programmed bias of CPU against human batter early in the count

Collapse

Recommended Videos

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • furball321
    Rookie
    • Apr 2015
    • 18

    #1

    programmed bias of CPU against human batter early in the count

    Briefly, I read over a month ago on the Operation Sports forums that the computer penalized a live player for swinging at the first two pitches of an at bat. I play offline Franchise, and I sorta got that impression on my own.

    I ran a test of 100 at bats swinging at the 1st pitch versus 100 at bats waiting til the 3rd pitch and got results that gave me a significantly higher batting average waiting til the 3rd pitch, even though that often put me significantly behind the count.

    Then I gave it more thought. What was going on, exactly? As I played dozens and dozens of Franchise games by hand, I found that, when I had GOOD results reported by the little at bat recorder for PCI placement and timing, I got fewer hits on the first two pitches of an at bat than I did on the third pitch of an at bat.

    So, I decided to do this test: Swing at the first pitch of an at bat. Record ONLY the swings that gave GOOD contact and GOOD timing. Then swing at the third pitch of an at bat and record ONLY the swings that gave GOOD contact and GOOD timing.

    I am not a hand-eye guru, so this took hundreds of at bats. These were not done in practice mode. They were done playing the SF Giants in Franchise mode in regular games against the rest of the league.

    The results are, I believe, fairly conclusive:

    In 200 at bats swinging at the first pitch when the result was GOOD placement and GOOD timing, I got an average of 88.7 speed of the hit and batted for a .266 average.

    In 200 at bats swinging at the third pitch - REGARDLESS OF THE COUNT - when the result was GOOD placement and GOOD timing, I got an average of 104.7 speed of the hit and batted for a .352 average.

    The numbers were compiled against the same few starting pitchers the Giants faced at the start of a Franchise season, as I began both tests from the season start. Again, that is with the entire Giants lineup, so it's not really scientific. But the results are so skewed as to convince me that the player is penalized for swinging at the first two pitches of an at bat.

    Look at it this way: If you que the ball up and get solid contact on the first pitch and hit ,266, but you que the ball up and get solid contact on the third pitch and hit .352, SOMETHING is going on besides physics.
  • 335TDC
    Rookie
    • Aug 2016
    • 352

    #2
    Re: programmed bias of CPU against human batter early in the count

    I believe this. It’s analogous to hitting the turbo button in Madden, which automatically disengages defenders from their blockers.


    Sent from my iPhone using Operation Sports

    Comment

    • Caulfield
      Hall Of Fame
      • Apr 2011
      • 10986

      #3
      Re: programmed bias of CPU against human batter early in the count

      FWIW these are MLB's 2017 batting averages based on the different ball/strike counts:
      1st pitch .348
      1-0 .341
      2-0 .360
      3-0 .414
      0-1 .330
      0-2 .152
      1-1 .340
      2-1 .349
      3-1 .364
      1-2 .159
      2-2 .181
      full count .216

      zero strikes .348
      zero balls .280
      3 balls .345
      2 strikes .176

      even count .273
      batter behind in count .202
      batter ahead in count .299
      swung at 1st pitch .270
      took 1st pitch .249
      Last edited by Caulfield; 06-24-2018, 02:57 PM. Reason: numbers transposed
      OSFM23 - Building Better Baseball - OSFM23

      A Work in Progress

      Comment

      • KBLover
        Hall Of Fame
        • Aug 2009
        • 12172

        #4
        Re: programmed bias of CPU against human batter early in the count

        Originally posted by furball321
        But the results are so skewed as to convince me that the player is penalized for swinging at the first two pitches of an at bat.
        I don't have definitive data as I don't track per count (maybe I should start if I can figure out how to manage it without spending more time in Excel than playing), but I know I've had 1st pitch successes. Sometimes, that's how I start rallies against a guy who's owning me and putting me 0-2, 1-2 all game or I put the throttle down looking to knock the pitcher out after he's rattled so I don't let him get strikes to work his confidence back up.

