Ode to Angelfan1961 (A Lorenzen Love Story)

[Inside Pitch]

Most people here will likely want to skip this, it's stat nerd heavy, but if you are interested in stuff like that you'd be well served to read this excellent article at FG...

https://library.fangraphs.com/pitching/stuff-location-and-pitching-primer/

This post leans heavily on the information in this article but for anyone not wanting to read the long form or that doesn't want to have to look at a bunch of graphs you can basically focus on these three excerpts/summaries on the three major pitch modeling components.

Stuff+

Stuff+ looks only at the physical characteristics of a pitch. Important features include, but are not limited to, release point, velocity, vertical and horizontal movement, and spin rate. A pitcher’s secondary pitches are defined based on their primary fastball — with “primary” defined by usage in an outing — and so are judged by velocity and movement differentials along with raw velocity and movement numbers. The model also includes “axis differential,” a statistic that attempts to describe the difference between the movement expected by spin alone and the observed movement affected by the phenomenon described as seam-shifted wake.

Stuff+ was trained against run values, so even if the research community is divided about how much a pitcher can control weak contact, the model includes an inherent nod to the possibility that they do possess some of that ability.

Location+

Location+ is a count- and pitch type-adjusted judge of a pitcher’s ability to put pitches in the right place. No velocity, movement, or any other physical characteristics are included in the statistic. A breaking ball should go to different parts of the strike zone in 2-0 and 1-2 counts, and Location+ captures that phenomenon. Stringer-based command statistics that attempt to judge what a pitcher was intending to do with each pitch do not add predictive value to those models, so Location+ only looks at actual locations and implicitly assumes the intent is generally the same across the league in certain counts with certain pitches.

Pitching+

The overall model, Pitching+, is not just a weighted average of Stuff+ and Location+ across a pitcher’s arsenal. Rather, it is a third model that uses the physical characteristics, location, and count of each pitch to try to judge the overall quality of the pitcher’s process. Batter handedness is also included in Pitching+, capturing platoon splits on pitch movements and locations.

So the short version of all the above is to TRY to measure movement/pitch type, location/pitch type, and how well a pitcher does at doing both.

If you have ever wondered why a guy that throws so hard gets lit up so much -- pitch modeling data is where you want to look.  If you wondered how guys can be coasting and all of a sudden a single baserunner can wreck it all -- again, pitch modeling is where you want to look.

Years ago I waxed poetic about Zach Wheeler despite what the ERA was because WATCHING him pitch it was obvious he was better than his numbers.  For a couple years I mentioned trading either Adell or Marsh in exchange for George Kirby, again because of watching him throw.  Now thanks to pitch modeling it's easier to argue that Pitcher A, should be better or will do well in the future.,

BTW, Wheeler ranks number 2 in MLB in Pitcher+, Kirby 3rd among MLB qualifiers.

So what does this have to do with Lorenzen?

Last we saw him, he was getting lit up the second half of 2023 -- he ended the season with an ERA over 5.50 after joining Philly and was dropped from the rotation -- awful except maybe it was just fluke results, fatigue, or simply bad pitch selection after swapping teams.

Stuff + is again the weighted results of every pitch thrown by every pitcher.  

Stuff+ Averages/Standard Deviations

Pitch Type	Average	Standard Deviation
Four-Seam Fastball	99.2	18.3
Changeup	87.2	16.4
Curveball	105.5	16.8
Cutter	102.1	14
Knuckle Curve	110.3	16.4
Sinker	92.5	13.6
Slider	110.8	15.6
Split-Finger	109.6	30.2

Lorenzen pitches graded out like this:

All FB types 98 (above average), Sinker 82 (below average), cutter 91 (below average) Slider 116 (above average), Curve 104 (above average)*, Change 95 (above average).  There seems to be a disconnect between what the stringers are calling a cutter and what Baseball Savant sees as a cutter because they only show him throwing 22 of them and that's not adding up.  it's likely that his sinker is being mistaken for a cutter and it's driving that grade down.

