Stats-Based CPU Sliders [MLB 17 Version]

[nomo17k]

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

I might update the OP with a "currently recommended set" or so, but I might want to check a little more. I'm never inclined to "release" a set quick so you should look at the spreadsheets and go from there for your own. The 5/14v2 set I think is okay for a stopgap set in any case.

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.

Overall I think it does play well, and I don't think I'm pushing the game to the edge of breaking down, as I might have been closer to do with other low Pitcher Control/high Solid Hit sets.



Removing high-scoring games where a middle relievers were likely left in with no confidence and zero energy for a bit long (e.g., pitched more than 2 innings poorly after the starter left), the blowout frequency is pretty good. I wanna see the right amount of high-scoring games, but not because CPU totally throws away a game leaving no energy reliever too long.

I'm now wondering if there is a way to make poor bullpen management less of an issue by giving relievers a little more stamina, for example. I think it's probably necessary to do this for optimal pitcher substitution regardless, but I generally do not like babysitting CPU and at least I want to find a compromise for "low maintenance" setting where I don't have to change settings in the middle of a game, etc.

[nomo17k]

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

Will try Reliever Stamina at 10 for a bit with 5/16. The extreme setting is to see the effect, if any, to understand it better. What I specifically want is decreasing the chance of middle-relief meltdown by allowing them to go a bit far without exhausting energy.

Otherwise the 5/16 set is identical to 5/14v2, so I expect them to play similarly for the most part.

[orye74]

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

Seen a ridiculous amount of strikeouts with this new slider set. I'll try it one more game then I'm reverting back

[nomo17k]

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

Seen a ridiculous amount of strikeouts with this new slider set. I'll try it one more game then I'm reverting back

I don't see anything particularly wrong with stats that lead to Ks (e.g., swing & miss %), so over a longer run it should stabilize to around 8 Ks per game per team.

I don't know what's "high" here, but in recent years we are seeing Ks increase at a record pace... when I was a kid a pitcher who had K/9 of 8 would be called a decent strikeout pitcher, but that's just an average these days...

[My993C2]

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

I don't see anything particularly wrong with stats that lead to Ks

I doubt there is a problem with Ks on your 5/14v2 set. I didn't see any abnormal number of Ks in my MoM game at Target Field yesterday. But I did see a lot of Big Fly HRs. However it was only one game and I do play with the Rockies who have a powerful offense but also not the best pitching staff who will give up runs themselves even away from Coors Field. Plus in all the CPU vs CPU games that were played after my MoM game, there was a mixture of low scoring and high scoring games. I did not pay attention to the Ks in those games, but I will have a glance at the post game pitcher stats after this evenings games to see if Ks are up. But as I said, I doubt there is a problem.

[nomo17k]

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

I doubt there is a problem with Ks on your 5/14v2 set. I didn't see any abnormal number of Ks in my MoM game at Target Field yesterday. But I did see a lot of Big Fly HRs.

...

I think there is still some tendency for the game to produce HRs easily, because the game in general tend toward producing more fly balls than other hit types of lower launch angle, and with the new ball physics I think their hang time is still longer than we see in real-life... this allows hitters with only moderate Power rating to go long distance when they square up. There are also some line drive shots that have some elevation that can go over the fence.

I'm trying to keep them in check over a long run, but occasionally there will be some spurts of HRs especially when pitchers hang pitches around the middle of the strike zone (which often happen with low confidence/energy relievers). When the pitcher's energy goes low, the batter PCI increases so all those lead to high offense.

[My993C2]

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

I'm trying to keep them in check over a long run, but occasionally there will be some spurts of HRs especially when pitchers hang pitches around the middle of the strike zone (which often happen with low confidence/energy relievers). When the pitcher's energy goes low, the batter PCI increases so all those lead to high offense.

I failed to mention that the majority of the HRs I had in my MoM game last night happened in the later innings of the game. There were a few HRs early on in the game (for both teams), but for the most part both starters did reasonably well as the game was tied 3-3 after 6 innings. It was only in the later innings when the relievers came into the game when things got a little out of control, especially for the Twins who were only down by 1 run in a 7-6 game going into the top of the 9th, but then surrendered 5 more runs after that to make what appeared to be a close game, a blow out. Likely not a slider issue.

[nomo17k]

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

I failed to mention that the majority of the HRs I had in my MoM game last night happened in the later innings of the game. There were a few HRs early on in the game (for both teams), but for the most part both starters did reasonably well as the game was tied 3-3 after 6 innings. It was only in the later innings when the relievers came into the game when things got a little out of control, especially for the Twins who were only down by 1 run in a 7-6 game going into the top of the 9th, but then surrendered 5 more runs after that to make what appeared to be a close game, a blow out. Likely not a slider issue.

I try to find the right compromise, but I do note that lowering Pitcher Control and increasing Solid Hit (by a small mount) both would make HRs a little easier to come by. In fact, if it doesn't affect batting average too much, I might want to increase Pitcher Control by one, to tighten things a bit. But I think it probably will affect batting average a little, so even if I improve pitchers' command, I'm more likely to increase Pitcher Consistency now...

