Stats-Based CPU Sliders [MLB 14 Version]

Collapse

Recommended Videos

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • nomo17k
    Permanently Banned
    • Feb 2011
    • 5735

    #76
    Re: Stats-Based CPU Sliders [MLB 14 Version]

    Originally posted by BrianU
    I like your approach to pitching sliders and not nerfing control into the ground. My only area of contention is the fielding and base running sliders. I know you are still in the stage of sticking to defaults and using defaults as a control to see what needs to be adjusted and I haven't gotten too much playing time in yet, but from my experience the base running speed results in some really funky plays. Non super fast guys beating out grounders that look relatively routine, and choppers too. I know you do not observe every game but curious if you have seen any instances of this. I just have a feeling if we had a statistic stating how many infield grounders made up total hits, we would see a an over-representation from whatever that number is in real life. We are able to track strike ratios on the various counts very closely but hits being only broken down by GB/LD/FB on contact leaves more to be desired in shaping the offensive side of the game.
    My observation on these plays so far is that I tend to see a lot more bang-bang plays at 1B than I remember before, but many more of them turn into outs than safe. This is just impression and I am not saying from evidence that I can supply. But I feel the reason may be that a few new fielding animations that I often see now physically take more time to make those plays, which in my view is a good thing because they remove the feeling of rush that I think was not really necessary on some plays.

    In my opinion, infield hits are an issue that cannot be solved just by tweaking BR Speed, and the real culprit affecting the amount of infield hits is the "poor" hit variety in the game.

    As far as I see, prior to MLB 14, most infield hits happened in (1) grounders to the left side of SS where he needs to make a long throw, (2) grounders to the right side of SS where he nonchalantly makes two-step throw, and (3) choppers around the plate fielded by either P or C. I think (1) is fine. The (2) was an animation issue which has been addressed this year by a new animation of SS making a snap throw while running, so it's probably okay. The nagging kind is (3), where a weakly hit chopper stays around the home plate.

    I think the game produces a bit too many of (3) as a result of poor contact, when in real life generating that type of hit is very difficult or nearly impossible. The balance of out/safe mostly depends on how fast P/C fields the ball and makes the throw (with varying results in the past). But in real life, choppers tend to bounce a lot harder off the ground, like this:

    Spoiler


    They often bounce even higher than infielders can reach. A lot of them are actually the reverse of fly balls, just the launch angle is down toward the ground instead of going up in the air. You still don't really see this kind of "poor" hits in the game.

    Choppers we see in the game tend to be of very weakly hit variety that stays and nerfed totally around the plate. I think that is the major issue right now, as far as infield hits of the kind you talk about are concerned. And I don't think this is physically realistic, because after all you are hitting a baseball pitch coming at you with a very high velocity and therefore a lot of energy. And that energy needs to go somewhere after a contact is made (i.e., conservation of energy), so if that energy is not in the batted ball, then it will go to either the bat (which sometimes breaks because of this) or the body of the hitter (who might feel sting). (As an aside, this is also why when squaring up with the sweetest spot, the hitter does not really feel much sting in his hands, because the transfer of energy from the bat/body is so smooth that much of it goes to the ball itself.)

    So that's my opinion. Rather than fixing through BR speed, the poor hit variety needs to be tweaked in the game.


    I appreciate greatly the statistically accurate approach which works well with some sliders, both the speed and reaction sliders I feel need to be looked at from an eye test approach as well. Something I am not the greatest at so I use others sound reasoning. Personally I can see a need to lower fielder reaction to 4, baserunning speed to 3 or 4, arm strength to 3 or lower, and fielder speed a bit. The challenge is finding a balance to reproduce accurate 2B/3B numbers in combination with passing the eye test of the gameplay.

    I don't mean to go on a tangent, just curious if you plan to touch those group of sliders at all?
    I actually lowered them for the newest set. I'm just not in a hurry to tweak those sliders till I have enough games played to see if the tweaks are necessary. But in general I think the fielding side of the game has been slowed down further, so I don't think it is as much of an issue as it were in the past.


