DIPS Engine ?

[andymac]

Originally Posted by Elendil

Not a lot of such players, maybe, but off the top of my head Ken Phelps and Dave Kingman were both low-double, high-homer guys:

http://www.baseball-reference.com/p/phelpke01.shtml

http://www.baseball-reference.com/k/kingmda01.shtml

I'd be interested to see wha their GB/FB numbers were. I'd bet they both hit the ball in the air quite a bit.

[sebastian0622]

I'm guessing the problem is in the way pitchers' ratings are created. That is to say that they seem to be independent of each other. That is fine for stuff and control, but I believe that movement should be correlated to both, primarily control. Movement could be distributed on an approximately normal curve of an equation based on stuff and control (ex. movement = .30*stuff + .70*movement) with a standard deviation of, say, 8 or so. (Just pulling numbers out of my @$$, they would need to be based on real life regression analysis before implementing). This way, you'll still see variation but you won't see completely unrealistic guys like yours. A pitcher with ratings of 75/84/57 would be 3 standard deviations from the mean (of 75/84/81), for instance. So that would be, practically, the limit of what you'd see on a semi-regular (.15% of pitchers, or one in about every 650 created pitchers) basis. Going the other way, a 75/84/105 would be a rare sight but would still pop up .15% of the time.

So you correlate movement to the other ratings, then you fix the g/f issue to add in a little more character and variation, then you have a realistic solution.

The first thing we have to do, though, as I mentioned very early on in this thread and was also mentioned by TonyJ, is to find out exactly how HR/9 is correlated to K/9 and BB/9 rates.

Oh, on top of all of that, BABIP should probably be based on control and movement. It is currently based on stuff. So we'd have to run regression on RL pitchers for that too. We might as well fix everything

[dougaiton]

Originally Posted by Elendil

Not a lot of such players, maybe, but off the top of my head Ken Phelps and Dave Kingman were both low-double, high-homer guys:

http://www.baseball-reference.com/p/phelpke01.shtml

http://www.baseball-reference.com/k/kingmda01.shtml

Indeed, there are more than a few, and certainly they aren't as rare as the pitching exceptions - plenty of guys delineate from the default. However, the problem with OOTP player generation is that there are no defaults - players are random modelled.

[BigCity]

Originally Posted by dougaiton

Indeed, there are more than a few, and certainly they aren't as rare as the pitching exceptions - plenty of guys delineate from the default. However, the problem with OOTP player generation is that there are no defaults - players are random modelled.

However, the whole issue of player generation can't be answered ONLY through comparisons of modern day players. The game generation, as defined today, is "diluted to produce "believable" characters throughout 100 years of history. To get the accuracy being discussed here, player generation has to be variable by Era.

[kingcharlesxii]

Originally Posted by BigCity

However, the whole issue of player generation can't be answered ONLY through comparisons of modern day players. The game generation, as defined today, is "diluted to produce "believable" characters throughout 100 years of history. To get the accuracy being discussed here, player generation has to be variable by Era.

Hmm that's a really good idea. An era setting before initial player generation.

[Ktulu]

There is an interesting article today at the Hardball Times which is relevant to some of the disscussion in this thread.

[dougaiton]

Originally Posted by BigCity

However, the whole issue of player generation can't be answered ONLY through comparisons of modern day players. The game generation, as defined today, is "diluted to produce "believable" characters throughout 100 years of history. To get the accuracy being discussed here, player generation has to be variable by Era.

Exactly. That's a very prescient point.


However, OOTP does throw up a few shapes that we might generally consider to be impossibilities - like that chap who walked under 20 times but hit 70 homers in that other thread. There are certain skillsets that being good at one thing necessarily means being at least okay at something else. This is of course more pronounced in different eras, but there's little doubt in my mind that whoever is the homerun leader, be it Frank Baker or Stan Musial or Yaz or Cecil Fielder, should have decent ability at getting doubles.

Again, this is all for 7, and I think the two things together - better 'generational' player creation, and more realistic 'norms' to deviate from - can help answer a lot of the problems we are hitting now, pitcher stats, age etc.

[rudel.dietrich]

Originally Posted by kingcharlesxii

Hmm that's a really good idea. An era setting before initial player generation.

You can do this already by adjusting the creation modifiers.

What there needs to be is the ability to 'link' ratings.
i.e. you can make it so that 90% of all generated players when they are created with a '5' power rating automaticly get a '5' or higher gap rating.
All players with a '8' power rating or higher automaticly get a '5' or higher eye rating.
etc


I would redo pitching ratings altogether

First the game needs to treat individual pitches as individual pitches.
Secondly I would get rid of two of the ratings we have now.
I would rate the pitches in the following area.

