Stats-Based CPU Sliders [MLB 17 Version]

[nomo17k]

Re: Stats-Based CPU Sliders [MLB 17 Version]

Right now I switched to dynamic difficulty (with these sliders) and playing many games to try and level it out...

Which of the current slider settings are you still tuning or looking at? It seems like they are settling in where some of them have remained the same for some iterations. I did notice Wind is back up to 4 and Control is down to 1.

Cliff notes: should I start running new season with these?

I'm mostly just trying to level out batting average/BABIP and HR frequency. Right now it's with just Pitcher Control and Solid Hit. If HRs come down a bit then the 5/12 set should be good enough. I might lower Wind again to 3 or lower if controlling HR frequency requires it.

Most other stats have been pretty stable for a long time and close to the MLB averages to the extent the game allows (some things like O-Swing %/aka chase %, it's too off from the MLB average which I'm not trying to correct), so unless that I decide another iteration with a major change is required (e.g., changing Contact slider), then the 5/12 set should play reasonably good baseball.

[doctorhay53]

Re: Stats-Based CPU Sliders [MLB 17 Version]

I've really enjoyed the several games I've been able to watch today with the 5/12 set.

I haven't played much since early in '15, but I must say I was utterly shocked by how amazing the game looks in CPU vs CPU. The commentary is pretty good, the presentation is just amazing (highlights of a homerun from the 1st inning when a guy bats later in the game? WHAT?), the graphics are solid and only a few of the animations have any weirdness. I'm pleasantly shocked by how realistic this is. A few stats being off a little bit is pretty much expected in any game, and I appreciate how much work you're doing to get everything as close as realistically possible. I'll certainly be spending a lot of time with my PS4 plugging away in this little fake universe of mine.

[nomo17k]

Re: Stats-Based CPU Sliders [MLB 17 Version]

I've really enjoyed the several games I've been able to watch today with the 5/12 set.

I haven't played much since early in '15, but I must say I was utterly shocked by how amazing the game looks in CPU vs CPU. The commentary is pretty good, the presentation is just amazing (highlights of a homerun from the 1st inning when a guy bats later in the game? WHAT?), the graphics are solid and only a few of the animations have any weirdness. I'm pleasantly shocked by how realistic this is. A few stats being off a little bit is pretty much expected in any game, and I appreciate how much work you're doing to get everything as close as realistically possible. I'll certainly be spending a lot of time with my PS4 plugging away in this little fake universe of mine.

Yeah, the only reason why I even bother to look at the stats and how the game plays to this details is because how well crafted this game really is.

You don't normally expect a console game, which often only aspires to be an arcade game, to get so much detail right... I'm still very impressed how close the vast majority of stats come to the real-life data. (If people aren't, they should look at those numbers in the spreadsheets...) Most non-text sports game of this nature don't even come close to this.

I often wonder if even the game designers realize how solid the game plays even as an authentic simulation of the sport. Not saying yielding realistic stats is more important than the process that leads to the results, and there of course are things that are still quirky in how the game plays in detail... but if the sport simulation cannot even produce realistic stats, the basic gameplay has no way of being realistic.

[doctorhay53]

Re: Stats-Based CPU Sliders [MLB 17 Version]

Yeah, the only reason why I even bother to look at the stats and how the game plays to this details is because how well crafted this game really is.

You don't normally expect a console game, which often only aspires to be an arcade game, to get so much detail right... I'm still very impressed how close the vast majority of stats come to the real-life data. (If people aren't, they should look at those numbers in the spreadsheets...) Most non-text sports game of this nature don't even come close to this.

I often wonder if even the game designers realize how solid the game plays even as an authentic simulation of the sport. Not saying yielding realistic stats is more important than the process that leads to the results, and there of course are things that are still quirky in how the game plays in detail... but if the sport simulation cannot even produce realistic stats, the basic gameplay has no way of being realistic.

Yeah, over the years I've went on many crusades trying to match stats to real-life in all kinds of video games; NCAA football, Madden, College Hoops 2k8, NBA 2kX. Lot's of spreadsheets, simming, running cpu vs cpu games, waking up in the middle of the night and going ahead and starting a new game because why not?

Looking at the spreadsheet, even for just a single day which in baseball is a tiny sample size, I'm pretty shocked by how close nearly everything is. I never saw anything like this in the other major sports 'sim' games.