    If you don't mind can I ask what made you raise solid hits to 7? I had thought that slider mainly impacted LD% but those numbers looked good on the default testing.
    It was a compromise that I made to increase BABIP without affecting a lot of other things, like plate discipline. I almost needed to account for .030 in BABIP which is quite huge, so I assumed increasing line drives is the most effective way of doing this. I would be more inclined to reduce FBs and increase GBs rather than just increasing the proportion on LDs, but you cannot do that with the current set of sliders.


    I know you addressed the BB being low still and how you are working on that, but how about strikeouts? Would lowering timing a notch be enough or would that have other unintended effects you wish to avoid?

    I always look forward to see what changes you make to balancing things and your approach. Thanks for another year of unparalleled work!
    I'm not really concerned about K% that much. It might be slightly down because of the changes that I have been making, but not outside of the realistic range. If I have to change this, then I will probably try using Foul Frequency slider first. Timing slider might affect solid hits, etc., which might not be desirable.
    The Show CPU vs. CPU game stats: 2018,17,16,15,14,13,12,11

    Comment

    • toyo301985
      Rookie
      • Jan 2012
      • 168

      #77
      Re: Stats-Based CPU Sliders [MLB 14 Version]

      where can i find your sliders in the vault?

      Comment

      • bcruise
        Hall Of Fame
        • Mar 2004
        • 23274

        #78
        Re: Stats-Based CPU Sliders [MLB 14 Version]

        Originally posted by toyo301985
        where can i find your sliders in the vault?
        I doubt Nomo will put anything in the vault until he's got something he's happy with (if he puts anything in there at all). He always takes a large sample size of CPU games and then makes small adjustments based on what he finds out.

        Comment

        • tessl
          All Star
          • Apr 2007
          • 5676

          #79
          Re: Stats-Based CPU Sliders [MLB 14 Version]

          Originally posted by Brian SCEA
          Quick update. I look at this yesterday and this morning, and also ran some deeper tests overnight.

          A few observations: Below average stealers (around 25 ability, 50 speed) fail more than they should, and it's partly hidden because they don't steal often to begin with. For decent stealers (50 ability, 75 speed) - let's call them "2nd-tier" stealers - a lot of steals are being decided by a very small margin (i.e. a few frames or <50-100ms), and this clustering is denser than it should be. Pitch speed can only account for a fraction of the difference, but still more than it would on paper because of the clustering. Something else must explain the rest of the gap. The most important lead I found was that certain pitchers appear to unrealistically hard to steal against.

          I did a search and found one significant change from last year that would account for some of this. It's intended to aesthetically prevent runners from getting "too good" a jump against pitchers, but the actual result is that for pitchers with fast stretch deliveries and also 2nd and 3rd tier runners the steal balance is affected. It causes clustering where they're frequently "almost safe" because of the cap, and even a small shift either way will change steal balance more than it normally should. In addition, I'm looking at other factors related to this that make it hard for lower tiered runners to steal. Unfortunately, since the issue is caused by a cap, the base running slider will only help whenever the cap is not happening (ironically).

          I'm working on other bugs as well, but when all things are done I'd be happy to go into more detail.
          That's interesting. I'm wondering if it is possible to revert to last year's animation of that's the problem and if the ps4 has the same animation.

          I won't have the game until May 6 but since I'm retired and I play manage mode I probably play - not sim - as many games as anybody in what essentially is cpu vs cpu mode. In mlbts13 I adjusted only one slider - wind. Gameplay was solid.

          Comment

          • ralphieboy11
            Pro
            • Jul 2005
            • 543

            #80
            Re: Stats-Based CPU Sliders [MLB 14 Version]

            Originally posted by tessl
            That's interesting. I'm wondering if it is possible to revert to last year's animation of that's the problem and if the ps4 has the same animation.

            I won't have the game until May 6 but since I'm retired and I play manage mode I probably play - not sim - as many games as anybody in what essentially is cpu vs cpu mode. In mlbts13 I adjusted only one slider - wind. Gameplay was solid.
            I'm really hoping this gets addressed in a patch. Even when I bump steal ability all the way up and steal frequency all the way down, I still get more people caught stealing that successful stolen bases.

