I can't get through 7th 8th or 9th?

[Brian SCEA]

The two keys to succeeding in late game are knowing when to pull your starter, and transitioning to relief pitchers.

Starters
As a starter gets tired, he loses control of pitch location more than anything else. In fact, some pitchers gain velocity late innings while very few lose more than 1mph (there's even a hidden attribute that handles this). Keeping this in mind, when a starter is tired you want to pitch around a bit more. This is because it's much more important to avoid throwing easy pitches down the middle than to avoid throwing balls.

How does that work? For simplicity, imagine a normal pitcher misses in a circle of 6 inches radius. Let's say in the 7th inning, he misses in a 6.5 inch radius. By moving all his pitches out .5 inch (for example), as far as deep strikes are concerned he's no worse off than before. The drawbacks are the batter can get ahead on the count more, you might bean or wild pitch by accident (which isn't that common all things considered), and you throw more pitches per batter faced. But since this strategy overall can keep a pitcher in the game you'd otherwise pull, it's a way of coping with a loss of control. This is especially more important for mid to low tier starters than for top starters.

Most starters can pitch 100, even 110 pitches until they hit 25% energy. It all depends on their stamina attribute. Below 25% energy, starters start losing control more rapidly. In a close game and a #5 starter, you might want to pull him at that point. But with a #1 starter or in a game you've sealed the deal you can certainly keep going. You'll underperform on average, but it won't matter because a win is a win.

It's worth noting that aside from pitcher attributes (HPer9, BBPer9, KPer9), energy is the second most important factor in pitching. I'd even say that in a typical game attributes are roughly 70% importance, energy is 20%, and confidence is 10%. The problem here is that if you drain every starter to 0 energy as an extreme, instead of being only 20% important it easily become 40% or 50% important. Don't let that happen in a game that matters - only run energy down to 0% because you don't care.

The penalties ramp up as you go below 25% energy. Likewise, from 100% down to 25% energy it's not a huge difference.

Relievers
Collectively speaking, mid relief pitchers are much worse than starters. In fact, the average mid reliever is a step worse than even the average 5th starter. And considering this is based on relievers pitching 1 inning each, they're even worse than their stats say. The main saving grace is that you use the above average relievers much more than the below average ones whenever possible.

With that in mind, in an important game you want to limit relievers to 1 inning each. A reliever with full energy is comparable to a bad starter with low energy - which is also why you don't just keep a starter in forever (he'll eventually cross the point of being worse than your relievers).

The closer is the exception, and that's why you only use him when it matters and save him otherwise. If you're going to lose, who cares if you lose by 5 runs or 10 after all? They're both the same as far as the season looks at it. Let the bad relievers handle the blowouts.

Relievers have to pitch, knowing that they're not as good. When the game is close, they have to pitch carefully rather than always to the batter like some top starter can. A good starter can miss the ball right down the middle and it'll still be un-hittable at 98mph. A reliever certainly can't, otherwise he'd be slated to start games himself.

The last issue is that as a video game, switching between any two pitchers can be jarring and takes some pitches to get used to. This applies to something as simple as jumping into an exhibition game, playing for 1 inning, and jumping into another with a different team. This was for me the hardest part about getting effective with relief pitching.

With all that said, even though relief pitchers are overall worse than starters this is only borne out in the long run. In any given game, it wouldn't be a surprise to see relievers shut out the side while the starter did poorly. It's just a matter of probability and how you manage that risk.

I used to have the same problems as far as relief pitching, but I found the above two keys to be the solution.

Other Notes
By the way, I do want to add that as far as the game plays and hits are determined, the hitting engine has no idea which team is AI or human controlled. The concept doesn't exist in the hitting engine, and it doesn't treat AI vs AI games any differently than human vs human games.

If it wasn't for the fact that the Show is a team-built video game with competitors, I'd publish the logic for hit determination and it wouldn't be that long. The rules are mainly the physical interaction between timing, pitch location, bat aim location and size, and player attributes. When you use a mode like timing-only, all the missing components are determined automatically using the AI code without any modification.

