Statistical Analysis | cpu vs cpu | MLB 20 | Bacon Fictional Roster

[bakersville123]

Posting this data in case anyone finds it interesting if not useful.



Using default sliders (except injuries turned down to zero as a control), MLB 20, and Bacon's Fictional Roster (Excellent and highly recommended), I played out all games cpu vs cpu in the first five days of the season. This got me through most teams rotations 1-5, with some teams only going 1-4 with an early off day.



I compiled daily box score stats in the same manner as Nomo used to do; in fact I used his excellent MLB 18 .xls as a base (Thanks Nomo!).



Below you will find my results and in parenthesis, the MLB 2018 / 2019 averages.



MLB20 MLB2018/19

AB 33.47 (34.1 / 34.3)

R 3.79 (4.65 / 4.83)

H 7.99 (8.69 / 8.65)

RBI 3.71 (4.44 / 4.63)

BB 3.37 (3.26 / 3.27)

SO 8.66 (8.25 / 8.81)

2B 1.58 (1.73 / 2.16)

3B 0.03 (0.16 / 0.16)

HR 1.00 (1.26 / 1.40)

HBP 0.29 (0.36 / 0.41)

SF 0.22 (0.24 / 0.24)

SH 0.04 (0.19 / 0.16)

GIDP 1.05 (0.78 / 0.71)

SB 0.37 (0.52 / 0.47)

CS 0.38 (0.19 / 0.17)

SB% 46.75% (73% / 73%)

STARTERS AVG INN PITCHED PER START 5.47 (5.51 / 5.18)

NUMBER PITCHERS PER GAME 3.9 (4.22 / ?)

WP 0.16 (0.37 / 0.37)

E 0.35 (0.57 / 0.60)

[bakersville123]

Re: Statiscal Anaylsis | cpu vs cpu | MLB 20 | Bacon Fictional Roster

Observations:
-Well tuned OOB.

-Offense is somewhat muted. Result of a few things.

--Stolen Base % too low (more runners will lead to more offense)
--Errors to low (Too much clean fielding. Not just actual errors, but in terms of fielding effectiveness, getting to a potential gapper, infielder diving to get to a hard grounder etc)
--GIDP too high...obvious inning killers.

--Bunts being fielded to successfully (high exit velocity, "small" field feel by fielders on default fielding sliders). Throwing lead runner out in many cases.
--2B/3B's too low.

--HR's a little low, though those could rise over the season as weather warms?




Possible Corrections:
-Raise Fielding Error Sliders:
--Infield Fielding Errors +2 (to 7)
--Infield Throwing Errors +1 (to 6)
--Outfield Fielding Errors +1 (to 6)
-Raise Steal Success +1 (to 6)
-Lower Fielder Speed -1 (to 4)



On the Fence:
-Lower Infielder Arm Strength -1 (to 4)
-and / or Lower Fielder Reaction -1 (to 4)


Based on the data statistical analysis in the OP, I'm very close to achieving MLB average OOB and thus do not want to make to many slider changes. However, does anyone have suggestions to make some minor tuning to get me more in-line with MLB averages? (Thanks Salsa for your feedback in my other post btw!)

[RobotRosKO]

Re: Statistical Analysis | cpu vs cpu | MLB 20 | Bacon Fictional Roster

When you post 2 great slider info and no one,replies, has to sting a little. I found this helpful but by not changing to many sliders from the onset probably lead to the no responses.

[Kenya1991]

Re: Statiscal Anaylsis | cpu vs cpu | MLB 20 | Bacon Fictional Roster

Observations:
-Well tuned OOB.

-Offense is somewhat muted. Result of a few things.

--Stolen Base % too low (more runners will lead to more offense)
--Errors to low (Too much clean fielding. Not just actual errors, but in terms of fielding effectiveness, getting to a potential gapper, infielder diving to get to a hard grounder etc)
--GIDP too high...obvious inning killers.

--Bunts being fielded to successfully (high exit velocity, "small" field feel by fielders on default fielding sliders). Throwing lead runner out in many cases.
--2B/3B's too low.

--HR's a little low, though those could rise over the season as weather warms?