            Comment

            • Heroesandvillains
              MVP
              • May 2009
              • 5974

              #81
              Re: Stats-Based CPU Sliders [MLB 14 Version]

              Originally posted by ralphieboy11
              I'm really hoping this gets addressed in a patch. Even when I bump steal ability all the way up and steal frequency all the way down, I still get more people caught stealing that successful stolen bases.
              Have you tried SB Ability at 0?

              Maybe the slider is reversed (again).

              Comment

              • BrianU
                MVP
                • Nov 2008
                • 1565

                #82
                Re: Stats-Based CPU Sliders [MLB 14 Version]

                Originally posted by Heroesandvillains
                Have you tried SB Ability at 0?

                Maybe the slider is reversed (again).
                looking at nomos 4/11 sliders which had SB Ability 10 resulting in Steal percentage 41.5% compared to his 4/15 data set with SB Ability 0 and Steal percentage 37.5% it looks like the issue is deeper than any thing we can do. Brian seems to have a good hold of what's going on though I know they will tune it.

                The 4/15 results are looking really damn good. Will be interesting to see what happens after patches. I am avoiding quick counts for now because I think that is gonna get adjusted the most of anything. Do you have any plans to do an abbreviated quick counts test once you solidify the CPU vs CPU full game set? I'd be interested in seeing that. I actually saw you can view CPU vs CPU games in franchise mode with quick counts on. I plan to use that a little for the end of the season and some playoff games if I get that far!

                A little side tidbit I have experienced with regard to trying to translate nomos sliders into a User vs CPU set.. I use auto everything I can along with timing hitting/classic pitching and i've found to get accurate foul ball numbers I have to boost foul ball frequency to at least 6 or 7 for both USER and CPU. I am not sure why this is, it must be my pitching style differs greatly from how the CPU pitches. I don't use API I decide pitches and location independently. It makes me wonder too how the CPU decides pitch effort in regards to choosing between minimum effort taps compared to maximum effort holds of the pitching buttons.

                My batting stats are not in line with MLB averages because quite frankly I suck and swing at bad pitches a lot but rather than adjust sliders I prefer adjusting my patience/choices so tracking my own batting stats helps me see what I am doing at the plate and what I should not be doing. The CPU batting stats such as swing rate, chased/miss rate, all look good. Same for the CPU pitching stats.

                Comment

                • nomo17k
                  Permanently Banned
                  • Feb 2011
                  • 5735

                  #83
                  Re: Stats-Based CPU Sliders [MLB 14 Version]

                  Originally posted by toyo301985
                  where can i find your sliders in the vault?
                  Yup, as bcruise wrote, I'm usually not in a hurry to upload a slider set to the vault. I usually do so when I think all there is left is minor adjustments.


                  Originally posted by BrianU
                  looking at nomos 4/11 sliders which had SB Ability 10 resulting in Steal percentage 41.5% compared to his 4/15 data set with SB Ability 0 and Steal percentage 37.5% it looks like the issue is deeper than any thing we can do. Brian seems to have a good hold of what's going on though I know they will tune it.
                  I don't have as big a sample for SBA at 0 so I cannot comment on whether the slider is still reversed or not. It does look like no slider adjustment is enough to "correct" the steal success issues though.


                  The 4/15 results are looking really damn good. Will be interesting to see what happens after patches. I am avoiding quick counts for now because I think that is gonna get adjusted the most of anything. Do you have any plans to do an abbreviated quick counts test once you solidify the CPU vs CPU full game set? I'd be interested in seeing that. I actually saw you can view CPU vs CPU games in franchise mode with quick counts on. I plan to use that a little for the end of the season and some playoff games if I get that far!
                  I am not planning to do much on quick count, as the data isn't as easy to gather (e.g., strike-ball count breakdowns, etc.). However, I don't really think a different slider set *should* be necessary, if the counts are generated realistically. All it does (on surface) is with each at bats, first few pitches are skipped, so whatever CPU does after that should be very similar regardless of whether that count was reached via QC or non-QC. If anything looks "off," then correcting that for realism really takes the algorithm for the count generation to be tweaked, which we cannot really do though sliders.


