Stats-based sliders for CPU

[nomo17k]

Re: Stats-based sliders for CPU

That attached file dictates a test using only 10 fielders on a single team. It is highly possible this team selected was a particularly slow team (ie last in team SB's), or that they just chose a "poor" set of fielders.

Yes that's a valid concern, but with quick googling we find some pieces of (not so reliable) info that time from contact to 1B is generall 4.3 sec for RHB, 4.2 sec for LHB, or something close like that. That appears to be in line with what scouts use when they time HS/college prospects, etc. So it's good enough for a rough guesstimate.

That test also states the timer was started on contact, not at bat drop. The in-game animation sequence between when a batter makes contact, and when they start running - ideally we would have to be certain that this animation is accurate, time-wise, and I highly doubt this to be the case.

It's not really accurate in game I think. But IRL what scouts want to measure is not necessarily how fast batter runners run, but how fast till they reach 1B (to avoid GIDP, increase infield hits, etc.), so I think that's what matters to them. And basically that's what I tried to imitate here.

Your use of a median speed as a reference point also relies on the assumption that in-game, every single player has an accurate speed rating.

Right that's an assumption, but we have to make one to proceed from somewhere. As a reference point, it's not a bad assumption. Not sure what's a better one to make.

As a side note, I think increasing BR speed can work wonder... I've been playing an RTTS player named Brett Butler, an Ichiro-type speed star, but having had problem getting infield hits, avoiding DPs, stealing bases, etc. Those things now happens quite noticeably now that I've been playing him with BR speed increased by a couple clicks.

My worry of doing this for the CPU sliders is balancing XBHs because I definitely want to keep fielder speed low, somewhere 0 - 3 range. Will see what happens.

I saw this thread a while ago and meant to post in here then but honestly I forgot . Nonetheless, I think this is an amazing topic, and look forward to following your research.

Bobhead, I never meant to do research. Just a way for nerds to have fun Friday evenings, while girls gone wild elsewhere.

[Bobhead]

Re: Stats-based sliders for CPU

How does the difficulty level affect CPU vs CPU games? I'm starting a CPU vs CPU season myself just out of curiosity and I currently have it set to All Star batting and HoF Pitching, and was wondering what this would mean for CPU vs CPU games.

[nomo17k]

Re: Stats-based sliders for CPU

How does the difficulty level affect CPU vs CPU games? I'm starting a CPU vs CPU season myself just out of curiosity and I currently have it set to All Star batting and HoF Pitching, and was wondering what this would mean for CPU vs CPU games.

The common knowledge is that the difficulty level doesn't matter in CPU vs CPU games. (If it does, it probably is a bug...)

When you think about it, it should not matter, because if the levels are different with pitching and hitting, you don't know which one to use right... Say I use Legend pitching but Rookie hitting. Rookie CPU hitters should be facing pitchers with Rookie AI (more hittable strikes), but Legend CPU pitchers want to throw like 15 year veteran of the Show, facing Legend hitters.... Not gonna happen.

Way to reconcile that dilemma is to not use difficulty levels at all...

[nomo17k]

Re: Stats-based sliders for CPU

A couple updates. I'm adjusting speed related sliders for better visuals and variety of plays. Another issue is a high rate of outfield assists. This is something DaiYoung from the CPU vs CPU thread alerted us before.

This set is done mostly before patch 1.16, only a few games are post patch.

Everything at default (5) except:

Power: 4
Foul Frequency: 4
Pitcher Consistency: 2
Strike Frequency: 4
Pitcher Hook: 2
Pitch Speed: 10
Fielding Error: 10
Throwing Error: 5
Fielder Arm Strength: 0
BR Steal Ability: 8
BR Steal Frequency: 8
Wind: 3

Comment: I lowered the fielder arm strength to zero in hope of reducing outfield assists, which it didn't sufficiently.