[mellis302]

Re: Stats-Based CPU Sliders [MLB 17 Version]

Nomo, I've been following this thread all year adjusting up and down with your sliders. BABIP has been the biggest outlier all year in my franchise as I'm sitting second to last in MLB in team BA and first in hits allowed(I sim the rest of the league games) through about 100 games now.

I really liked what I saw in hit diversity with solid hits at 7 so instead of shift that down to decrease the inflated HR% in your 5/9 set I adjusted Power down to 4, Control down to 2 and wind back up to 5 to weaken the amount of moonshots but still get a realistic amount of HR's. The eye test on this looks great. I wish I was collecting as much data as you to show the difference but just wanted to let you know what I was running now because that 5/9 set was close except for BB% and HR%.

I'd be interested to see you run a set on this or your thoughts on to why it wouldn't work.

[doctorhay53]

Re: Stats-Based CPU Sliders [MLB 17 Version]

My general sense in ~10 franchise games is that hits and scoring are low in what I'm seeing.

Are you testing in the play now games, and do the play now games include the matchup advantage stuff that's in diamond dynasty?

[nomo17k]

Re: Stats-Based CPU Sliders [MLB 17 Version]

Yeah, over the years I've went on many crusades trying to match stats to real-life in all kinds of video games; NCAA football, Madden, College Hoops 2k8, NBA 2kX. Lot's of spreadsheets, simming, running cpu vs cpu games, waking up in the middle of the night and going ahead and starting a new game because why not?

Looking at the spreadsheet, even for just a single day which in baseball is a tiny sample size, I'm pretty shocked by how close nearly everything is. I never saw anything like this in the other major sports 'sim' games.

Ultimately someone has to do the job of translating real-life sport data into code for the game, and I think the combination of people who can properly interpret data and turn into code and willingness to execute on the vision of authenticity is difficult to come by.

Too much "this is just a game" attitude can thwart the push toward authenticity, but if it doesn't play well as a game then the project won't be profitable to sustain the effort itself... so I think it's on a kind of very fine balance for a consumer-oriented game to be on.

With the recent push toward more accessibility, not sure which direction The Show is going from here on though. I really hope things won't get mucked up too much in an effort to play it like a game instead of a baseball simulation.

[nomo17k]

Re: Stats-Based CPU Sliders [MLB 17 Version]

My general sense in ~10 franchise games is that hits and scoring are low in what I'm seeing.

Are you testing in the play now games, and do the play now games include the matchup advantage stuff that's in diamond dynasty?

I don't think the Inside Edge type of match-up matters at all outside DD.

10 games is a way too small sample, especially if you aren't rotating teams/ball parks.

[doctorhay53]

Re: Stats-Based CPU Sliders [MLB 17 Version]

I don't think the Inside Edge type of match-up matters at all outside DD.

10 games is a way too small sample, especially if you aren't rotating teams/ball parks.

Yeah, it's been spread over different teams, with 2 different franchises, including one that had a fantasy draft. So it's been pretty spread around. I was just adding an anecdote to the previous poster who said they were getting low numbers of hits, which you can weight however you like Even 0 if you wish.

I was going to do a full update after ~2 in-game months with a stable franchise. At that point, things like 1st in ERA last in runs scored are indicative of glaring issues, especially in a fantasy draft where most teams are pretty even. But it's going to take me quite a while to get there.

[nomo17k]

Re: Stats-Based CPU Sliders [MLB 17 Version]

Nomo, I've been following this thread all year adjusting up and down with your sliders. BABIP has been the biggest outlier all year in my franchise as I'm sitting second to last in MLB in team BA and first in hits allowed(I sim the rest of the league games) through about 100 games now.

I really liked what I saw in hit diversity with solid hits at 7 so instead of shift that down to decrease the inflated HR% in your 5/9 set I adjusted Power down to 4, Control down to 2 and wind back up to 5 to weaken the amount of moonshots but still get a realistic amount of HR's. The eye test on this looks great. I wish I was collecting as much data as you to show the difference but just wanted to let you know what I was running now because that 5/9 set was close except for BB% and HR%.

I'd be interested to see you run a set on this or your thoughts on to why it wouldn't work.