The interesting thing about the info above is that if you look at his location+ data you'll see he's been doing an even better job of locating his pitches.. Location is the same for everyone and so -- 100 is league average regardless of the pitcher.  All FB types 100, Sinker 108, Cutter 66, Slider 102, Curve 111, Change 101.  This is where you can start to pick up how good the pitch has been for him.

Pitching+ Averages/Standard Deviations

Pitch Type	Average	Standard Deviation
Four-Seam Fastball	98.1	8.2
Changeup	98.7	8.4
Curveball	103.9	7.2
Cutter	98.6	6.2
Knuckle Curve	104.5	7.2
Sinker	95.4	6.7
Slider	106	6.9
Split-Finger	107.6	10.3

This is probably the most important category -- what he's been able to do when you look at the quality of his stuff and how he's located it...

All FBs 98, Sinker 100, Cutter 62, Slider 106, Curve 110, Change 102 -- don't look now but that sinker/cutter isn't helping him -- but is it even the sinker???

Basically Lorenzen has four quality pitches Four Seamer, Slider, Curve, Change but there is some question what that fourth quality pitch may be.

These are the averages allowed for the individual pitch types at Baseball-Savant.

Pitch type -- wOBA (number of pitches)

Four Seamer - .289m (794) 
Slider - .282, (525)
Change up - .270, (489)
Sinker - .338, (277)
Sweeper - .427 (142)
Curveball - .962 (28)
Cutter - .742 (22)

When you compare his averages allowed from 2022 to last year it seems likely that there there is a large disconnect between what human stringers are charting .vs what statcast is charting.  It seems likely that the stringers are getting the sinker wrong as the averages allowed on it have actually been extremely consistent while the sweeper has seen pretty big swings year to year.

Whatever.   The point is Lorenzen's actually pitched better than it may seem and most of the predictive data likes him.  Last year was his first time throwing more than 97 inning since 2015, he may have just been gassed.  If it's a tunneling issue that can be fixed, if it was fatigue then one can hope another year of added innings means he can keep pitching well longer.  But he's a relatively fresh arm career wise and he's got quality stuff.

Plus if the Angels did sign him -- we'd see a lot less spam from our friend, Fredo.

[jeskola]

8 minutes ago, Inside Pitch said:

Most people here will likely want to skip this, it's stat nerd heavy

you're right

[Inside Pitch]

1 minute ago, jeskola said:

you're right

I normally am.  

[totdprods]

I like that he also wants to be here, or at least he did last winter. He's still fairly early in his career as a SP, so wouldn't be surprised if there was another level to unlock in there. Totally cool with the Halos bringing him back.

[DCAngelsFan]

Interesting to me is the use of the term "trained" with regards to the model.

Applying machine learning in this way is a way to understand "what" happened - but it's a very short step to using models in real-time to make predictive pitch selections - factoring/weighting what the pitcher is good at (that day), what the hitter is bad at, against all pitchers, similar pitchers, and this pitcher, weighing count, pitch sequence, time of day, game situation, his bat speed today, what the last pitch was/result, what the desired outcome is (strikeout, popup, ground ball), what he had for breakfast - so, not static pitch charts, but real-time. 

If it's not being done yet, it will be soon.  

Sure, the Luddites will roll their eyes, but this will increasingly be an advantage to those that make use of it, "increasingly" as the models get better, and are fed more data.  

 

 

[T.G.]

17 minutes ago, DCAngelsFan said:

Interesting to me is the use of the term "trained" with regards to the model.

Applying machine learning in this way is a way to understand "what" happened - but it's a very short step to using models in real-time to make predictive pitch selections - factoring/weighting what the pitcher is good at (that day), what the hitter is bad at, against all pitchers, similar pitchers, and this pitcher, weighing count, pitch sequence, time of day, game situation, his bat speed today, what the last pitch was/result, what the desired outcome is (strikeout, popup, ground ball), what he had for breakfast - so, not static pitch charts, but real-time. 

