by tersignf » Sat Oct 08, 2005 10:47 am
There's only one way that I know of to attempt to get a reliever in to start, besides an injury, and that isn't the way. It's not sure-fire, but then again people shouldn't try to do that anyhow in my opinion.
I agree with everyone on HAL, but one thing you have to realize: the more settings you make static (i.e., choosing a tendency), the less options you give the AI. Sometimes that's good, but sometimes there can be very unintended consequences because of the limitations you've set. A primitive AI should be given flexibility in some cases, and in others as little as possible.
For instance, when we use the avoid lefty/right preference, many times we're expecting way too much of the AI--especially in the case of reverse pitchers. The only way to maximize success with this, and some have done it a few times with very successful bullpens, is to have a boatload of relievers for HAL to choose from. On some teams I had, he would go through 5 relievers in 3 innings for great matchups and then give it to the closer. I've had some great...and some terrible, splits. So in a nutshell--have plenty of relievers available for avoid lefty righty situations--likely won't work as expected with more expensive guys, including starters.
Further...think about how basic the check marks are and don't expect them to be cure-alls for one-sided types. If a team has 2 LH / 7 RH bats, and you have Beckett pitching in the 5th with a 1 run lead, with 2 LHs coming up in the 8 and 9 hole, do you want HAL to walk those 2 guys to get to Mora or Kendall or Freel at the top of the order?
My point here is the AI can be used to your advantage in some situations, but it's very limited. It still comes down to a balanced team--because if you have a weakness HAL and your opponents will find it.
Now this doesn't answer your question on Radke. It's stupid...and unfortunately, expected every now and then--but there was a reason for it...somewhere transparently buried in the roots of the AI's decision tree was a logic trail that led to it, but you'll never know.
All the more reason that if I can't see HAL's decision algorithms, I'm not going to depend on or expect much out of him. That dog won't hunt.
Cheers...don't get mad at HAL--just come to grips with the fact that it sucks and you won't be surprised when he does something stupid!