise if we took the Pacific Coast League and changed those stats to major league. Would the AI treat the PCL stat line X differently than the MLB .csv stat line Y... even if X = Y. Know what I mean?