Quote:
Originally Posted by Markus Heinsohn
Couple of things, the AI does not pick the best player in the inaugural draft at #1 all the time, there is some randomness involved, otherwise all drafts would be the same which is something I don't want.
Regarding general draft AI in historicals, there are several issues here and it depends quite a lot by the way you set up your league (recalc on or off, use draft values etc) so I'd need to know which settings exactly do not deliver expected results, otherwise I'd be fishing in the dark.
|
The Gus Dorner/Joe Tinker/Johnny Evers example was from a 1901 amateur draft in a straight historical that Reed was playing with the "Use pre-defined draft value for AI" box checked. In no universe is Gus Dorner in the same league as Joe Tinker and Johnny Evers. No way, no how, no where. He probably should've been close to the last player picked in the draft, but he was in the top ten, and Tinker and Evers were not. Something was changed radically about the pre-defined draft values between OOTP18 and OOTP19, and that is why Scoman started this thread and why others have jumped on board.
With that box unchecked anything can happen in a historical league. Babe Ruth can even go 8th overall in a random debut draft. This is actually more plausible than the Dorner/Tinker/Evers example because things will be much more randomized when that box is left unchecked. Babe Ruth is arguably the best player to ever lace up a pair of cleats (relative to those players that he competed against), but once that box is unchecked, anything can, and does happen. That's why I said (currently) the historical is damned if he does check the box, and damned if he doesn't.
I'm actually enjoying putting together my own draft values. The process is much different for a random debut inaugural draft than a random debut amateur draft because of the players coming in at all different stages of their careers, whereas in the amateur draft, everybody is at the beginning of their respective careers. I'll throw my support behind those that don't want to do what I'm doing though because I believe the "Use pre-defined draft value for AI" function can work much better than it currently does, and did work much better than it currently does in OOTP18 and other previous versions. Somewhere between OOTP18 and OOTP19, the draft values were overhauled, so that's where any search should begin IMHO.
Here's what I'm talking about from
this thread in the Bug Reports forum:
"Here's what I get from moseying through the OOTP19 MLB Master.csv file.
1. George Mullin, 2331 Draft Value
2. Addie Joss, 1587 D.V.
3. Bob Ewing, 1501 D.V.
4. Bob Rhoads, 1156 D.V.
5. Charlie E Smith, 1017 D.V.
6. Otto Hess, 993 D.V.
7. Carl Lundgren, 984 D.V.
8. Andy Coakley, 835 D.V.
9. Fred Glade, 690 D.V.
10. Gus Dorner, 633 D.V.
11. Johnny Evers, 598 D.V.
12. Chappie McFarland, 562 D.V.
- - - - - - - - - - - - - - - - - -
??. Joe Tinker, 353 D.V.
Here's what I get with the same 13 players in OOTP18 Master.csv file.
1. George Mullin, 1958 D.V.
2. Joe Tinker, 1806 D.V.
3. Johnny Evers, 1784 D.V.
4. Bob Ewing, 1130 D.V.
5. Addie Joss, 1102 D.V.
6. Bob Rhoads, 819 D.V.
7. Otto Hess, 808 D.V.
8. Charlie E Smith, 726 D.V.
9. Carl Lundgren, 660 D.V.
10. Andy Coakley, 548 D.V.
11. Fred Glade, 516 D.V.
12. Gus Dorner, 474 D.V.
13. Chappie McFarland, 404 D.V.
Hmmm...Something must've happened between OOTP18 and OOTP19 to change these draft values so drastically. The best person to ask about that is the database master himself, Spritze in the Historical Database Issues thread.
In OOTP18 1902 debuts, there's also Harry Bemis (704 D.V.), Ernie Courtney (558 D.V.), Red Dooin (1290 D.V.), Patsy Dougherty (1233 D.V.), Tom Jones (1058 D.V.), Ed Phelps (633 D.V.), Lew Ritter (462 D.V.), Doc Smoot (680 D.V.), and Heinie Wagner (983 D.V.). What I'm saying is something changed drastically between OOTP18 and OOTP19. Whether this was intentional or a bug is what we need to find out."