ZiPS and OOTP Ratings Generation

[hjrrockies]

I have a question about how the developers implement ZiPS projections in ratings generation. Here's what I understand so far:

1. ZiPS projections are NOT park-neutral
2. ZiPS projections form the baseline for in-game ratings
3. The simulation engine is essentially a random number generator, weighted by player ratings
4. The simulation engine is also weighted by the current ballpark

What I want to know: do the player ratings use a park-adjusted version of ZiPS? Because, if not, that would have a drastic effect on the ratings for Rockies hitters and pitchers. To illustrate:

Suppose the "Stuff" rating corresponds roughly to:

70 ~ 10.0 K/9
65 ~ 9.5 K/9
60 ~ 9.0 K/9
55 ~ 8.0 K/9
etc.

In game, Jon Gray has a 60 Stuff Rating. ZiPS projects him have a ~9.0 K/9, including Coors Field. Meaning, ZiPS thinks his park-neutral, true-talent strikeout rate is nearly 10 K/9 (or a 70 rating). Since Jon only gets a 60, when the game actually simulates, his strikeout rate is usually considerably less than 9.0 K/9, because of Coors Field.

I would love to hear from the developers on this issue, and would hope there's a good way it could be resolved so my Rockies playthroughs don't involve an offense scoring 7 runs a game, but a team ERA in the high 5's. Even though the net effect means the Rockies win as many games as expected, the splits are more extreme than in real life.

[Mat]

I sense a Rockies fan.

[Lukas Berger]

I don't believe we use a park adjusted version of ZIPS, but we do try to manually adjust the ratings to an extent to adjust for issues like this.

It's possible Gray's stuff is in fact a bit too low as of now, I'll take a look and tweak that a bit if needed.

[Lukas Berger]

I don't think this is a systemic issue with the Rockies though.

f.e. in my last sim, they scored 762 runs and allowed 826.

That's only one sim but it's very much in-line with their recent performances in reality and what they could reasonably be expected to do this year: http://www.baseball-reference.com/teams/COL/

I'll take a look at some other sims to see what their results look like, but it certainly doesn't seem from what I'm seeing now as if the Rockies results are out of line with reality with too many runs being scored on either side.

They look pretty solid overall.

[X3NEIZE]

On that note, and I hate doing this...

Masahiro Tanaka ratings are a bit on the low end... his stuff is way above what he shows in game....I know i'm a Yankee fan but i'll also say that Carter is way off and on the overated side..

[Lukas Berger]

Originally Posted by X3NEIZE

On that note, and I hate doing this...

Masahiro Tanaka ratings are a bit on the low end... his stuff is way above what he shows in game....I know i'm a Yankee fan but i'll also say that Carter is way off and on the overated side..

My last test sim Tanaka had a K9 of 7.7. He K9'd 7.4 irl last year.

Sim before the last Tanaka had a 7.8 K9

I'd say that pretty much spot on, no?

[Lukas Berger]

On Carter, last sim had him at .186/.280/.395 Sim before the last he was at .209/.294/.529. Hit 49 HR's but only had a 1.5 WAR.

Last year irl he hit .222/.321/.499

Obviously it's just one sim, and the results are different each time, but I haven't seen anything too far off with either guy except that maybe Carter is a bit too bad.

[Lukas Berger]

To do the same with Gray as well, my last two sims had him at 9.2 K9 and 7.1 K9.

So I do agree there might be an issue there, and I've bumped up his stuff a bit for next build.

[X3NEIZE]

Originally Posted by Lukas Berger

My last test sim Tanaka had a K9 of 7.7. He K9'd 7.4 irl last year.

Sim before the last Tanaka had a 7.8 K9

I'd say that pretty much spot on, no?

I haven't sim'd i was just going by scoutung reports and atributes... but if that's the case, looks about right. Just seems a bit off with his 60 stuff when compared to Pineda for example... on paper and in game Pineda is the better pitcher, when we clearly know that's not the case.

[Lukas Berger]

Originally Posted by X3NEIZE

I haven't sim'd i was just going by scoutung reports and atributes... but if that's the case, looks about right. Just seems a bit off with his 60 stuff when compared to Pineda for example... on paper and in game Pineda is the better pitcher, when we clearly know that's not the case.