If it's not being done yet, it will be soon.  

Sure, the Luddites will roll their eyes, but this will increasingly be an advantage to those that make use of it, "increasingly" as the models get better, and are fed more data.  

 

 

Using ML in my line of work is becoming more and more prevalent.  I keep thinking someone is going to us it for predictive data in MLB, if they aren't already.  Any time you do predictive modeling without ML, you're limited by the number of data points you can use and when you have to pick data points, the model is prone to bias.  ML eliminates the bias (for the most part) because it's not limited by the number of data points you can use. 

 

[Angel Oracle]

IP has peaked my interest somewhat in Lorenzen.    

I could see a 2024 rotation of Detmers, Canning, Sandoval, Lorenzen, and Anderson.    Then Silseth can use at least the first half of 2024 in AAA to build up innings.

Then by 2025: Burnes, Detmers, Canning, Sandoval, and Silseth; with Anderson traded and possibly Dana in waiting?

[Inside Pitch]

7 minutes ago, Angel Oracle said:

IP has peaked my interest somewhat in Lorenzen.    

I could see a 2024 rotation of Detmers, Canning, Sandoval, Lorenzen, and Anderson.    Then Silseth can use at least the first half of 2024 in AAA to build up innings.

Then by 2025: Burnes, Detmers, Canning, Sandoval, and Silseth; with Anderson traded and possibly Dana in waiting?

The thought of Daniel, Silseth, Bachman, even Rosenberg, all waiting in the wings is a welcomed sight after a near decade of counting on reclamation projects and 4A types for the actual rotation.  

 

[Angel Oracle]

8 minutes ago, Inside Pitch said:

The thought of Daniel, Silseth, Bachman, even Rosenberg, all waiting in the wings is a welcomed sight after a near decade of counting on reclamation projects and 4A types for the actual rotation.  

 

I should have at least included Daniel in that next group with Dana.

Still waiting for Bachman to stay healthy.

[T.G.]

16 minutes ago, Angel Oracle said:

IP has piqued my interest somewhat in Lorenzen.  

If you're going to represent "The Beach."

[Angel Oracle]

Just now, T.G. said:

If you're going to represent "The Beach."

Yep, my version of your/you’re

[Taylor]

Very interesting thoughts. I didn't read more than the first sentence though.

[ettin]

@Inside Pitch NERD.

[AngelsLakersFan]

8 hours ago, DCAngelsFan said:

Interesting to me is the use of the term "trained" with regards to the model.

Applying machine learning in this way is a way to understand "what" happened - but it's a very short step to using models in real-time to make predictive pitch selections - factoring/weighting what the pitcher is good at (that day), what the hitter is bad at, against all pitchers, similar pitchers, and this pitcher, weighing count, pitch sequence, time of day, game situation, his bat speed today, what the last pitch was/result, what the desired outcome is (strikeout, popup, ground ball), what he had for breakfast - so, not static pitch charts, but real-time. 

If it's not being done yet, it will be soon.  

Sure, the Luddites will roll their eyes, but this will increasingly be an advantage to those that make use of it, "increasingly" as the models get better, and are fed more data.  

 

 

The model needs about 250 pitches to become more predictive than any forecasting model or statistic. 

Impressive but a long ways from the 4 or 5 pitches I need to see before I would start screaming at Maddon or Nevin.

[T.G.]

1 hour ago, ettin said:

@Inside Pitch NERD.

In the Chris Knight/Real Genius mode.

[tomsred]

4 hours ago, Taylor said:

Very interesting thoughts. I didn't read more than the first sentence though.

Attention deficit disorder (ADD) alert.

[Tank]

16 hours ago, T.G. said:

In the Chris Knight/Real Genius mode.

IYKYK

[Angel Oracle]

1 hour ago, Tank said:

IYKYK

[Slegnaac]

But how is he as a DH?

[Mark68]