                  A little side tidbit I have experienced with regard to trying to translate nomos sliders into a User vs CPU set.. I use auto everything I can along with timing hitting/classic pitching and i've found to get accurate foul ball numbers I have to boost foul ball frequency to at least 6 or 7 for both USER and CPU. I am not sure why this is, it must be my pitching style differs greatly from how the CPU pitches. I don't use API I decide pitches and location independently. It makes me wonder too how the CPU decides pitch effort in regards to choosing between minimum effort taps compared to maximum effort holds of the pitching buttons.

                  My batting stats are not in line with MLB averages because quite frankly I suck and swing at bad pitches a lot but rather than adjust sliders I prefer adjusting my patience/choices so tracking my own batting stats helps me see what I am doing at the plate and what I should not be doing. The CPU batting stats such as swing rate, chased/miss rate, all look good. Same for the CPU pitching stats.
                  To correct for your own playing tendency and make the stats in line with real-life MLB numbers, you generally need to make some exaggerated changes.

                  When you control all 9 players, each of them plays according to your own tendency, whereas in real life all of them have different tendency. The stats you collect are aggregate, so the former will be just be how you play but the latter is an average of all 9 with different tendencies playing. There will be big differences because of this.

                  And the difference between the minimum and maximum pitch effort is not actually that big it seems:

                  http://www.operationsports.com/forum...&postcount=194
                  The Show CPU vs. CPU game stats: 2018,17,16,15,14,13,12,11

                  Comment

                  • tessl
                    All Star
                    • Apr 2007
                    • 5676

                    #84
                    Re: Stats-Based CPU Sliders [MLB 14 Version]

                    Originally posted by ralphieboy11
                    I'm really hoping this gets addressed in a patch. Even when I bump steal ability all the way up and steal frequency all the way down, I still get more people caught stealing that successful stolen bases.
                    The best developer is working on it - Brian so I suspect it will get patched.

                    I'm still wondering if the patches for the ps3 will apply to the ps4. If not that would be a giant headache for the devs.

                    Comment

                    • BrianU
                      MVP
                      • Nov 2008
                      • 1565

                      #85
                      Re: Stats-Based CPU Sliders [MLB 14 Version]

                      Originally posted by nomo17k
                      Yup, as bcruise wrote, I'm usually not in a hurry to upload a slider set to the vault. I usually do so when I think all there is left is minor adjustments.




                      I don't have as big a sample for SBA at 0 so I cannot comment on whether the slider is still reversed or not. It does look like no slider adjustment is enough to "correct" the steal success issues though.




                      I am not planning to do much on quick count, as the data isn't as easy to gather (e.g., strike-ball count breakdowns, etc.). However, I don't really think a different slider set *should* be necessary, if the counts are generated realistically. All it does (on surface) is with each at bats, first few pitches are skipped, so whatever CPU does after that should be very similar regardless of whether that count was reached via QC or non-QC. If anything looks "off," then correcting that for realism really takes the algorithm for the count generation to be tweaked, which we cannot really do though sliders.




                      To correct for your own playing tendency and make the stats in line with real-life MLB numbers, you generally need to make some exaggerated changes.

                      When you control all 9 players, each of them plays according to your own tendency, whereas in real life all of them have different tendency. The stats you collect are aggregate, so the former will be just be how you play but the latter is an average of all 9 with different tendencies playing. There will be big differences because of this.

                      And the difference between the minimum and maximum pitch effort is not actually that big it seems:

                      http://www.operationsports.com/forum...&postcount=194
                      I play with only the Mets so I should obviously have more K's and lower offensive output than the MLB average but I do not think that means I can't still try to emulate the swung%, chased%, or have the same amount of swinging strikes, etc.. I could be wrong but I assume the numbers wouldn't vary that much on the whole. I haven't checked for team specific numbers in these areas I will look on Fangraphs I think they have that there. It may be that tuning the User hitting sliders in small ways based on the team you use is a good idea. In Madden I would use a different QB accuracy based on the QB I was using. There are lots of ways to use sliders it is a big reason why I love sports games so much.

