AI Strategic Mistakes List For Markus
 

#1 Playing defense In

With a runner on third and less than two outs late(6th inning on) in a close game,the AI needs to play their defense in to protect themselves from allowing a run on a ground out.This is very important.

#2 Bunting

With runners on first and third,and one out,the AI should NEVER bunt unless it's a suicide/safety squeeze.

With a runner on first and one out,bunting is acceptable only with the pitcher at bat.

With runners on first and second and one out,bunting is again acceptable with the pitcher at bat or a very weak hitter.

with a runner on second and one out once again only the pitcher or a very weak hitter


#3 Intentional Walks

It has been proven that an intentional walk is a generally poor strategy except with runners on second and third and two outs with a great hitter at the plate.The AI issues way way too many intentional walks that more often than not come back to bite them.



#4 [B]Stealing bases

Never attempt a steal of second with runners on first and third and one or two outs,Never.especially late in a close game.

Never attempt to steal third with no outs, never.

These strategies go for double steals as well.



Ok Markus,these are the most glaring problems,by fixing these,you'll make the AI's in game decisions much more life like which in turn will make the game more challenging.Please implement these ASAP.

I agree with most of this post but never is a very strong word.



I think Vince Coleman or Ricky Henderson would tend to disagree with never steal third with no outs. I use the 20 scale for player ratings and if the AI had a 20\20 speed\stealing rated player I wouldn't have a problem with it trying to swipe a base in almost any situation as long as it wasn't an automatic 100% he's going thing.

In my current career I have a perennial steal leader that isn't even rated 20/20 and I have no problem anytime the AI sends him. Off the top of my head he's probably a 20/18 or at least in that neighborhood. I would like the AI to recognize it has a once in a generation player (the 20/20 type) and use him.

So instead of never maybe have the AI go with never for anyone below a certain speed\steal rating and the guys with 20-speed/20-steal no special restrictions and maybe prorate down to guys with a 17 steal rating going very rarely. Perhaps only against low rated throwing catchers and pitchers that don't hold runners well.

This is a case though where hard data could probably be found to support your "never" option or reveal what real MLB tendencies have been in these steal situations. I'm sure it's a very limited chance of players attempting steals in your example and Coleman and Henderson are extreme outliers in the history of MLB, the OOTP 20/20 type player that FWIW, in my dynasty, I don't think I've had.

Not trying to knock your ideas just that word never tends to bother me.

Now, maybe the AI cannot be made to take into account the 20/20 Coleman-Henderson, once in a million type player. Or maybe it's too hard to code or not worth the effort? I don't know I'm not a programmer.
If the AI cannot do this and the choice is to stay with the current system or go to "never", than I would go with never. It's the more realistic of the two choices.

Obviously,there are exceptions to the rule,I mean the players you mentioned are two of the best base stealers in the history of baseball so maybe they get a pass but 99% of the time its a terrible idea to try to steal third with no outs when a base hit will score them from second anyway.Hell,two groundballs and they'll score.In general there's a 50% chance that you'll be thrown out on a steal attempt,making it a sketchy proposition at best.

In real life, I believe that for steals of third. In the game, it seems to be no more difficult than steals of second (ie. about 30% thrown out).

Actually, the succes % of stealing third is higher than stealing second on league average. The reason is obvious, only good stealers try stealing third.

In nearly two seasons worth of play time,the league average has around a 53% success rate for stolen bases.Unfortunately,there was no breakdown for 2nd and 3rd base.Mind you I'm in a 1906 historical league so fictional leagues as well as different eras in historical leagues may differ.Bottom line is this,the 53% is more or less inline with R/L

This sounds more like the cliche "You should never make the first or last out at third base," which has some wisdom, but isn't gospel truth.

I also think that, regardless of what Sabermetrics says about the foolishness of intentionally walking batters, the game should do it as much as it happens in real life. (My personal managerial strategy is pitching around batters more frequently, and rarely issuing an intentional walk, but I want the computer to act like most real-life managers, when at default settings.)

Actually thats all that I want,the AI to behave like a real manager would.I would never have the audacity to think that Markus would take my recommendations as gospel,and implement them word for word.these were just suggestions to be looked into.

I think I've seen the gist of this thread posted in the fundamental flaws thread, so I'll close this one now.