Stats-Based CPU Sliders [MLB 14 Version]

[nomo17k]

Re: Stats-Based CPU Sliders [MLB 14 Version]

Nomo, for stat heading L, I believe it says HR/PA (I don't know 100% because it's cut off a bit but if that's the case, it's showing that's it's around half the MLB league average. Is there a reason for this and is it fixable with a slider tweak or are you happy with where this is at? This is in your current recommended set...

I think you are referring to the entry HR/FB. That's a fraction (in %) of fly balls that becomes HRs. I've been computing just because it's easy to do.

I wonder why it's consistently low compared to the MLB ave, but one major factor may be that the ways The Show and Fangraphs identify fly balls are very different. For one, I don't put too much faith in how the game distinguish fly ball, line drive, and grounder.

In the end, I'm just computing but not really using it for slider adjustments.

[Threeebs]

Re: Stats-Based CPU Sliders [MLB 14 Version]

I think you are referring to the entry HR/FB. That's a fraction (in %) of fly balls that becomes HRs. I've been computing just because it's easy to do.

I wonder why it's consistently low compared to the MLB ave, but one major factor may be that the ways The Show and Fangraphs identify fly balls are very different. For one, I don't put too much faith in how the game distinguish fly ball, line drive, and grounder.

In the end, I'm just computing but not really using it for slider adjustments.

Well that explains it. You're absolutely right. Playing RTTS for instance I always see quirky calls on what the out actually is. I don't think it's a stat we can really put much weight into, like you said... Really liking the numbers your new slider set is coming up with...

[Aensland]

Re: Stats-Based CPU Sliders [MLB 14 Version]

Hey everyone, long time lurker here; decided to finally register. I've been using your sliders the past couple years and giving it a shot again now. Few things I've noticed:

Using it with quick counts, the number of home runs (especially opposite field shots barely clearing walls) seem to be way inflated, compared to real life and the results posted on this board. To counter that, I bumped the CPU home run power down to 4, and the overall offense took a major nosedive. In 8 games, all but one had a score or either 1-0 or 0-0 by the sixth inning, using starters 1 through 5. Most runs scored was 5, and that only happened once. Granted, 8 games is a small sample, but atm it appears getting this year's game to produce realistic results is more difficult than last year.

[KBLover]

Re: Stats-Based CPU Sliders [MLB 14 Version]

I'm the type that fist-pumps when I draw a walk, as most people would do if they hit a HR.

Me too, and walks are a sore spot with me and this game (that and BABIP by CPU on HoF Classic Pitching...)

Going to keep an eye on this thread as I hope to get MLB 14 soon, and maybe see what applies to MLB13 in the meantime.

I LOVE walks and working the count and trying to run up pitch counts, even if I can't hit the starter well.

[kcsam]

Re: Stats-Based CPU Sliders [MLB 14 Version]

I see you've began testing with QC. That's how I've been playing since using your sliders. So far, it looks like results are a lot different than your previous sets. I know the sample size is still really small, but it still looks pretty extreme. Since I'm on my phone I can't check the sliders from previous iterations. Have you changed them since the last set on 5/7?

I wouldn't think QC would make a difference in how the game would play based off slider alignments, but then again, what do I know. I didn't code the game! I'll be interested in seeing what you find. At least now you'll be able to get results from the cpu VS cpu games quicker.

[kcsam]

Re: Stats-Based CPU Sliders [MLB 14 Version]

This is a placeholder for an upcoming thread talking mostly about how to brew coffee so that it doesn't taste too sour. Another focus will be on what fast food chain cooks the best-tasting french fries. There will be occasional tangent on MLB 14 The Show sliders as well.


The on-going data collection can be followed here:

https://docs.google.com/spreadsheet/...=sharing#gid=2


Meanwhile, here is the "currently recommended" slider set:

This set is based on the collected data through May 7. Everything is at 5 (default) except:

Foul Frequency: 3
Solid Hit: 6
Starter Stamina: 6
Reliever Stamina: 6
Pitcher Control: 3
Pitcher Consistency: 4
Strike Frequency: 4
Manager Hook: 4
Pitch Speed: 0
Fielding Errors: 7
Throwing Errors: 3
Fielder Run Speed: 3
Fielder Reaction: 4
Fielder Arm Strength: 4
BR Steal Ability: 0


Note that Pitch Speed is set to 0 in order to obtain realistic SB%.

Are you using reliever stamina at 6 or 5? On here you say 6, but the spreadsheet for 5/7 you say 5.

