Too many two out inning rallies by AI!! (Comeback AI Discussion)

[Segagendude]

re: Too many two out inning rallies by AI!! (Comeback AI Discussion)

I could've used a "miracle user comeback" today.
Tribe v. Rockies, ST, All Star default.
Sowers gives up not one.....not two.....but three HRs. Along with 7 hits.

Final score:
Rockies 11, Tribe 1.

Hits: Rockies: 17. Tribe: 2.

OMG, I got shelled bigtime.
The one run was a measly RBI single.
Rockies had a total of 5 HR's.......
I really stunk today; thank goodness it's only ST.....

[DarthRambo]

re: Too many two out inning rallies by AI!! (Comeback AI Discussion)

Game 4 of testing... season m.ode this time!

CLE 10 0 0 0 0 0 0 0 0 10 13 1
TEX 0 0 0 0 0 2 0 0 0 2 7 0


Gave tribe 10-0 lead, was able to hold easily, as well as cliff lee pitching a CG!

Is there a comeback code, obviously not. if so, it would happen every game right? imaybe the lead has to be no more than 9 or 8, who knows. but for me, this gives me assurance there's nothing in the game. its just a really hard game.

I'm done testing for now, may do more later with other pitcher, but i doubt it.

[SoxFan01605]

re: Too many two out inning rallies by AI!! (Comeback AI Discussion)

Game 4 of testing... season m.ode this time!

CLE 10 0 0 0 0 0 0 0 0 10 13 1
TEX 0 0 0 0 0 2 0 0 0 2 7 0


Gave tribe 10-0 lead, was able to hold easily, as well as cliff lee pitching a CG!

Is there a comeback code, obviously not. if so, it would happen every game right? imaybe the lead has to be no more than 9 or 8, who knows. but for me, this gives me assurance there's nothing in the game. its just a really hard game.

I'm done testing for now, may do more later with other pitcher, but i doubt it.

lol...43 pages of argument to get here, but at least you finally did

Jokes aside, glad you got it figured out (or at least resolved for you). I imagine it's frustrating to play a game when you feel handicapped.

Hopefully now you can get more enjoyment out of things.

[DarthRambo]

re: Too many two out inning rallies by AI!! (Comeback AI Discussion)

ya, i had to do it. it became a personal determination to figure this out. Game is just literally the hardest sports game, most realistic sports game ever!

[phillyfan23]

re: Too many two out inning rallies by AI!! (Comeback AI Discussion)

hey trinity....

good that u finally figured it out.

last game I played :

7-3 win using Hamels against the nats scrub pitcher and i dominated them like i should.

No late inning babe ruth appearance by the cpu, no counter-punching runs by the CPU.

felt real.....

8 hits I had....4 HRs 2 doubles 4 walks...

[swaldo]

Comeback AI: A scientific analysis

Here is the data from the tests I have been working on:

The numbers below are from 15 CPU vs CPU games on Hall of Fame level. Each team in both the NL and AL played a game. Default everything but I did turn ump variation off so all balls and strikes are called perfectly. Why CPU vs CPU? By testing in this matter nobody can accuse me of playing the CPU poorly.

I then compared this data to 15 real MLB games played on June 1, 2008. Why June 1? I figure both batters and hitters are in good form by then. Also, all NL and AL teams played a game on this date as well. My goal is to get up to 50 games tested and see how close the data merges together.

I realize this sample is low. However it took over 15 hours for my PS3 to churn out these results (they are not quick simmed games.) I first selected human vs CPU then when the game started I switched my controller to CPU. In regards to pitchers I chose a mix of good, fair and poor to replicate real life. I think it's best to start posting this data now because it's a hot topic in the forum. From here on I will post results weekly.

I tried to copy/paste the box scores here but since it's formatted in tables I found it better to just attach the word doc. Page 1 is the 15 games played in The Show and page 2 consists of 15 real MLB games played on June 1, 2008. Note: any numbers you see on the far right are extra inning results.

Before I get to the stats I want to make clear of what the definition of a comeback is. A comeback is when a team that is losing comes back to tie or take the lead at any given time. For example, here is a fictitional box score.

101005000
110040000

In the above example there were two 1 run comebacks and one 4 run comeback.

