Stats-based sliders for CPU

[nomo17k]

If you are looking for the information about the June 3 version of this slider set, please go to this post:

http://www.operationsports.com/forum...post2042505133

If you are looking for the information about the April 16 version of this slider set, please go to this post:

http://www.operationsports.com/forum...post2042313913




###################

Hi everyone. First time poster here. I haven't been much of a gamer lately but recently got MLB 10 thinking it'd be perfect just to kill time when it was too cold out this winter, but oh boy I never thought a baseball video game has gotten this far and I've been hooked.

I prefer realism over anything, so that has led me to tweak things using sliders, thanks to the wealth of info in this forum. I like playing/watching obscure strikeout artists like Sid Fernandez and Hideo Nomo (remember them??), so it's important for me to get pitcher-batter matchup realistic. It's actually difficult for these unusual players, as MLB 10 CPU hitters are a bit too well disciplined at the plate and don't chase enough pitches out of the zone.

Thinking ahead as I'm definitely going to get a copy of MLB 11, I'm coming up with a strategy for my own slider making, hoping to receive feedback and share my own progress at the same time here in this thread.

My initial focus will be to get CPU vs CPU matchup right at All-Star level. After all humans play with wildly different abilities (I already pitch decent games at any levels but can merely hit .240 as a Rookie) and stats can get skewed that way.

So my initial strategy would be: (1) Play a bunch of CPU vs CPU games (> 30 hopefully), first with the default slider set and then (2) Rack up game by game stats and compare to the following 2007 – 2010 per game, per team stats (compiled from Baseball-Reference or FanGraphs):



These standard stats are obvious, but there are some other numbers we can easily get at the end of each game from the pitching chart:



For balls hit in play, percentages for flyball (FB), line-drive (LD), and groundballs (GB), % of pitches swung at (Swing%), and % missed among those swung at.

(3) I obviously just run these games in the background and not watch every game, but at the end of the game I can look at the replays and how the game physics looks like, taking notes of any glaring deviations from realism (are balls hit too hard, are all outfielders running like Carl Lewis, etc.), to find which sliders may need adjustments.

Using a couple sliders from this forum I've played around some with MLB 10, and it's quite cool it already plays close to those real-life numbers. People at SCEA obviously are using these real-life numbers to tune the game, which is fantastic.

Anways, now I have to wait another month to start this thing, don't I. If I'm still alive in a month I'll report my findings. Please wish me luck then.

[Heroesandvillains]

Re: Stats-based sliders for CPU

If you're still alive???

Do you have a dangerous job or something?

[nomo17k]

Re: Stats-based sliders for CPU

If you're still alive???

Do you have a dangerous job or something?

Yeah, I climb Mt Everest every other week... No, I have as dull a job as sitting in front of a computer all day long. It's just that I haven't excluded the possibility of that happening. After all a quadruple play happens in baseball does it.

I cannot wait till the release day!

[nomo17k]

Re: Stats-based sliders for CPU

I'm going to do 75 games until I decide which sliders to adjust (to cover the pitching by the bottom of rotation), but the trend probably won't change much so here is the intermediate results, through 43 games played with the default sliders (I maxed the pitch speed to accelerate the game tiny bit), the SCEA roster downloaded on the release date. Allstar all the say, but that should not matter:



Offensive outburst in the last few games really inflated the stats; just an example you cannot make sound judgement with only 10 or so games.

Offense numbers are consistently up. Not nearly enough CS. Very low BB. High WP. All consistent with what others have been saying so far.

[DaiYoung]

Re: Stats-based sliders for CPU

Interesting stuff Hideo. Indeed very similar to what I found. The good news about the CPU offense is that lowering the Power and Solid Hit sliders will have a dramatic effect in quelling the power surge. I'd recommend taking them down one notch at a time.

Walks are more of a problem since reducing the offense seems also to reduce the amount of BB. My theory is that as pitchers give up less hits they become more confident and throw more strikes. Maybe there is a solution in the pitcher consistency slider, I'll be testing that slider out in the next couple of days.

[Heroesandvillains]

Re: Stats-based sliders for CPU

I'm going to do 75 games until I decide which sliders to adjust (to cover the pitching by the bottom of rotation), but the trend probably won't change much so here is the intermediate results, through 43 games played with the default sliders (I maxed the pitch speed to accelerate the game tiny bit), the SCEA roster downloaded on the release date. Allstar all the say, but that should not matter:



Offensive outburst in the last few games really inflated the stats; just an example you cannot make sound judgement with only 10 or so games.

Offense numbers are consistently up. Not nearly enough CS. Very low BB. High WP. All consistent with what others have been saying so far.

The lack of CS is alarming. Have any one you tested the effects of the stolen base success slider? It's hard to gauge attempts, though, over such a small sample size. Lowering it should also minimally bring runs per game down. That would also be a good thing.