[doctorhay53]

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

Just a little feedback, after the first month in my fantasy draft franchise, half with the 5/12 set and half with the 5/14v2 set, my team is top 5 in all major pitching stats and bottom 5 in all major hitting stats. I don't think I drafted quite that skewed. We're number 1 in ERA, SO, BB, Hits, and HR on the pitching side.

I'll keep an eye on it, but I must say that watching the games, it just seems like there's never a big hit with runners on. They've actually been pretty boring games to watch.

Like I said, in alot of years of slider testing, these things make the antennae stand up. But keep doing what you're doing and I'll keep going with what you have.

[nomo17k]

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

Just a little feedback, after the first month in my fantasy draft franchise, half with the 5/12 set and half with the 5/14v2 set, my team is top 5 in all major pitching stats and bottom 5 in all major hitting stats. I don't think I drafted quite that skewed. We're number 1 in ERA, SO, BB, Hits, and HR on the pitching side.

I'll keep an eye on it, but I must say that watching the games, it just seems like there's never a big hit with runners on. They've actually been pretty boring games to watch.

Like I said, in alot of years of slider testing, these things make the antennae stand up. But keep doing what you're doing and I'll keep going with what you have.

I don't think there is specifically an issue with "big hits," as the game really doesn't pre-determine outcomes and the best we can do with slider adjustments is to influence outcomes by changing frequency of certain type of events happening here and there... so if you feel hits are not coming as frequently as you feel it should, that probably means the batting average is low.

I don't think it's unrealistically low, but even with all the adjustments to bring up batting average, it is probably still on a lower side... perhaps it's still be around .250 if things continue as they have so far. I would like to bring that up by .005, but sliders aren't as granular (i.e., one click change is too big), so an adjustment this fine is a little difficult to make.

But if so desire, I believe there is a better combination where things sit a little more on an offensive side than the current set allows.

Also, comparing stats from games that you actually play out and the games that you sim... they actually run on different game engine (so-called gameplay vs simulation engines), so there will be some differences due to the fact that they are actually not driven by the same engine...

I haven't tested sim engine this year yet, so I don't know what the league averages are when purely using the sim engine... it's usually pretty close to the real-life MLB averages of recent years.

[doctorhay53]

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

I don't think there is specifically an issue with "big hits," as the game really doesn't pre-determine outcomes and the best we can do with slider adjustments is to influence outcomes by changing frequency of certain type of events happening here and there... so if you feel hits are not coming as frequently as you feel it should, that probably means the batting average is low.

I don't think it's unrealistically low, but even with all the adjustments to bring up batting average, it is probably still on a lower side... perhaps it's still be around .250 if things continue as they have so far. I would like to bring that up by .005, but sliders aren't as granular (i.e., one click change is too big), so an adjustment this fine is a little difficult to make.

But if so desire, I believe there is a better combination where things sit a little more on an offensive side than the current set allows.

Also, comparing stats from games that you actually play out and the games that you sim... they actually run on different game engine (so-called gameplay vs simulation engines), so there will be some differences due to the fact that they are actually not driven by the same engine...

I haven't tested sim engine this year yet, so I don't know what the league averages are when purely using the sim engine... it's usually pretty close to the real-life MLB averages of recent years.

Yeah, good point; the sim engine could be the issue too. In the past I've actually compared the results of played out games to the results of simming. I.e. I used to load the same game and sim it 10 times, then play it out and adjust sliders until results were acceptably similar (and, if possible, somewhat realistic). I would track correlations across runs (i.e. even though a results was within 1-sigma each time, if it was always near 1-sigma high it probably still meant there was a problem).

I guess the point is that it's also important to me that the sliders fit somewhat within the sim engine, so that stats aren't too crazy.

I should have clarified the 'big hits' comment; I know the sliders can't really control that, what I was suggesting is that perhaps there's still been a bit of unluckiness to my results and things will even out. And then again, maybe my results are pretty decent and the sim engine is out of whack. Just takes quite a while to play out games, even on fast presentation, so I won't have more info for a little while.

[My993C2]

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

Another MoM 8-5 game last night full of HRs. Three for me playing with the Rockies and I also surrendered a grand slam HR by the Twins. I am not complaining here. Part of the issue is I am playing with a team with a potent offense but also a team that does not have one of the better pitching staffs. Last night also produced the usual a mix of high scoring and low scoring CPU vs CPU QC games so if anything I could lower the solid hits a bit for my Rockies MoM games and keep the settings as is for the CPU games.

Oh and strike outs aren't an issue with the 5/14v2 set. There was one CPU vs CPU game when one team got 14 Ks while their opponents only got 5. But that can happen in baseball. I'd say on average there was about 8 Ks per team per game which I believe it what you were shooting for. So I am happy with the strike outs as they are.