        I don't believe I've had more success swinging at, say, 0-2 pitches than 0-0 or 1-0, which would be before the 3rd pitch of the at-bat.

        I tend to hit deep into counts unless for specific reasons, as I like driving up pitch counts and working walks/seeing what the pitcher is going to do to me.

        So that might be part of it - I'm not an aggressive hitter, more a patient one that sees more 2-2, 3-2 than 1-1, 2-0 swings.
        "Some people call it butterflies, but to him, it probably feels like pterodactyls in his stomach." --Plesac in MLB18

        Comment

        • underdog13
          MVP
          • Apr 2012
          • 3222

          #5
          Re: programmed bias of CPU against human batter early in the count

          Not sure where you read about that on here. Not going to dismiss your results but as you mentioned there are numerous holes within the testing method. I will also add from a programmers perspective it doesn't make sense for them to program something like this in.

          Sent from my SM-G965U using Tapatalk
          PSN: Dalton1985
          Steam: Failure To Communicate

          Comment

          • underdog13
            MVP
            • Apr 2012
            • 3222

            #6
            Re: programmed bias of CPU against human batter early in the count

            Oh I also realized that I've screenshoted my rtts by ab count stats that I'll post here later

            Sent from my SM-G965U using Tapatalk
            PSN: Dalton1985
            Steam: Failure To Communicate

            Comment

            • Caulfield
              Hall Of Fame
              • Apr 2011
              • 10986

              #7
              Re: programmed bias of CPU against human batter early in the count

              Originally posted by underdog13
              Not sure where you read about that on here. Not going to dismiss your results but as you mentioned there are numerous holes within the testing method. I will also add from a programmers perspective it doesn't make sense for them to program something like this in.

              Sent from my SM-G965U using Tapatalk
              a new version of the old scene ~ s tyler

              but I wouldnt be surprised if there was more success when even a modicum of patience at the plate were displayed. and not for nefarious reasons but natural ones.
              OSFM23 - Building Better Baseball - OSFM23

              A Work in Progress

              Comment

              • bcruise
                Hall Of Fame
                • Mar 2004
                • 23274

                #8
                Re: programmed bias of CPU against human batter early in the count

                Originally posted by underdog13
                Oh I also realized that I've screenshoted my rtts by ab count stats that I'll post here later

                Sent from my SM-G965U using Tapatalk
                Good call, I've done that too:



                Legend difficulty, no slider changes except for fielder run speed down 2 (which, yes, inflates offense but it was still a consistent slider value through the whole season, therefore the comparison works).

                I'm not really buying into this. I was pretty bad with 2 strikes on me. If anything I probably hit better than I should have early in the count.

                3-0 is the biggest surprise there though - I guess I get a little too swing-happy and hit into some unlucky outs in that count.
                Last edited by bcruise; 06-24-2018, 04:56 PM.

                Comment

                • underdog13
                  MVP
                  • Apr 2012
                  • 3222

                  #9
                  Re: programmed bias of CPU against human batter early in the count

                  Originally posted by bcruise
                  Good call, I've done that too:



                  Legend difficulty, no slider changes except for fielder run speed down 2 (which, yes, inflates offense but it was still a consistent slider value through the whole season, therefore the comparison works).

                  I'm not really buying into this. I was pretty bad with 2 strikes on me. If anything I probably hit better than I should have early in the count.

                  3-0 is the biggest surprise there though - I guess I get a little too swing-happy and hit into some unlucky outs in that count.
                  3-0 is probably so low because you are probably taking 3-0 alot and draw a walk

                  Sent from my SM-G965U using Tapatalk
                  PSN: Dalton1985
                  Steam: Failure To Communicate

                  Comment

                  • KBLover
                    Hall Of Fame
                    • Aug 2009
                    • 12172

                    #10
                    Re: programmed bias of CPU against human batter early in the count

                    Originally posted by bcruise
                    Good call, I've done that too:




                    Sigh I'd love to see stats by count in franchise.
                    "Some people call it butterflies, but to him, it probably feels like pterodactyls in his stomach." --Plesac in MLB18

                    Comment

                    • bcruise
                      Hall Of Fame
                      • Mar 2004
                      • 23274

                      #11
                      Re: programmed bias of CPU against human batter early in the count

                      Originally posted by KBLover
                      Sigh I'd love to see stats by count in franchise.
                      The perks of RTTS - only having to track one player means you can get in a whole lot more detail for that one player.