Possible Corrections:
-Raise Fielding Error Sliders:
--Infield Fielding Errors +2 (to 7)
--Infield Throwing Errors +1 (to 6)
--Outfield Fielding Errors +1 (to 6)
-Raise Steal Success +1 (to 6)
-Lower Fielder Speed -1 (to 4)



On the Fence:
-Lower Infielder Arm Strength -1 (to 4)
-and / or Lower Fielder Reaction -1 (to 4)


Based on the data statistical analysis in the OP, I'm very close to achieving MLB average OOB and thus do not want to make to many slider changes. However, does anyone have suggestions to make some minor tuning to get me more in-line with MLB averages? (Thanks Salsa for your feedback in my other post btw!)

Thank you for your work.
How do your sliders look now ?

[bakersville123]

Statistical Analysis | cpu vs cpu | MLB 20 | Bacon Fictional Roster

Wanted to share an update. I've spent the past few months testing cpu vs cpu sliders on MLB 20 w/ Bacon's roster, and after roughly 1,500 games and 44 official slider variations I've landed on a set that gets extremely close to a six year MLB average (2015-2021, excluding 2020).



In the end I went with the third variation I tested, which was actually just a curiosity and one I never expected to produce the stunningly, MLB accurate cpu vs cpu results that it did.



It requires only a single slider adjustment.


This supporting data is from a 142 game sample size which went through each team's rotation nearly twice (a handful of #5's did not go twice).


My Slider Results | (MLB2015-21 avg in parentheses)


BA .247 (.252)


BABIP .300 (.297)


SLUGGING% .403 (.417)


OB% .315 (.320)



BB% 8.51% (8.51%)


SO% 23.1% (21.9%)


G_Number (Runs by way of HR) 42.9% (Tough to find data but ideally want upper 30's%)


SB_ATT .81 (.69)


SB% 79.2% (72.5%)

AB 34.44 (33.98)

R 4.30 (4.53)

H 8.52 (8.55)

RBI 4.22 (4.32)

BB 3.25 (3.17)

SO 8.82 (8.33)

2B 1.77 (1.70)

3B .18 (.17)

HR 1.07 (1.20)

HBP 0.25 (0.38)

SF .19 (.25)

SH .08 (.19)

GIDP 1.0 (.74)

SB .64 (.50)

CS .17 (.19)

STARTERS AVG INN PITCHED PER START 5.42 (5.43)

NUMBER PITCHERS PER GAME 4.17 (4.28)

WP .12 (.37)

E .46 = .37 IF, .09 OF (.58)


The sliders I used to achieve the above are:


*BASERUNNER SPEED = 10 (Max)


That's it. All other sliders default (ie 5). I did have both pitch speeds at 10 also, which doesn't affect cpu vs cpu. Visual only.



Like I said, I certainly did not expect these results but the data speaks for itself and is what I'm locking in and continuing my MLB20 cpu vs cpu Bacon's roster journey with. The errors were consistently around .43-.5 which is interesting as all sliders are default. It seems as if putting the fielder under stress with faster runners causes more consistent fielding mishaps. You will NOT see batters beating out routine grounders, as evident in the data. You will still see OF's challenge and sometimes throw out runners, especially guys with cannon arms. The hits per game are nearly identical to the six year MLB average *8.52 mine vs MLB 6-yr avg 8.55. You WILL see guys going from first to third, taking the extra base, being aggressive on base paths, and taking advantage of defensive miscues. I just saw a bunt with a runner on first, fielded by 1B who threw to second for force, ball goes into left center, runner advances to 3b, throw back into the IF goes past cutoff and beyond 3b coach, runner on 3b takes off and scores. All on default fielding and throwing sliders.



From my testing, as soon as you start muddling with fielding sliders, homeruns spike to the 1.4+ range with default hitting sliders, and everything starts going out of whack and you get taken down a never ending rabbit hole of cause/effect.



Special recognition note; TrueSims MLB 20 sliders perform very well. I ran them through a full 142 game test, only changing pitcher stamina and manager hook to default. They will give you a good game however in my testing with Bacon roster, offense will be somewhat muted. runs per game around 4.0, slugging around .380. But consistent results.

[Caulfield]

Re: Statistical Analysis | cpu vs cpu | MLB 20 | Bacon Fictional Roster

I always find projects like this fascinating.
also, makes me miss Nomo17k

[jcar0725]

Re: Statistical Analysis | cpu vs cpu | MLB 20 | Bacon Fictional Roster

Wanted to share an update. I've spent the past few months testing cpu vs cpu sliders on MLB 20 w/ Bacon's roster, and after roughly 1,500 games and 44 official slider variations I've landed on a set that gets extremely close to a six year MLB average (2015-2021, excluding 2020).