[nonoirie]

Re: Stats-Based CPU Sliders [MLB 14 Version]

Nomo,

Couple of questions;

1. When you start a new batch, do you restart the season or do you continue for the previous batch and adjust the slider?

2. IP-SP and NP was dead-on at default and 4/9. IMO, other sliders you tested on later batches should make IP-SP higher and NP lower since you increased starter stamina and lowered the manager hook but your #s suggest it's going the opposite direction. What I believe is that Pitcher Control/ Consistency/ Strike Freq effects IP-SP and NP more than ST stamina and manager hook. What's your take on this?

[nomo17k]

Re: Stats-Based CPU Sliders [MLB 14 Version]

Hey everyone, long time lurker here; decided to finally register. I've been using your sliders the past couple years and giving it a shot again now. Few things I've noticed:

Using it with quick counts, the number of home runs (especially opposite field shots barely clearing walls) seem to be way inflated, compared to real life and the results posted on this board. To counter that, I bumped the CPU home run power down to 4, and the overall offense took a major nosedive. In 8 games, all but one had a score or either 1-0 or 0-0 by the sixth inning, using starters 1 through 5. Most runs scored was 5, and that only happened once. Granted, 8 games is a small sample, but atm it appears getting this year's game to produce realistic results is more difficult than last year.

I see you've began testing with QC. That's how I've been playing since using your sliders. So far, it looks like results are a lot different than your previous sets. I know the sample size is still really small, but it still looks pretty extreme. Since I'm on my phone I can't check the sliders from previous iterations. Have you changed them since the last set on 5/7?

I wouldn't think QC would make a difference in how the game would play based off slider alignments, but then again, what do I know. I didn't code the game! I'll be interested in seeing what you find. At least now you'll be able to get results from the cpu VS cpu games quicker.

I'm running a few CPU vs. CPU games with QC just out of curiosity now that I'm mostly satisfied with the main slider set (except the SB issue, of course).

From how things look (only several games so far), the game does seem to play a bit differently in terms of the tendency to produce walks and strikeouts. Both are very high compared to non-QC games. Also, the batting average seems to be a bit down, but the sample size is too small yet.

I guess these are kind of expected given that batters are given less chance to hit basically by starting from a deeper count, closer to being struck out or walking.

[nomo17k]

Re: Stats-Based CPU Sliders [MLB 14 Version]

Are you using reliever stamina at 6 or 5? On here you say 6, but the spreadsheet for 5/7 you say 5.

I bumped up Reliever Stamina to 6 (to be on par with Starter Stamina) pretty much only because I have been seeing consistently the number of pitchers used in each game (Np in the spreadsheet) is higher than the MLB average. While I have to look deeper into how pitchers are used to get a better sense of why this is so, I have been noticing that at times (often enough that I notice when I scan box scores to records stats) the CPU manager tends to do successive one-point relievings (i.e., pitching to just one batter and getting relieved). This happens even with Manager Hook at 4.

I suspect that this happens when the energies of relievers are generally low... below the threshold that CPU uses to replace a pitcher. I notice this often with teams after a long extra inning game when even the next-day starter gets brought in. So I'm giving relievers a bit more stamina so that they don't get exhausted as easily.

Besides, I think most of the time pitchers are relieved because of situation, not because of lack of energy. Giving them less stamina may cause trouble like above but I don't think giving them a bit more stamina is ever really harmful.

Nomo,

Couple of questions;

1. When you start a new batch, do you restart the season or do you continue for the previous batch and adjust the slider?

2. IP-SP and NP was dead-on at default and 4/9. IMO, other sliders you tested on later batches should make IP-SP higher and NP lower since you increased starter stamina and lowered the manager hook but your #s suggest it's going the opposite direction. What I believe is that Pitcher Control/ Consistency/ Strike Freq effects IP-SP and NP more than ST stamina and manager hook. What's your take on this?

I'm continuing with the same season file when I change sliders. Only this time PS3 and PS4 versions are branched out separately. Generally I want to accumulate enough stats with a single CPU vs. CPU season and look at stats purely generated by the gameplay. And smile like a creep.

As for IP-SP and Np, I think there are a couple effects in play. One is what you mentioned, the increased pitch count due to me making them throw a bit more pitches out of the strike zone to increase walks. Another is that once you get deeper into a season, your pitchers are not always fully rested because of the rotation and occasional extra inning games, etc. I think that effect may be showing up a little after the initial 75-game set.

And that's basically the reason why I bumped up the Reliever Stamina slider. Whether that yields an intended effect, I'm not really sure but I don't want to see the CPU manager making pitching changes one at-bat after another only because most relievers are not fully rested.