                      I know each player is different though. The different PCI sizes plus choosing which players to let use contact swing/power swing (I try to keep them specific to guys based on walks/low Ks and HRs/Dbls, for example I wouldn't use contact with Granderson, and I wouldn't use Power with Murphy.) I wish there was more ways SCEA differentiated using different hitters as the user. Right now you are just as good at drawing walks with Votto as you are with Granderson, other than contact swing being a bit tighter for Votto. Your eye is still the same it's not the hitters eye but your own eye. I am not sure how they could make Plate Discipline mean more when you control the batter and highlight how different players have a different eye for the ball. I wish also that they implement a bat speed rating and a reach rating.

                      On the pitching side MLB 2k did something really innovative towards the end of their cycle. They had a rating for each pitchers' usage of each pitch. If you started to overuse a pitch compared to how that pitcher uses it in real life, it would diminish the effectiveness of it. This would benefit both the user and the AI if they added usage. The user would be forced to mix it up more and the AI would be much more varied as well. It bothers me that we can't have the CPU pitchers simulate usage rates but instead they all rely on whatever pitch is in their #1 slot, #2 slot, #3 slot.

                      With regard to how the CPU uses minimum/maximum effort I wasn't thinking so much in terms of effect on stamina but instead on location/strikeout potential. If the CPU always uses max effort but the way I pitch is using min effort 80% of the time then I can see why the divergence in my results compared to theirs. That is interesting though to see because I did think stamina would be affected more. I still use min effort to start the game as a way to locate pitches better and build up confidence. Once a pitch is feeling good I switch over to max effort if it's a pitch I want contact or a swinging strike.

                      Comment

                      • DarthRambo
                        MVP
                        • Mar 2008
                        • 6630

                        #86
                        Re: Stats-Based CPU Sliders [MLB 14 Version]

                        Through 16 games in my test franchise I've played all btw...the biggest problem I'm having is batting average. I'm batting .230 and the cpu is at .238

                        I also lead the league in strikeouts. I'm playing with the Indians so they aren't a suberb pitching staff by any means. Hit variety is weak imo this year. I'm going to raise solid hits to 7 like u have. But has that helped the low batting average you were seeing??

                        I use quick counts which I know effects some I'm sure but regardless the batting averages shouldn't be so low.

                        Sent from my SPH-L710 using Tapatalk
                        Last edited by DarthRambo; 04-19-2014, 01:41 PM.
                        https://www.youtube.com/DarthRambo

                        Comment

                        • kcsam
                          Pro
                          • Feb 2011
                          • 676

                          #87
                          Re: Stats-Based CPU Sliders [MLB 14 Version]

                          Hey Nomo. Any more recent changes to your sliders based off further testing?

                          Comment

                          • nomo17k
                            Permanently Banned
                            • Feb 2011
                            • 5735

                            #88
                            Re: Stats-Based CPU Sliders [MLB 14 Version]

                            Originally posted by BrianU
                            ...

                            I know each player is different though. The different PCI sizes plus choosing which players to let use contact swing/power swing (I try to keep them specific to guys based on walks/low Ks and HRs/Dbls, for example I wouldn't use contact with Granderson, and I wouldn't use Power with Murphy.) I wish there was more ways SCEA differentiated using different hitters as the user. Right now you are just as good at drawing walks with Votto as you are with Granderson, other than contact swing being a bit tighter for Votto. Your eye is still the same it's not the hitters eye but your own eye. I am not sure how they could make Plate Discipline mean more when you control the batter and highlight how different players have a different eye for the ball. I wish also that they implement a bat speed rating and a reach rating.
                            I think that's the price you have to pay for having all players under your control. That by itself is already very unrealistic (all players controlled by a gamer with fixed skill set and thought process).

                            Other than the fact that CPU pitches around power threats, Plate Discipline is pretty much the only attribute that could differentiate good walkers from hackers who don't walk. I think that's a good approach, but I feel the effect of having high and low Plate Discipline could be made a bit wider, so that if you check with a very patient hitter IRL, then you can get away with your own bad reaction a bit. That kind of adjustment I feel can still be made.