A/INN = 0.01328 (0.00644)

So outfield assists still occur almost twice the rate in real life! For comparison, at default it was A/INN = 0.01075, so assists actually increased! Do I believe that lowering outfield arm produces more assists? Not really, but perhaps it is safe to say the fielder arm strength doesn't have a strong effect to reduce outfield assists at least.

Since I'm adjusting gameplay speed, I also want to keep XBH fractions in check. The MLB averages in parentheses.

2B/H = 0.206 (0.203)
3B/H = 0.023 (0.021)
HR/H = 0.138 (0.111)

So these are still okay, even with fielder arm strength minimized.

I also lowered pitcher hook so that CPU uses roughly 3.92 pitchers in a game. At default, CPU uses slightly more pitchers. Looking good there.

[nomo17k]

Re: Stats-based sliders for CPU

This set is done after the patch 1.16.

Everything at default (5) except:

Power: 4
Foul Frequency: 4
Pitcher Consistency: 2
Strike Frequency: 4
Pitcher Hook: 5
Pitch Speed: 10
Fielding Error: 10
Throwing Error: 5
Fielder Run Speed: 0
Fielder Arm Strength: 0
BR Steal Ability: 8
BR Steal Frequency: 8
Wind: 3

Comment: I lowered the fielder run speed as well as fielder arm strength to zero. Here is the assists per inning:

A/INN = 0.00682 (0.00644)

So outfield assists are still a tiny bit inflated, but much closer to real life... maybe outfielders being slower leads to a lot less close plays at third...

Since I'm adjusting gameplay speed, I also want to keep XBH fractions in check. The MLB averages in parentheses.

2B/H = 0.228 (0.203)
3B/H = 0.036 (0.021)
HR/H = 0.106 (0.111)

So doubles and triples are now inflated, as expected. (HRs are actually down from before, likely caused by the reduced wind effect by the recent patch.) The number of outfield assists are now reasonable, but there are a bit too many doubles and triples. Perhaps I shouldn't lower the fielder run speed too much to avoid this issue. However, a good side effect of reducing that slider is that I see a lot more border line plays in the outfield, which leads to more play variety. It's fun. So I want to keep it around 0 - 3 range, if I can keep XBH fractions reasonable.

Pitcher hook slider is set to default for this set. Upon reflection this wasn't necessary but the patch fixed the closer being pulled too early glitch, so I was thinking the CPU reliever usage might have changed (which I don't think changed in general... just closer).

[nomo17k]

Re: Stats-based sliders for CPU

I lowered the base runner speed as well for this set. (Pitcher hook is back to 2.)

Everything at default (5) except:

Power: 4
Foul Frequency: 4
Pitcher Consistency: 2
Strike Frequency: 4
Pitcher Hook: 2
Pitch Speed: 10
Fielding Error: 10
Throwing Error: 5
Fielder Run Speed: 0
Fielder Arm Strength: 0
BR Speed: 0
BR Steal Ability: 8
BR Steal Frequency: 8
Wind: 3

Comment: What's the outfield assists now?

A/INN = 0.00699 (0.00644)

This seems good enough. Slightly increased compared to the last set, perhaps within the range of random fluctuation though. So, maybe if I keep the fielder run speed sufficiently low, it is possible to reduce outfield assists to a reasonable level after all without doing anything else...

Since I'm adjusting gameplay speed, I also want to keep XBH fractions in check. The MLB averages in parentheses.

2B/H = 0.223 (0.203)
3B/H = 0.022 (0.021)
HR/H = 0.106 (0.111)

Doubles are still up, but triples are right on the MLB ave. So fast base runners really help to increase triples, but doubles are caused more by the speed of outfielders. (HRs are still down; the game isn't as power friendly as before the patch apparently.)

Also note the very high caught stealing! BR speed does have a significant impact on the steal success as well.

These minute details are kinda good, because we have two independent ways of adjusting double & triple fractions. The only problem is that I actually want to *raise* BR speed, because it is closer to real-life speed and also it created more close plays at 1B, reduce GIDP, etc.