Take a look at the Swing Analysis for the AI after a pitch. Whatever hit or miss resulted, it would have been the same for you had you made the same inputs. Keep in mind player attribute and the difficulty setting both play a role here as well in scaling how big or small your batting cursor is.

This means the AI is responsible for picking the right timing and right batting cursor aim, then deciding if he should swing or not. The way the AI does this, it doesn't know where the ball is exactly or what the result of his swing will be. The AI wears "beer goggles" so even if the pitch is inside the AI might think it's down the middle. This is exactly why the AI will take strikes or chase balls, and (less trivially) why pitch break fools the AI correctly. A slider that lands outside might get the AI to swing, whereas a backwards slider on the inside might get the AI to take the pitch. A canned system is incapable of getting this to work correctly and naturally in every aspect. And that in summary is exactly why the system is double-blind. The AI doesn't know where the ball is and what the hitting engine will spit out, and the hitting engine has no idea who's controlling the bat.

[nomo17k]

Did I just see a god?

Damn that's quite some info!!

[brett the jet favre]

Quote: Originally Posted by merchant1874

Im the same. I seem to give up most of my runs in the 1st 2nd and 3rd.

Sounds like King Felix, after the 3rd, its OVER!

[bcruise]

Uh oh, someone let B-Ma out again.

Thanks for the great info Brian.

[miamikb2001]

Great info thanks Brian

[Se7enth_Sinn]

Quote: Originally Posted by Saviour13

I've played probably 60-70 games since release. I've probably won 25 of those. But i've lead before the 7th in probably 50. Why is it, whenever I get to the 7th or 8th innings, my pitchers hit a wall(doesn't matter if energy his full or low, or confidence is high or low) and I lose the lead. It just feels like everything hits a hole or I'll make errors I can't control (pop ups to the infield dropped) Any tips?

I've noticed that too, just get someone ready and take him out. In real life its the same, the 3rd and 4th time seeing the same pitcher is always an advantage for the hitter.

[stealyerface]

Wow.. Great info, and thanks for the insight into the behind-the-scenes coding and AI goodies.

This is not a "Hey Developer" thread question but more of a "Let me bouce this off you please" type of question.

With API set to the ON position, is there additional coding, tweaks, or adjustments made by the AI on the second, or third time through a lineup?

Meaning, if I am rolling with Josh Beckett and I get him into the 6th inning, does the aforementioned AI concepts you outlined, give a bump in the cursor, or the PCI for the computer, based on my selection of pitches throughout the game?

I know this is complicated, and maybe you cannot answer it, but when I was pitching competitively, I sometimes knew in warm ups that I had a very nasty split finger working on that day. But the truth is, I may not run it out in front of a hitter until the 4th inning, or halfway through the second time through a line up, and certainly for fastball hitters, I would wait til the tank was getting dry, and then start to throw that in the mix.

My question is that for the AI hitters, if I do not use a certain pitch at all through 3 innings, the confidence meter is pretty much gone to zero. So I might warm up with 5 curveballs and dial it in, but not use it at all in the game. Can I "fool" or have a better percentage of a dice roll against the AI if I "save" a pitch for later innings, that the AI has yet to be able to recognize in the API, and therefore, give my starter an advantage on the size of the batter's PCI, as well as his algorithmic advantage over a Guess-pitching AI batter?

Phew.. Hope that made any sense.

If you are able to shed some info without divulging anything proprietary, that would be great.

~syf

[stealyerface]

Also, you brought up a great point, and an even better reason that I have gone back to the Analog Pitching.

One of the things that I feel The Show has really gotten very good at, is the difference between pitchers, with regards to their wind up, and more importantly, out of the stretch.

After pitching with a guy like Lester, you can see that from the stretch, his yellow line on the Analog Meter is fairly close to the top line. This means, as a shorter guy, and with a compact pitching motion, the timing between the initial pull-back, and having to bring the stick forward is fairly short. This also keeps runners on base from stealing him wild, as his delivery timing to the plate is quite quick.

Micheal Bowden on the other hand, has a very long stride from the stretch, and as a result, you have to make an adjustment on his motion, and this makes pitching with different guys totally different.

You do not get this awesome difference with Classic or Pulse pitching.

Very well done.

~syf