In the end I went with the third variation I tested, which was actually just a curiosity and one I never expected to produce the stunningly, MLB accurate cpu vs cpu results that it did.



It requires only a single slider adjustment.


This supporting data is from a 142 game sample size which went through each team's rotation nearly twice (a handful of #5's did not go twice).


My Slider Results | (MLB2015-21 avg in parentheses)


BA .247 (.252)


BABIP .300 (.297)


SLUGGING% .403 (.417)


OB% .315 (.320)



BB% 8.51% (8.51%)


SO% 23.1% (21.9%)


G_Number (Runs by way of HR) 42.9% (Tough to find data but ideally want upper 30's%)


SB_ATT .81 (.69)


SB% 79.2% (72.5%)

AB 34.44 (33.98)

R 4.30 (4.53)

H 8.52 (8.55)

RBI 4.22 (4.32)

BB 3.25 (3.17)

SO 8.82 (8.33)

2B 1.77 (1.70)

3B .18 (.17)

HR 1.07 (1.20)

HBP 0.25 (0.38)

SF .19 (.25)

SH .08 (.19)

GIDP 1.0 (.74)

SB_att .64 (.50)

CS .17 (.19)

STARTERS AVG INN PITCHED PER START 5.42 (5.43)

NUMBER PITCHERS PER GAME 4.17 (4.28)

WP .12 (.37)

E .46 = .37 IF, .09 OF (.58)


The sliders I used to achieve the above are:


*BASERUNNER SPEED = 10 (Max)


That's it. All other sliders default (ie 5). I did have both pitch speeds at 10 also, which doesn't affect cpu vs cpu. Visual only.



Like I said, I certainly did not expect these results but the data speaks for itself and is what I'm locking in and continuing my MLB20 cpu vs cpu Bacon's roster journey with. The errors were consistently around .43-.5 which is interesting as all sliders are default. It seems as if putting the fielder under stress with faster runners causes more consistent fielding mishaps. You will NOT see batters beating out routine grounders, as evident in the data. You will still see OF's challenge and sometimes throw out runners, especially guys with cannon arms. The hits per game are nearly identical to the six year MLB average *8.52 mine vs MLB 6-yr avg 8.55. You WILL see guys going from first to third, taking the extra base, being aggressive on base paths, and taking advantage of defensive miscues. I just saw a bunt with a runner on first, fielded by 1B who threw to second for force, ball goes into left center, runner advances to 3b, throw back into the IF goes past cutoff and beyond 3b coach, runner on 3b takes off and scores. All on default fielding and throwing sliders.



From my testing, as soon as you start muddling with fielding sliders, homeruns spike to the 1.4+ range with default hitting sliders, and everything starts going out of whack and you get taken down a never ending rabbit hole of cause/effect.



Special recognition note; TrueSims MLB 20 sliders perform very well. I ran them through a full 142 game test, only changing pitcher stamina and manager hook to default. They will give you a good game however in my testing with Bacon roster, offense will be somewhat muted. runs per game around 4.0, slugging around .380. But consistent results.

Man this is good stuff. I was always convinced that the slider tweaking is out of control.

Sent from my thoughts

[bostonboy_003]

Re: Statistical Analysis | cpu vs cpu | MLB 20 | Bacon Fictional Roster

Would this theory / slider base with speed at 10 as the only change work for ‘21 too? Love this!


Sent from my iPhone using Operation Sports

[bostonboy_003]

Re: Statistical Analysis | cpu vs cpu | MLB 20 | Bacon Fictional Roster

Also what difficulty are these for? All star?


Sent from my iPhone using Operation Sports

[bakersville123]

Re: Statistical Analysis | cpu vs cpu | MLB 20 | Bacon Fictional Roster

Also what difficulty are these for? All star?


Sent from my iPhone using Operation Sports

Since I play (and did all of this testing) cpu vs cpu, the difficulty doesn’t really matter. Its dynamic with the slider at zero. I did not test these user vs cpu as I don’t play that way.

As for would this work on 21’, someone would have to do the same data driven approach as I did, as you need ample sample size to draw conclusions. Each version of the game is a bit different under the hood, although they share many similarities.