I hope CPU manager manages the bullpen a bit more cleverly.

[nonoirie]

Re: Stats-Based CPU Sliders [MLB 14 Version]

I hope CPU manager manages the bullpen a bit more cleverly.

This.................. Bottom line is this....... Boy, I wish........

[kcsam]

Re: Stats-Based CPU Sliders [MLB 14 Version]

I'm running a few CPU vs. CPU games with QC just out of curiosity now that I'm mostly satisfied with the main slider set (except the SB issue, of course).

From how things look (only several games so far), the game does seem to play a bit differently in terms of the tendency to produce walks and strikeouts. Both are very high compared to non-QC games. Also, the batting average seems to be a bit down, but the sample size is too small yet.

I guess these are kind of expected given that batters are given less chance to hit basically by starting from a deeper count, closer to being struck out or walking.

I totally agree with that. I am seeing an awful lot of K's and walks per game. In some stretches, I've seen #3 or #4 pitchers records 5 K's in a row. Walks also seem very high as well. I know it's not out of the realm of reality, but I've seen a lot of times where a pitcher will walk the bases loaded and then walk in a run.

And I also agree with batting averages. A lot of my games are 0-0 until around the 5th or 6th inning. Although, last night while I was doing an MLB The Show Love game with the White Sox vs. The A's, the Sox had 8 runs and 15 hits. By far the most hits I've seen since playing the game. Outside of that, I'm only seeing 5-10 hits per team per game.

I'll be very interested in seeing what your QC tests yield and how the sliders need to be adjusted to make up for the differences. Once we get that all ironed out, I think I'll finally start my franchise

[nomo17k]

Re: Stats-Based CPU Sliders [MLB 14 Version]

I totally agree with that. I am seeing an awful lot of K's and walks per game. In some stretches, I've seen #3 or #4 pitchers records 5 K's in a row. Walks also seem very high as well. I know it's not out of the realm of reality, but I've seen a lot of times where a pitcher will walk the bases loaded and then walk in a run.

And I also agree with batting averages. A lot of my games are 0-0 until around the 5th or 6th inning. Although, last night while I was doing an MLB The Show Love game with the White Sox vs. The A's, the Sox had 8 runs and 15 hits. By far the most hits I've seen since playing the game. Outside of that, I'm only seeing 5-10 hits per team per game.

I'll be very interested in seeing what your QC tests yield and how the sliders need to be adjusted to make up for the differences. Once we get that all ironed out, I think I'll finally start my franchise

I'm not going to try to make a slider set for QC, because I don't think sliders will change CPU strategy enough, if all, to make it play realistically against each other.

For example, in QC games I'm seeing full count (3-2) nearly three times more than I have seen in non-QC games. How should I adjust the sliders so that CPU puts ball in play before going deep in count? That's an aspect of the game you cannot really change very much even with sliders.

IMO, the kind of gamers who can end up playing realistically are those who don't take a lot of pitches and put balls in play more than average MLB players do... and that's likely the vast majority of gamers, so if developers are tuning the game for them, then that is the right thing to do. But to cater to those who can already play the game realistically (that includes CPU AI), I think the game needs to offer a slightly different logic than what it does.

One thing that I find interesting though is that the CPU behavior with QC isn't the same as fully played out game with the same count/situation. For example, what I'm seeing seems to suggest that, given a 2-1 count with everything else being equal, CPU with and without QC seem to approach the game differently. In other word, QC is not a simple fast-forward mechanism.

I'm actually not very fond of the way QC is implemented as a different game and not a mechanism to fast-forward the pace of the existing game. It's as if we now have three different game engines, the gameplay, sim, and QC, for which tuning needs to be done separately. In my wet dream, I would rather see each mode based on the full gameplay engine, but stripped out so that it does not need to slow down to render the graphics (for sim) and some mechanism to skip/fast-forward the portion of the game that you don't want to see (somewhat like QC/result play).

That way, we'd have more seamless experiences across different modes. But that's not likely to happen because of technical complications from what I heard... hence I keep wet dreaming.

[kcsam]

Re: Stats-Based CPU Sliders [MLB 14 Version]

I'm not going to try to make a slider set for QC, because I don't think sliders will change CPU strategy enough, if all, to make it play realistically against each other.

For example, in QC games I'm seeing full count (3-2) nearly three times more than I have seen in non-QC games. How should I adjust the sliders so that CPU puts ball in play before going deep in count? That's an aspect of the game you cannot really change very much even with sliders.