Pineda shouldn't be better on paper or in-game. Do you have scouting turned on maybe? If so that throws everything off.

That being said, they may be a bit too close, but the issue is probably more that Pineda is a teeny bit overrated.

[X3NEIZE]

Originally Posted by Lukas Berger

Pineda shouldn't be better on paper or in-game. Do you have scouting turned on maybe? If so that throws everything off.

That being said, they may be a bit too close, but the issue is probably more that Pineda is a teeny bit overrated.

is it on by default? if so i do have it on as this year I haven't touched anything... all default..

If you're looking at the Yankees, the Girardi should always pick Chapman as a closer... doesn't happen in the Beta, he seems to always go for Betances....

[Lukas Berger]

Originally Posted by X3NEIZE

is it on by default? if so i do have it on as this year I haven't touched anything... all default..

If you're looking at the Yankees, the Girardi should always pick Chapman as a closer... doesn't happen in the Beta, he seems to always go for Betances....

It is on by default.

Chapman starts as the closer. What happens as the game goes on is harder to control.

[hjrrockies]

Thanks a ton Lukas! One thing I noticed was that the Rockies, as a team, are actually quite on-point with strikeout rates (Gray aside) and homerun rates, but their BB rates are much higher than I expected. In 3 full-season sims, the Rockies posted BB/9 rates of 4.08, 3.96, and 4.02.

You guys do a great job, and I don't want to pile on more work than is worth your time, but it might be worth a quick pass at BB rates for the Rockies, depending on what you see on your end.

Thanks!

[hjrrockies]

Originally Posted by Mat

I sense a Rockies fan.

I very much am

[X3NEIZE]

Originally Posted by Lukas Berger

It is on by default.

Chapman starts as the closer. What happens as the game goes on is harder to control.

That's not happening... I tried three different saves, when I start the save Bentances is ALWAYS the closer and Chapman the setup guy... I can send you the save...

[wallewalls]

Originally Posted by X3NEIZE

That's not happening... I tried three different saves, when I start the save Bentances is ALWAYS the closer and Chapman the setup guy... I can send you the save...

I started a Yankees save and Chapman was the closer with Betances and Holder I believe as the set up guys

[X3NEIZE]

Originally Posted by wallewalls

I started a Yankees save and Chapman was the closer with Betances and Holder I believe as the set up guys

Let me guess, you were Manager? Because as GM only it seems to always pick Betances... either than or my luck sucks.

[NoOne]

you guys should remember this is a mostly automated process and uses only objective facts and not all the info that a fan may have about a home town player. (some of which being most defintely false, lol.. humans err)

they typically do make some adjustments, but don't expect them to meet any "homer's" feelings.

if you feel something is off, then change it. i'll change velocities if i see different info on fangraphs or something... i try to base it on RL fact, but i deviate from that on occasion.. that's certainly not a requirement... its' a video game.. make it as you want it.

J.D. Martinez is a perfect example for me. OotP's version of this player was, in my opinion, underrated the last few years (i think they finally beefed him up in '17). by coincidence the changes i made to him were almost an exact mirror of what ootp eventually beeefed him upto... a guy that can hit 30+hr with regularity and K's 150-200times a year. high babip, low k's... makes sense for him.

they weren't wrong before... and i certainly wasn't right even though he's proven me right since. i choose one of many possibilities and ran with it... at that time any could have been correct, therefore they are all correct at that time. it's simply incredibly hard to speculate about these things. change is constant... there are likely as many overrated and underrated and only time will show which is true.

if you tracked individual gut-feelings and compared to zips on all dispted players over the next few years, i'd put my money on the zips projection doing better than most human eye-tests per player... and nearly 100% certain better in an overall sense.

(some may know their local players better, if they can also strip emotion from the evaluation. i.e. jd actualyl made soem significant changes to his approach and his swing mechanics - unfortuanltey, these combinations of words are used all the time by organizations about struggling players with absolutely no change in results, too. he showed a consistent shift in his statistics, though... and as time bore on, it became obvious that it was a real shift)

[hjrrockies]

Saw that Jon Gray got a stuff buff in the update! Thanks!

[Lukas Berger]

Originally Posted by hjrrockies

Saw that Jon Gray got a stuff buff in the update! Thanks!

Np, thanks for the suggestion.