In many ways, there really ins't one combination of sliders that is fully "correct," but there are a few different combinations that can produce similar results (at least as we can see superficially in data like stats), so if the results are within realism, I'd say go for it you know what I mean.

Purely from stats seen, I do like 5/8 and 5/9 sets which are mostly the same sliders, except that they were done pre and post v1.06 patch. A substantial change was only in HR production, so using just Power slider might make sense... but just one click of that slider can affect overall offense (compare 4/25 and 4/26 set). So it's not a simple change regardless I think.

But the major reason why I'm exploring options with Solid Hit below 7 is that with the 5/9 set, I think the potential for high-scoring games where pitchers kind of meltdown is substantially higher, due to the combination of low Pitcher Control and high Solid Hit.

Very roughly speaking I want to keep double-digit scoring games to around 5 games for a 75-game set, as that's roughly how often they happen in MLB today. The 5/9 set had 4 in just 30 games, so while the majority of games may play well okay, I think the setting may be making the game vulnerable to the balance becoming "off" from where it plays more stably...

If I can (by such measure as keeping sliders closer to the default), I want to avoid being too close to the setting that exposes quirkiness of the game balance that may be suppressed if we play on a more "normal" setting. And for some reason I have found Solid Hit to act weird at times... that's why I'm investing a little more time exploring Solid Hit 6.

It's kinda in a difficult phase because I've been pursuing a rather small change using sliders where one click introduces changes big enough to over/undershoot things. At some point I probably have to make some trade-offs. The flurry of patches coming in without being totally clear on what has been touched also make it difficult to set an expectation.

[nomo17k]

Re: Stats-Based CPU Sliders [MLB 17 Version]

...

But the major reason why I'm exploring options with Solid Hit below 7 is that with the 5/9 set, I think the potential for high-scoring games where pitchers kind of meltdown is substantially higher, due to the combination of low Pitcher Control and high Solid Hit.

...

Responding to myself I suppose, but one way a blowout can happen in this year's game is that when a starter leaves early and the run differential is big, the CPU manager has a tendency to leave in a long reliever to pitch for an extended time, regardless of how he performs. They usually are not that good as pitchers, so they may also get knocked around, losing confidence, and exhausting stamina... so I think in some cases I should consider this possibility and attribute pitching meltdown to subpar bullpen management by CPU.

This is from watching such a game... CPU is leaving in a middle reliever even though his confidence/energy is totally zeroed out. And the Padres is knocking him left and right... :P


Another thing about increasing Solid Hit is that the new ball physics I think introduced a kind of (unrealistic) line drives that have long hang time, and some extra HRs might be coming from them.

I don't remember increasing Solid Hit just by one changes the HR aspect of game so much in the past (if anything, line drives are less likely to become HRs than fly balls are...)... the slider might be a little difficult to tame.

[nomo17k]

Re: Stats-Based CPU Sliders [MLB 17 Version]

In addition to collecting stats, I'm doing quite a bit of eye tests recently, and out of all sets I think I like the 5/14v2 set the best so far...

The offense feels neither anemic nor poppy, and overall It plays tight.

There are a few blowouts, perhaps as frequent as I've seen with Pitcher Control set lower, but they don't feel hopeless, as in those moments when CPU just hit hard anything throwing at it (something many would interpret as scripting or artificial boost for CPU). Some games are clearly the cases when CPU manager leaves zero energy reliever for too long, so I shouldn't weigh them too much anyways.

Looks like there is a "beta" set now... the 5/14v2 set looks pretty good so far. Hopefully the tide doesn't dramatically change as I add more games.

[Ebpmd]

Re: Stats-Based CPU Sliders [MLB 17 Version]

Is beta set posted anywhere or in vault?

[rjackson]

Re: Stats-Based CPU Sliders [MLB 17 Version]

Is beta set posted anywhere or in vault?

Click the link in nomo's signature for '17

[My993C2]

Re: Stats-Based CPU Sliders [MLB 17 Version]

the 5/14v2 set looks pretty good so far. Hopefully the tide doesn't dramatically change as I add more games.

I have yet to play a MoM game with the 5/14v2 set (I'll be playing one today once I get home from work), but yesterday I got about a dozen QC Fast Play CPU vs CPU games in using the 5/14v2 set and I liked the results I have seen so far.