So far in my 15 game 'The Show' test here is the breakdown:

1 run CB: 14
2 run CB: 5
3 run CB: 3
TOTAL COMEBACKS: 22

CB's in last 3 innings: 6
CB's in 9th inning: 2

Total runs: 151 (5.03 AVG per team)
Total hits: 297 (9.9 AVG per team)
Total Errors: 11 (0.37 AVG per team)

Total Runs - first 3 innings: 42
Total Runs - middle innings: 51
Total Runs - last 3 innings: 51
Extra Inning runs - 7

REAL MLB GAMES ON JUNE 1, 2008 BREAKDOWN:

1 run CB: 11
2 run CB: 4
3 run CB: 2
4 run CB: 1
TOTAL COMEBACKS: 18

CB's in last 3 innings: 3
CB's in 9th inning: 2

Total runs: 142 (4.73 AVG per team)
Total hits: 265 (8.83 AVG per team)
Total Errors: 15 (0.5 AVG per team)

Total Runs - first 3 innings: 59
Total Runs - middle innings: 44
Total Runs - last 3 innings: 33
Extra Inning runs - 6

Run distribution (How many times a certain amount of runs was scored in an inning (The Show : Real MLB):

1 run - 38 : 43
2 runs - 31 : 18
3 runs - 9 : 7
4 runs - 2 : 3
5 runs - 2 : 3
6 runs - 0 : 1
7 runs - 0 : 0
8 runs - 0 : 0
9 runs - 0 : 1

To conclude this test there are more comebacks in The Show thus far, but it's not outrageous. There is a bit more offense in The Show so this could have something to do with it. The data that strikes me the most is the total runs broken down by thirds (first 3 innings, middle innings & last 3 innings.) These numbers are:

The Show: 42,51,51
Real MLB: 59,44,33

The real MLB allowed runs early and then it tapered off. The Show is the complete opposite - starting off well and allowing more runs as the game went on. Perhaps the starting pitcher and reliever ratings need to be looked at? It's too early to tell for sure but there seems to be something going on with that. I wonder what would happen if you started your closer and closed with your starter? Crazy I know.

Other than that the numbers look very realistic. The MLB games had more big innings but the sample size is too low to be concerned with that just yet. It was fun watching the games as I strolled by the living room here and there. You gotta love it when Moyer has his circle change working!

Here is the word doc of the box scores - if I have time I'll post screen shots of all the games played on the PS3. I have files of them all because I also want to look at doubles, triples, walks, strike % etc. but that's a subject for another day...

Show09-Test.doc

[Speedy]

Re: Comeback AI: A scientific analysis

Another thread...

[KOL]

Re: Comeback AI: A scientific analysis

Very nice work,
Must of taken a hell of a lot of time, I'm sure everyone will appreciate it.
So that all but proves it, MLB 09 is only a fraction out.
The devs did a great job. Hopefully this will end all the arguments......
and the last thread.

[DallasRocks]

Re: Comeback AI: A scientific analysis

Very nice work I guess we just have to get out to a big lead in the first few innings or else we're screwed

[jim416]

Re: Comeback AI: A scientific analysis

Very nice work I guess we just have to get out to a big lead in the first few innings or else we're screwed

Sigh, not so. PM me if you need some help.

[DallasRocks]

Re: Comeback AI: A scientific analysis

Sigh, not so. PM me if you need some help.

I'm not one of those saying there is a comeback code but I have had my share of AI comebacks, but most of the time I'm a good enough hitter to come back

[Knight165]

Re: Comeback AI: A scientific analysis

I think we might see even closer games to real life games if we raised the pitching clutch a bit.
It would cut down on the hits...and a little less unraveling in the late innings for runs.

M.K.
Knight165

[DallasRocks]

Re: Comeback AI: A scientific analysis

I think we might see even closer games to real life games if we raised the pitching clutch a bit.
It would cut down on the hits...and a little less unraveling in the late innings for runs.

M.K.
Knight165

hmm that sounds like a great idea, I would probably only edit my own team though

[Matt Diesel]

Re: Comeback AI: A scientific analysis

Props for using real data to back up your stance. I knew there were a lot of comebacks IRL.

You could do it again raising the pitching clutch +1 or +2. I bet that would have a significant effect.

[Speedy]

Re: Comeback AI: A scientific analysis

I think we might see even closer games to real life games if we raised the pitching clutch a bit.
It would cut down on the hits...and a little less unraveling in the late innings for runs.

M.K.
Knight165

That's a point I made in the other thread which seemed to be glanced over. I believe the clutch rating might be skewed to favor the batter more which makes users think there's a comeback code.

On a side note, if I was a dev I'd be thrilled people are upset they can't win the game b/c of stupidity, lack of focus or whatever by the user. For a video game, it's much more fun (at least to me) to really work in winning a game than know it's a given you can win and have to work to lose just to make it more competitive/realistic.