Oh I did tweak the "BR Steal Ability" down to a 7 from the 10 you have and it produced an interesting side effect. I did this because I felt stealing bases was too easy for my MoM team. It's interesting, I don't have any Billy Hamilton or Trey Turner sort of speed on my team, but I am rated one of the faster teams according to the player ratings SDS assigned the rosters yet the real world team does not steal many bases. So now at a value of 7 I will really only send my fastest guys and the moderate speed guys are susceptible to pick offs. But what was the side effect? Pitches in the dirt. With the slider at 7 (and the rest of your sliders settings at your recommended values), if the ball is in the dirt far away enough from the catcher where the base runner will take off, the fast guys will still be able to take the extra bag. But the moderate or slower speed base runners appear to be susceptible to getting thrown out. Prior to this, I was only ever seeing the base runners successfully taking the extra base whenever the ball got away from the catcher. Now it is sort of leveled out at least where some base runners get picked off.

[nomo17k]

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

Yeah, good point; the sim engine could be the issue too. In the past I've actually compared the results of played out games to the results of simming. I.e. I used to load the same game and sim it 10 times, then play it out and adjust sliders until results were acceptably similar (and, if possible, somewhat realistic). I would track correlations across runs (i.e. even though a results was within 1-sigma each time, if it was always near 1-sigma high it probably still meant there was a problem).

I guess the point is that it's also important to me that the sliders fit somewhat within the sim engine, so that stats aren't too crazy.

I should have clarified the 'big hits' comment; I know the sliders can't really control that, what I was suggesting is that perhaps there's still been a bit of unluckiness to my results and things will even out. And then again, maybe my results are pretty decent and the sim engine is out of whack. Just takes quite a while to play out games, even on fast presentation, so I won't have more info for a little while.

The sim engine is probably fine. If not, people who use it to do player rating and progression and stuff usually make an issue out of it.

And if the sim engine doesn't produce reasonably good results out of box, we cannot really do much (aside from global attributes editing), since most sliders do not affect simmed games.

[nomo17k]

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

Another MoM 8-5 game last night full of HRs. Three for me playing with the Rockies and I also surrendered a grand slam HR by the Twins. I am not complaining here. Part of the issue is I am playing with a team with a potent offense but also a team that does not have one of the better pitching staffs. Last night also produced the usual a mix of high scoring and low scoring CPU vs CPU QC games so if anything I could lower the solid hits a bit for my Rockies MoM games and keep the settings as is for the CPU games.

While that level of HR fest probably won't continue, I think the 5/14v2 and 5/16 sliders are still a little HR-happy than I'd like them to be.

If you are playing in Colorado, then there may also be some park factors in play.

Oh I did tweak the "BR Steal Ability" down to a 7 from the 10 you have and it produced an interesting side effect. I did this because I felt stealing bases was too easy for my MoM team. It's interesting, I don't have any Billy Hamilton or Trey Turner sort of speed on my team, but I am rated one of the faster teams according to the player ratings SDS assigned the rosters yet the real world team does not steal many bases. So now at a value of 7 I will really only send my fastest guys and the moderate speed guys are susceptible to pick offs. But what was the side effect? Pitches in the dirt. With the slider at 7 (and the rest of your sliders settings at your recommended values), if the ball is in the dirt far away enough from the catcher where the base runner will take off, the fast guys will still be able to take the extra bag. But the moderate or slower speed base runners appear to be susceptible to getting thrown out. Prior to this, I was only ever seeing the base runners successfully taking the extra base whenever the ball got away from the catcher. Now it is sort of leveled out at least where some base runners get picked off.

This may actually be a point to check.

While I'm pretty confident that setting BR Steal Ability to similar value (well say within +/-1) as Pitch Speed sliders would yield SB success rate to roughly that of MLB average (72%), CPU managers do like employing hit-and-run as strategy, and mediocre base stealers often get caught because of failed hit-and-runs.

So it is possible by increasing BR Steal Ability, I may be making the best base stealers too successful, while mediocre ones "stealing" only on hit-and-runs continue to get caught, skewing the average that I'm computing (which includes both legit steals and failed hit-and-runs).

[nomo17k]

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

On steal success...

By now I've run most of my games with Pitch Speed 10 and BR Ability 10 (basically only the initial default set is different), so I just took the current top 10 base stealers in terms of SB from each league, and here is the list:

Player SB CS
---------------------------
Cain	12	2
Kiermaier	9	0
Dyson	8	0
Segura	8	2
Burns	8	1
Ellsbury	8	3
Davis	8	2
Revere	8	2
Brantley	6	0
Gomez	6	2
Andrus	6	2
Upton Jr	6	0
Naquin	6	0
Gordon	18	0
Villar	10	1
Blackmon	10	1
Hernandez	10	0
Nunez	10	1
Marte	7	4
Herrera	7	0
Diaz	7	0
Turner	6	1
Inciarte	6	0
Jankowski	5	0
Hazelbaker	5	1
Harrison	5	0
Myers	5	1
Dahl	5	2
Pollock	5	0

So the SB% of this group is 89%... is this higher than typically seen in the best base stealers in MLB?

It does look it's much higher. In 2016, top 20 base stealers from both leagues swiped at SB 79%.

So perhaps I should lower BR Steal Ability to make the best stealers a little more realistically not as good at stealing bases... that might actually lower the overall league SB%, but I think it's a good compromise given that a lot of caught steals may be coming from failed hit-and-run.