[nomo17k]

Re: Stats-based sliders for CPU

Another thing is the patch apparently over-corrected the WPs. Pre-patch, a WP happened as frequently as a HR, but now it's a rare event. Passed balls don't occur really (only once after ~ 200 games).

Also, I may be seeing too many high-BB games (like 6, 7 walks)... I don't think the patch has anything to do with it, but now I want to see if I can click up strike frequency or pitcher consistency slider by one to see if it maintains ~ 3.3 BB... I actually want the early-count strike % to be consistently in 59% range (which isn't lately).

[bedwardsroy19]

Re: Stats-based sliders for CPU

Hey Nomo,

This might be a dumb question thats already been answered but I am curious, by CPU vs CPU are these only effective in games that are CPU vs CPU that you watch, or are they effected when simming a season as well?

[nomo17k]

Re: Stats-based sliders for CPU

Hey Nomo,

This might be a dumb question thats already been answered but I am curious, by CPU vs CPU are these only effective in games that are CPU vs CPU that you watch, or are they effected when simming a season as well?

Only injuries, stamina, and manager hook sliders affect simmed games.

So if you want to make changes to sliders in hope of changing the game balance in simmed games (i.e., simulating "dead ball" era by reducing power to zero or steroid era by increasing it, etc.), I don't think you can.

I did something similar with simmed game stats, and they appear to be tuned to be in line with the recent MLB ave:

http://www.operationsports.com/forum...post2042365425

[bedwardsroy19]

Re: Stats-based sliders for CPU

Only injuries, stamina, and manager hook sliders affect simmed games.

So if you want to make changes to sliders in hope of changing the game balance in simmed games (i.e., simulating "dead ball" era by reducing power to zero or steroid era by increasing it, etc.), I don't think you can.

I did something similar with simmed game stats, and they appear to be tuned to be in line with the recent MLB ave:

http://www.operationsports.com/forum...post2042365425

Darn, that would have been convenient!

Thanks Nomo

[bubs3141]

Re: Stats-based sliders for CPU

Another thing is the patch apparently over-corrected the WPs. Pre-patch, a WP happened as frequently as a HR, but now it's a rare event. Passed balls don't occur really (only once after ~ 200 games).

Also, I may be seeing too many high-BB games (like 6, 7 walks)... I don't think the patch has anything to do with it, but now I want to see if I can click up strike frequency or pitcher consistency slider by one to see if it maintains ~ 3.3 BB... I actually want the early-count strike % to be consistently in 59% range (which isn't lately).

Any luck with this. Played a manage only mode using your sliders and the balls were high in the beginning of the game but towards the end it evened out nicely. I think the confidence rating had a lot to do with it. Just wondering if you made any changes to the consistency slider.

[nomo17k]

Re: Stats-based sliders for CPU

Any luck with this. Played a manage only mode using your sliders and the balls were high in the beginning of the game but towards the end it evened out nicely. I think the confidence rating had a lot to do with it. Just wondering if you made any changes to the consistency slider.

I'm wondering about this too... AFAIK the patch only addressed analogy pitching issue and I don't think it should affect CPU vs CPU games, but I feel the game has become a little "walk friendly" recently...

So I'm running my games now with strike freq slider at 5 (default), and walks are coming plenty, so I can suggest you doing this already. If it keeps this rate, I may actually start raising the pitcher consistency slider by one as well (not doing this yet though). I'll keep you posted.

It's puzzling what changed though. It could be that the analog pitching control is related to CPU pitch command. Can't confirm.

[bubs3141]

Re: Stats-based sliders for CPU

ok, will raise consistency back to 5

[nomo17k]

Re: Stats-based sliders for CPU

ok, will raise consistency back to 5

FYI so far I only changed strike frequency, not consistency. From what it looks like that might also work as well though...

[bubs3141]

Re: Stats-based sliders for CPU

I will do both.