Max velocity
Min velocity
Control
Effectivness (replacing 'stuff') this would also work as a movement rating since how effective a breaking pitch is, is how well it breaks.


Each pitch would be rated in these 4 areas. All 4 would come together to form an overall rating on each pitch for easy reference.
Then all of the overall pitch ratings would come together to form an overall pitching rating.

One final rating not tied to actual throwing would be 'pitching IQ'
Alot of people will flame me for this idea since it can't be quanitfied, but there is something to be said for knowing when to throw what pitch.
I would have a 1-10 rating on how well a pitcher studies hitters and teams and known when to throw what pitch. That would be the 'pitching IQ'.


I have never understood the movement rating. I find it completly inane. If you have a high stuff rating and you are a finese pitcher you are obviously going to gave good movement of your breaking balls.
Stuff as far as I understand is how good your pitches are. If you have high stuff and throw a curve then it is simple logic that that curve is going to break well. That is what a curve does.
Movement to me is a totaly uneeded ratings.


So using my system EBH hits would be tied to all the ratings instead of one rating controling one aspect.
First the game would take the velocity of the pitch into account, then it would take the pitchers effectivness rating and the pitchers control rating into account and weigh that against the hitters ratings.
So if a power pitcher tries to throw a curve ball (in which he has a bad rating in) to a good cureball hitter (his pitching IQ determened that he would select that pitch to throw) and he put it in a bad location, and it had something extra on it. HR

The games hitting ratings are spot on.
But I would add addtional ratings to determine how good a hitter is at hitting certain pitches.
Some hitters are good at hitting fastballs some at hitting the slider, etc.


Oh, one last thing. I would allow the ability for pitchers to learn new pitches in the minor leagues. And the GB/FB% would dynamicaly change as new pitches were added.
If a pitcher picks up a good sinker then expect the GB% to shoot up 25% or something along those lines.



Thoughts?

[El Cid Part II]

Most of what you guys are saying is over my head
But I do have MLB for the PS2, and it has each pitch with 3 characteristics
Velocity
Accuracy
Break

Plus you have Clutch, Hits Allowed which I do not like and HR Allowed

I think that works well

It be nice if pitches were more developed in OOTP

[Crapshoot]

Originally Posted by Aadik

Hmm- will check his XBH when I get back home...

Update: he has given up 24 doubles and 9 triples, which may well explain a lot of that. As always, the OOTP PBP tells us nothing actually about what happened. Today's game is similarly frustrating - he has pitched 7 innings, 1 walk, 5 K's, and 8 hits, including 2 solo HR's.

Ratings fyi:

Stuff: 65
Control: 111
Movement: 30

[draven085]

Can you check some of his teammates' EBH/9 IP numbers against his? I'm curious if pitchers with better movement ratings are allowing as many doubles and triples as he is. As it is I'd say that the low movement rating is directly responsible for dragging his performance down.

[Crapshoot]

Yeah, I was always under the impression that movement correlated to HR allowed, and Milwaukee is a hitter's park. I'll check when I get back. The more I think about it, the more the GB% and the high HR rate together bother me though- that should be an inverse correlation.

[Joshv02]

Ok, so I can’t find any studies out there (though I’m sure they exist) that show if SOr or BBr or SO/BB have any correlation to HR rate. (Great article, btw, from Studes; thanks, Ktulu).

So, I took the Lahman database for every season of every pitcher with >200 BFP since 1955 (when we track IBB). I had 12,291 seasons of data.

I added some columns: HR/IP; HR/H; HR/BIP*; and, HR/BFP.

By HR/BIP, I do not actually mean Balls in Play, as HRs are not in play. I mean HR/(BFP-SO-BB-HBP).

I also added in (BB-IBB)/BFP and SO/BFP.

I then ran a simple Correl test in excel. I got the following numbers:

SO/BB	         BBr	         SOr	 
0.023362287	0.1312753	0.1048	HR/H
-0.080309341	0.0891128	-0.0906	HR/IP
-0.025859706	0.0368653	-0.0656	HR/BFP
0.011514666	0.1339985	0.0938	HR/BIP

Now, I learned as much math as a college freshman, and I know as much math as a high school freshman. So, I can’t do much to interpret these. I’d guess that HR/BIP and BBr have the highest correlation, but it still is fairly weak. I.e., I don't think there is much of a correlation at all in there.

I don’t know what OotP does. If anyone wants my sheet to really look at the data, feel free to PM me.

[IatricSB]

Originally Posted by BigCity

I can understand not liking it, but as long as it's an option, I wouldn't say it's the worst thing that ever happened. It'll be interesting to see if Markus stays with the option idea in V7 or goes 100% DIPS though.

Yeah, I went over the top with that. But it's not something I've ever been completely sold on.