                      Maybe some day...

                      Comment

                      • HighCmpPct
                        Denny 3K
                        • Oct 2011
                        • 3593

                        #12
                        Re: programmed bias of CPU against human batter early in the count

                        Not to mention, you have to account for pitch type, perks like first ball hitter, FB hitter, Off-speed hitter...

                        I know you put some time in on this but honestly you wasted it, there's really not gonna be a way to test this on the user side. You'd need developer tools to get the same pitch same spot every time with the same batter, even then there are so many variables as to what results in hits or outs.

                        Besides common baseball knowledge dictates working the count produces better results. Only guys trying to kill first pitches are guys that are just natural first pitch swingers and/or if the pitcher has been laying first pitch strikes deep in the zone all game.

                        Sent from my SM-G955U using Operation Sports mobile app

                        Join us in the 3K Gaming Discord for the best Sim Sports Setups!!
                        3K Gaming Discord

                        Link to my YouTube.
                        Denny 3K Gaming

                        Comment

                        • countryboy
                          Growing pains
                          • Sep 2003
                          • 52728

                          #13
                          Re: programmed bias of CPU against human batter early in the count

                          While I can appreciate the testing that was done, there are still too many unknown variables to definitively say this is an issue. A few missing variables are:

                          * Pitcher ratings/attributes
                          * Hitter ratings/attributes
                          * Pitch location
                          * Player hot/cold zones


                          All of these factor into the "formula" that determine hit type. Its not as simple as I took 200 swings on first pitch and 200 swings on hitter's count to determine issue or not. If it were, then player ratings/attributes/quirks would have no reason to be in the game.
                          I can't shave with my eyes closed, meaning each day I have to look at myself in the mirror and respect who I see.

                          I miss the old days of Operation Sports :(


                          Louisville Cardinals/St.Louis Cardinals

                          Comment

                          • TheWarmWind
                            MVP
                            • Apr 2015
                            • 2620

                            #14
                            Re: programmed bias of CPU against human batter early in the count

                            While the difference in average can be explained away for several reasons, but the difference in bat exit velocity is somewhat alarming.

                            Are you playing with fielder run speed at default (5)? If so, that would go a long way to explaining why you have such a big gap in average from first pitch to 3rd+. When runners in the field are really fast, power becomes a way more important factor in determining a hit.

                            However, even if that does fully explain the massive gap in average, the question still remains: why is there a difference in power?

                            Sent from my SM-G930W8 using Operation Sports mobile app

                            Comment

                            • saucerset
                              Rookie
                              • Jul 2010
                              • 482

                              #15
                              Re: programmed bias of CPU against human batter early in the count

                              Originally posted by countryboy
                              While I can appreciate the testing that was done, there are still too many unknown variables to definitively say this is an issue. A few missing variables are:

                              * Pitcher ratings/attributes
                              * Hitter ratings/attributes
                              * Pitch location
                              * Player hot/cold zones


                              All of these factor into the "formula" that determine hit type. Its not as simple as I took 200 swings on first pitch and 200 swings on hitter's count to determine issue or not. If it were, then player ratings/attributes/quirks would have no reason to be in the game.
                              The pitchers confidence at the time also matters. It is much harder to get a hit when the pitchers confidence is high. After a few first pitch swinging outs, the pitchers confidence goes up and his energy also doesn't drop as fast.

                              Isn't there also and AI learning aspect? Like if you keep swinging at the first pitch then the pitcher will start throwing junk on the first pitch because he knows you're swinging. If your taking more pitches then the odds of getting a sweet pitch goes up. At least that is my experience, I may be way off.

                              Comment

                              Working...