[DaiYoung]

Re: Stats-based sliders for CPU

Any updates Nomo? I'm very curious to see what you're trying out.

[nomo17k]

Re: Stats-based sliders for CPU

I'm almost finishing with my 75-game run with default. Slow I know haha but will be posting the spreadsheet for those who can make use of it (call it a gift from nerd). It will be a good baseline stats for reference.

Then, I'll re-read the CPU vs CPU thread thoroughly to see where people are at now, and decide which combination of sliders to test out for my next set. It's good to have stats data points with slightly different slider combo, so I'm likely to deliberately set the sliders slightly off from suggested in the thread. We already have a good baseline set in the CPU vs CPU thread, so that makes it easier for me to think where I might want to deviate.

I'll be hanging around in that thread for a while!!

[nomo17k]

Re: Stats-based sliders for CPU

Through 75 games with default sliders:



Only started keeping track of fielding and throwing errors (FE & TE) for the last 19 games. I didn't distinguish IBB and regular BB (not in boxscore).

These numbers are mostly still consistent with the CPU vs CPU thread findings and haven't changed much from my last post.

On the stats few people are tracking. Compared to real-life MLB numbers:

* slightly more foul balls.
* slightly more groundballs than flyballs & linedrives.
* CPU swings at pitches slightly more aggressively and miss less.

I don't think there are obvious ways to change the latter two with sliders.

[Heroesandvillains]

Re: Stats-based sliders for CPU

Lowering contact will lower hits/per 9 (good), make the CPU miss on swings more/foul less (good), and effect the hitters vision/discipline, causing hitters to possibly be less aggressive (good). I'm not positive on the last one, but I think:

You should lower contact. It should help. Don't you think?

[PsychoBulk]

Re: Stats-based sliders for CPU

Great work and findings.

Very helpful to those of us with more than a passing interest in sliders...

[DaiYoung]

Re: Stats-based sliders for CPU

Through 75 games with default sliders:



Only started keeping track of fielding and throwing errors (FE & TE) for the last 19 games. I didn't distinguish IBB and regular BB (not in boxscore).

These numbers are mostly still consistent with the CPU vs CPU thread findings and haven't changed much from my last post.

On the stats few people are tracking. Compared to real-life MLB numbers:

* slightly more foul balls.
* slightly more groundballs than flyballs & linedrives.
* CPU swings at pitches slightly more aggressively and miss less.

I don't think there are obvious ways to change the latter two with sliders.

Thanks for the spreadsheet Nomo, very comprehensive.

Just eyeballing it you can see the effect of the wind. Almost every time the wind was 15mph or more blowing out, the game became a dinger fest. From the research into the effect of wind on home run rates that I have read, it just doesn't have that much effect. Additionally, winds of that speed are much less common in real life games.

[DaiYoung]

Re: Stats-based sliders for CPU

Lowering contact will lower hits/per 9 (good), make the CPU miss on swings more/foul less (good), and effect the hitters vision/discipline, causing hitters to possibly be less aggressive (good).

You should lower contact.

The problem with lowering contact is that it reduces plate vision/discipline and therefore results in even fewer walks. At least that was what I found when I lowered it last year.

[nomo17k]

Re: Stats-based sliders for CPU

Just eyeballing it you can see the effect of the wind. Almost every time the wind was 15mph or more blowing out, the game became a dinger fest. From the research into the effect of wind on home run rates that I have read, it just doesn't have that much effect. Additionally, winds of that speed are much less common in real life games.

Wow, you are using the weather info! Good that I kept track. Wasn't sure I was gonna use it myself.

Yeah, looking at gameplay, I feel the wind effect is quite exaggerated. Wasn't sure it had any effect on HRs, but nonscientific observation was that I saw quite a few power fest in Chicago ballparks. I don't know if I got the info from the same source, but I also remember the wind effect isn't so great on HRs, but they detect slight but significant increase in the wind effect in Chicago ballparks... Should've kept the web page for reference.

[DaiYoung]

Re: Stats-based sliders for CPU

Wow, you are using the weather info! Good that I kept track. Wasn't sure I was gonna use it myself.

Yeah, looking at gameplay, I feel the wind effect is quite exaggerated. Wasn't sure it had any effect on HRs, but nonscientific observation was that I saw quite a few power fest in Chicago ballparks. I don't know if I got the info from the same source, but I also remember the wind effect isn't so great on HRs, but they detect slight but significant increase in the wind effect in Chicago ballparks... Should've kept the web page for reference.

Yes, that was the article I read. Was in the Hardball Times. Actually I posted a link to it on these boards last week.

Et voila:

http://www.hardballtimes.com/main/article/when-the-wind-blows/