IMO, the kind of gamers who can end up playing realistically are those who don't take a lot of pitches and put balls in play more than average MLB players do... and that's likely the vast majority of gamers, so if developers are tuning the game for them, then that is the right thing to do. But to cater to those who can already play the game realistically (that includes CPU AI), I think the game needs to offer a slightly different logic than what it does.

One thing that I find interesting though is that the CPU behavior with QC isn't the same as fully played out game with the same count/situation. For example, what I'm seeing seems to suggest that, given a 2-1 count with everything else being equal, CPU with and without QC seem to approach the game differently. In other word, QC is not a simple fast-forward mechanism.

I'm actually not very fond of the way QC is implemented as a different game and not a mechanism to fast-forward the pace of the existing game. It's as if we now have three different game engines, the gameplay, sim, and QC, for which tuning needs to be done separately. In my wet dream, I would rather see each mode based on the full gameplay engine, but stripped out so that it does not need to slow down to render the graphics (for sim) and some mechanism to skip/fast-forward the portion of the game that you don't want to see (somewhat like QC/result play).

That way, we'd have more seamless experiences across different modes. But that's not likely to happen because of technical complications from what I heard... hence I keep wet dreaming.

Interesting. I've never played a non QC game, so I can't speak on the full count piece, but I have noticed there are a lot in QC. Are you saying there are 3 times more full counts in QC off the bat? As in, when a new batter comes up he is automatically setup with a 3-2 count? If so, then yeah, I totally agree. Nothing you can do with sliders to prevent that. You'd have to dig deep into the code to change that. That's sort of a drag because I really enjoy the fast play of QC in Mom and cpu vs. cpu (the 2 ways I play)

Are you going to continue with your cpu vs cpu testing with QC? I'd still be interested in seeing how the numbers play out. I don't necessarily care about the 3-2 counts. As long as we can get AVG., HR's, RBI's, Walks, K's and the other "important" stats ironed out, I may stick with it. If not, then I guess I am going to have to go back to full games and just skip QC all together. Who knows, maybe with a larger sample size some of these #'s will average out like your other slider set (wishfully thinking) ?

[nomo17k]

Re: Stats-Based CPU Sliders [MLB 14 Version]

Interesting. I've never played a non QC game, so I can't speak on the full count piece, but I have noticed there are a lot in QC. Are you saying there are 3 times more full counts in QC off the bat? As in, when a new batter comes up he is automatically setup with a 3-2 count? If so, then yeah, I totally agree. Nothing you can do with sliders to prevent that. You'd have to dig deep into the code to change that. That's sort of a drag because I really enjoy the fast play of QC in Mom and cpu vs. cpu (the 2 ways I play)

Are you going to continue with your cpu vs cpu testing with QC? I'd still be interested in seeing how the numbers play out. I don't necessarily care about the 3-2 counts. As long as we can get AVG., HR's, RBI's, Walks, K's and the other "important" stats ironed out, I may stick with it. If not, then I guess I am going to have to go back to full games and just skip QC all together. Who knows, maybe with a larger sample size some of these #'s will average out like your other slider set (wishfully thinking) ?

I'll just be running a bit more test with QC to see where the stats stabilize. If the deviations aren't too big, then there might be some way to adjust things a bit, but I feel the main issue is that QC is basically a different mode of gameplay. Otherwise, the same sliders should just work.

As for the number of 3-2 count ABs, it's how many times they are reached, regardless of what count is generated at the beginning of ABs. Clearly, the numbers seem to indicate that CPU is either taking or not putting ball in play as often as it does IRL, given whatever count it's given at the beginning of an AB. Or the generated counts tend to push things deeper to full count to begin with. Probably both.

[kcsam]

Re: Stats-Based CPU Sliders [MLB 14 Version]

I'll just be running a bit more test with QC to see where the stats stabilize. If the deviations aren't too big, then there might be some way to adjust things a bit, but I feel the main issue is that QC is basically a different mode of gameplay. Otherwise, the same sliders should just work.

As for the number of 3-2 count ABs, it's how many times they are reached, regardless of what count is generated at the beginning of ABs. Clearly, the numbers seem to indicate that CPU is either taking or not putting ball in play as often as it does IRL, given whatever count it's given at the beginning of an AB. Or the generated counts tend to push things deeper to full count to begin with. Probably both.

That'll be great. Hopefully the differences with QC aren't too great and you can seemingly adjust accordingly. I'd be much appreciative of that.

It's too bad there's not a slider you can adjust for "putting ball in play" or something along those lines. That way the deep counts wouldn't throw things off. This must be what is inflating the walks and K's in QC and limiting .AVG.