Taking one for the Team (Too Often)

[motnahp]

Running with 1.41 Beta. Set up as multiplayer, with all teams under human control. I have noticed that AI has been allowing relievers to give up 15-20 runs per outing in some cases. This is likely a result of the tweaking done to limit relievers' appearances.

For example, I have one reliever for San Francisco who has given up more than 10 runs in an appearance nine times. His most ridiculous line was: 1 2/3 IP, 10 H, 30R, 28 ER, 26 BB, 0K. He is not the only one, just the most extreme I have noticed. His stats for the year: 25 appearances, 17 2/3 innings pitched, 105 hits, 239 R, 234 ER, 205 BB, and 14 K. He has a 119.21 ERA. For some reason, even with four better pitchers in the minors, AI won't demote him. He's not locked on the roster, so it makes no sense.

I know the risks with Betas, so I'm not upset. I just feel this is somehing that [X(]MUST [X(]be corrected before the next official release. Prior to d-loading 1.41 Beta, I was running the same setup with 1.20 and this issue never arose.

Shaun, if you'd like, I can zip this assn up and send it to you. Other than this issue, I've noticed nothing unusual with 1.41 Beta.

[motnahp]

Any doubt (and I had very little) that this problem was not introduced with a recent Beta is now gone. The San Francisco reliever has poisoned the association. In a recent game, Philadelphia beat the Giants 159-4. No, that is not a typo. The Phillies had innings of 63, 36, 15, 26, and 16 runs during the game. Our hero was allowed to toil for 5 innings, allowing 43 hits, 156 runs (153 earned), walked 123 and struck out no one. He allowed 5 homers.

On the SAME DAY, a Chicago reliever pitched 7 innings and gave up 18 runs in a 25-3 loss. He came into the game in the first inning when the starter got shelled. I'm sure there are dozens more examples of this lunacy, so I'll just shut down the assn. until I can revert back to 1.20 or another official release. I've learned my Beta lesson.

[puresimmer]

It must have something to do with the fact that you run multiplayer mode with all teams human controlled? Would you mind doing a little experiment for me where you setup a league EXACTLY as you normally would, but don't set all the teams to human controlled?

That must be it, because nobody else is reporting such craziness.

Thanks for sending the file BTW.

Shaun

[motnahp]

Will do, and thanx for the prompt response. I'll need to begin from my last backup, prior to the 1970 season. I should have something to report on this tomorrow.

[XCom]

Guys, take a look at this thread...

tm.asp?m=1298608

Seems like the same kind of runaway problem. Just thought I'd throw it out there.

[motnahp]

I think I finally got it figured out, but it took some trial and error. I tried making all teams computer-controlled, but I was prompted that all teams had to have exactly 35 players and a certain number of hitter and pitchers. That was not possible in my assn. I hit "OK" and got stuck in a loop. I escaped with CTL-ALT-DEL.

I remembered having all teams' Hook set to "Very Patient". I had done this to counter the effects of too many reliever appearances. Changing all teams to Average Hook had no noticeable affect on weird innings.

I went into the corrupt assn and looked at the pitchers' ERAs. I had about a half-dozen guys with 10.00+ ERA. They had all experienced weird innings. A couple of them didn't really belong on the MLB roster, having only pitched briefly in the bigs. The worst offender was Chicago's Jim Cosman'70. He had only one real-life appearance in 1970, but the AI insisted on keeping him in the big leagues. I just decided to NOT import him at all. After a six-week test, there were still several pitchers with the strange innings and high ERAs. To the AI's credit, when I went to each team and Auto Adjusted the rosters, most of these clowns were sent down to the minors. The Cubs kept two, only because their minor leaguers were scrubs.

I then went to every team and made almost all of them have 4-man rotations. This seemed to give them better relievers, although in a couple cases quality starters went to AAA while mediocre relievers were in the bigs. I also looked at every MLB pitcher and adjusted the Player Usage settings on all relievers with a Stuff rating less than 50. I made all such pitchers mop-up men. Another six-week test resulted in only one clown pitcher, Jim Colborn of the Cubs. He had only one weird inning, allowing 15 earned runs.

With Colborn and three others, I decided to retire the 1969 varieties and import them again for 1970. Thanks to Shaun for making this option possible with 1.41 Beta. I did another 6-week test and had ZERO weird innings. I'm not sure what change made the difference.

Early in the testing, I had one incident that I could not explain at all. Pete Mikklesen of LA was a solid reliever with 98 Stuff. He had made 16 "normal" relief appearances and had an ERA of 3.66. In his 17th appearance of the season, he allowed 16 earned runs in 3 innings, raising his ERA to more than 7.00. Nothing about LA's pitching staff was suspect, so I just counted it as a fluke. I'll run a few more (longer) tests over the next few days.

[motnahp]

Running with 1.65b, doing a 2006 replay and acting as manager of the Braves. The problem mentioned in this thread is still popping up. AI managers will put a scrub pitcher in the game and leave the poor soul in to receive abuse no pitcher should endure.

After I manage my game, I sim the other games of the day. I try to keep my eye open for any games where the winning team scores 20+ runs. When I catch one, I look at the box score and see what happened.

Almost without exception, an AI pitcher gives up double-digit runs over an inning or two, often with an unrealistic number of walks. A recent example:

Atlanta and Houston were tied 1-1 going into the top of the 7th. The Houston AI manager brought in Phil Barzilla to start the 7th inning. 16 runs, 5 hits, and 14 walks later, the Braves had a 17-1 lead. Incredibly, AI sent Barzilla out to start the 8th inning! As the Atlanta manager, I bunted a couple of times, the only way I could think of to end the madness.

In fairness, I know that the AI would keep this pitcher in the minors indefinitely if given the opportunity to do so. Because I want the AI teams to use the rosters they had available on a given day in real MLB, I forced Barzilla and his "5 Stuff / 5 Control" to stay on the roster. Luckily for AI Houston, this pitcher only stayed on the roster for a few days, then went back down to the minors to stay.

I'm not sure what the perfect solution would be to this dilemma. In a past version of PS, those seldom-used pitchers and batters were changed to "5 point-scrubs". IMO, a conversion/lowering of these players was needed, but to take them all the way down to 5-point scrubs was a bit of overkill. A player good enough to make it to the major leagues is very rarely so much worse than his peers. Even the worst pitchers in history aren't allowed to walk double-digit numbers of batters in an inning.

I don't claim to be a prophet, but with the new feature of September call-ups.....guess what? More AI teams will have more of these low-rated players playing more often, leading to more of these anomalies. A scrub batter doesn't have a great impact on a game, but a 5-point scrub pitcher, left in the game too long........different story.

I'll do an experiment soon, copying my association and manually raising the ratings on 5-point pitchers to see how they fare. I'll post my results later.