                            On the pitching side MLB 2k did something really innovative towards the end of their cycle. They had a rating for each pitchers' usage of each pitch. If you started to overuse a pitch compared to how that pitcher uses it in real life, it would diminish the effectiveness of it. This would benefit both the user and the AI if they added usage. The user would be forced to mix it up more and the AI would be much more varied as well. It bothers me that we can't have the CPU pitchers simulate usage rates but instead they all rely on whatever pitch is in their #1 slot, #2 slot, #3 slot.
                            I am actually not really fond of that approach. It may look good and kind of make sense if the game is truly meat as an MLB simulation so that all existing players are meant to play like their real-life counterparts. But I actually find that approach very limiting. What if the player in real life is struggling because he actually has a bad approach? In the game, then you might be able to use him more effectively by using a better approach. But if you are always forced to mold your playing style into what the game think how the in-game player plays, then you aren't in full control of your players. I don't know if that's fun.

                            I for sure do think the game will benefit from having more variety in CPU strategies, and the major reason why the game still fails to replicate certain types of real-life players is because CPU lacks variety in strategies that it employs. It's consistently good, but not all real-life baseball players think and approach the game consistently. So what I am hoping is that the game starts evolving in that direction, where a few different types of CPU AIs are trained to simulate different types of real-life players (or manager or even GM).
                            The Show CPU vs. CPU game stats: 2018,17,16,15,14,13,12,11

                            Comment

                            • nomo17k
                              Permanently Banned
                              • Feb 2011
                              • 5735

                              #89
                              Re: Stats-Based CPU Sliders [MLB 14 Version]

                              Originally posted by IrishSalsa
                              Through 16 games in my test franchise I've played all btw...the biggest problem I'm having is batting average. I'm batting .230 and the cpu is at .238

                              I also lead the league in strikeouts. I'm playing with the Indians so they aren't a suberb pitching staff by any means. Hit variety is weak imo this year. I'm going to raise solid hits to 7 like u have. But has that helped the low batting average you were seeing??

                              I use quick counts which I know effects some I'm sure but regardless the batting averages shouldn't be so low.

                              Sent from my SPH-L710 using Tapatalk
                              I would first decrease the hitting difficulty level and increase the pitching difficulty level if things continue the same way. I am not an advocate of tweaking sliders until you find yourself a good difficulty level. (You could also use the dynamic difficulty thing added this year.)

                              Hit variety actually is vastly better this year. Bloopers finally started happening last year, but this year there are a lot more poor/weak hit variety you can often see. To me this is one of the biggest improvement I see in this year's game.


                              Originally posted by kcsam
                              Hey Nomo. Any more recent changes to your sliders based off further testing?
                              I'm closing in on wrapping up with the first release-worthy set (about 10 games to go). I am not quite satisfied with the "walk" department yet (I want to see a bit more of them), but the rest seems pretty good, and I think it's fairly usable already.

                              We cannot do much with stealing issues... that needs a patch but the issue is in a good hand so hopefully things return to normalcy in a week or two.

                              You can always check the slider set I'm testing by going to the first cell (A1) in the spreadsheet tab and look at the comment.
                              The Show CPU vs. CPU game stats: 2018,17,16,15,14,13,12,11

                              Comment

                              • bcruise
                                Hall Of Fame
                                • Mar 2004
                                • 23274

                                #90
                                Re: Stats-Based CPU Sliders [MLB 14 Version]

                                Saw a great example of the stealing situation Brian was referring to just now in a CPU season I'm running - Anthony Rizzo and his 46 SPD, 15 BR ABIL just got caught stealing. Wasn't a pickoff or a hit and run, it was a straight steal. It's the second time he's tried in this season I'm running, and was cut down both times.

                                This was on default sliders, but I think you guys said this kind of thing was happening even with frequency turned down? Anyway, that's the kind of thing you don't need a sample size for to know something's wrong.

                                Rizzo, btw, has 11 stolen bases in his career and I'd wager most, if not all are on missed hit and runs.

                                Comment

                                Working...