Kieran Liming
@KieranLiming
Followers
1,170
Following
388
Media
124
Statuses
407
Blue Jays | Tread Athletics | Macalester Baseball Alum | MS in Statistics
Joined January 2021
Don't wanna be here?
Send us removal request.
Explore trending content on Musk Viewer
Medina
• 387337 Tweets
Walz
• 274488 Tweets
#WWERaw
• 202071 Tweets
#AllEyesOnBangladeshiHindus
• 200172 Tweets
Protecting
• 163095 Tweets
Tati
• 148869 Tweets
原爆投下
• 86927 Tweets
Tucker
• 52704 Tweets
#IRIAMメンテ中のフォロー祭り
• 49844 Tweets
ファミチキ
• 42646 Tweets
篠原ともえ
• 38211 Tweets
Fortaleza
• 35742 Tweets
Cruzeiro
• 32960 Tweets
Aiyuk
• 30606 Tweets
ハムの日
• 30361 Tweets
Nassau County
• 27232 Tweets
Cedric
• 20327 Tweets
Browns
• 18974 Tweets
Wyatt Sicks
• 18694 Tweets
Goyang
• 18553 Tweets
49ers
• 18519 Tweets
#DesafioXX
• 17234 Tweets
はだしのゲン
• 16771 Tweets
過去最大の上昇幅
• 14984 Tweets
おたる水族館
• 12929 Tweets
Bronson Reed
• 12074 Tweets
時給2500円
• 11623 Tweets
Odyssey Jones
• 11266 Tweets
Bray
• 10615 Tweets
Carlyn
• 10552 Tweets
सुषमा स्वराज
• 10159 Tweets
Josh Smith
• 10074 Tweets
Pinned Tweet
My last thread predicted an increase in SSW sinkers in the next few years. This thread will be about how to throw one and physically how this effect is achieved in the context of fastballs.
1
12
223
Really excited to be joining the Toronto Blue Jays as a Minor League Pitching Analyst this upcoming season!
The last year has been an exciting run, and I will always be grateful for those who spent the time to help me grow as a person and as an analyst.
🇨🇦🇨🇦🇨🇦
9
19
331
With Ben Joyce making his MLB debut, I wanted to take some time to talk about arsenal design for guys like Joyce–pronators. In his debut he flashed two of the three pillars of what I like to call the pronator triangle.
9
19
309
With Alex Speas making his MLB deput yesterday, I want to take the opportunity to talk about elite supinators and pitch usage.
Speas usage in his debut:
19 pitches
16 gyros
1 cutter
1 fastball
1 sweeper
The sweeper:
4
17
213
Are sweepers bad now?
Shohei Ohtani returns next year as a DH and I don't think he has much trouble with them.
This a quick reminder that if you have a Stuff model, you might want to be conscious of shifts in the game.
3
8
100
A lot of time is spent talking about good fastballs, and a lot of training economy is used trying to develop them. I want to highlight a couple guys who have productive fastballs and some pitfalls into fastball pitch design.
1
6
92
Having just eclipsed one year working at Tread, I wanted to highlight some non-data accomplishments I have made and some of the coaches that have helped me get there.
This time last year my PR was a sus 91 on a pocket radar. This summer I hit 93.7.
5
7
85
Lateness of break, Is it a myth?
I will take some time to discuss the limitations of 9-parameters philosophy when it comes to pitch movement, and then pine hopelessly for exact pitch location by time data.
3
5
82
With the rise of Stuff+ models I think it is important to understand how boosted tree models work and some of their limitations.
Consider:
Clay Holmes' Sinker at 3 IVB and 6.5 ft Release height and
Camilo Doval's Sinker at 3 IVB at 5.3 release height.
2
5
84
Deadzone fastballs are not a monolith.
The general rule of thumb is 12"-15" of IVB puts a fastball in the "deadzone" where the fastball does not perform.
However, this is not the full story. You can throw a 13" IVB FF and get results. You can also throw a 1" IVB SI and not.
3
8
81
Logan Webb is a prominent and successful adopter of SSW. His career reflects two major themes I predict we will see in the next few years.
1. Seam effect sinkers are going to become more commonplace.
2. Overall fastball usage will continue to decrease.
5
2
66
Throwing from a low arm slot gives a pitcher access to maybe my favorite pitch in baseball.
A variant of the well known Sweeper, the PNW (as coined by known PNW-thrower
@T_Zombro24
) allows pitchers to generate unique profiles with breaking balls.
4
9
54
Velocity to the midline is very important. I have been on the hard gyro train for a while, but for supinators, it can be a staple pitch in an arsenal.
Dane Dunning (the GOAT):
2022: Cutter at 9.3/-1 at 87.1, 11.9% freq
2023: Cutter at 10/-1.5 at 88.7, 23.2% freq
5
5
49
What is the death pitch? (TM
@Alexkachler10
) Is it a hard vertical curveball with little side to side movement or is there more to it?
I will be exploring pitch design from the lens of approach angles to help answer this question.
2
4
49
Speaking of sweepers, let's dispel some of the confusion about the seam effects of this pitch, and how you can throw one.
First you will need supination bias, then you will need a marker.
Sweeper ➡️ A slower slider variation with significant horizontal break (generally, >= 10 inches of glove side movement) (LEFT)
Gyro ➡️ A harder slider variation with minimal horizontal break (generally, <= 3 inches of glove side movement) (RIGHT)
Sweeper craze breakdown 🧵⬇️
7
18
187
3
8
46
As part of an investigation into what makes certain pitches more deceptive than others, I took a look at pitchers that defied the norm. These pitchers, when compared to others with similar release points, have significantly higher vertical breaks on their fastballs.
1
3
37
Leaning into motor preferences is usually the fastest way to develop new effective pitches.
Chris Cortez is an electric arm (PC:
@mattm1302
) with significant supination bias. Our goal was to find him a better fastball shape.
(Sub max pen)
4
9
38
Clickbait aside, we have been seeing a pretty dramatic decline is performance for the internet's most popular pitch. With this in mind, I want to talk about Stuff models and how you train your models.
2
6
30
So whats the solution? Throwing true gyros 90-93. Lean into the motor preferences. He wants to supinate. So let him supinate.
Fastball for show. Slider for dough.
1
0
19
My final tip for mastering the pronator triangle is to throw 100. It'll help!
2
1
19
The first point on the pronator triangle is a high efficiency fastball. The shape of the fastball will be largely determined by your armslot, but as a pronator you benefit from having a fastball that is actually worth throwing.
Think of guys like Strider:
1
0
17
Pitch design principles when working on a time frame:
1) Have a metrics/movement goal in mind
2) New pitches should come relatively easily (lean into motor preferences)
3) Know enough about the forces that act on the ball to make decisions on how to alter grips
1
4
16
Part of the reason he can produce such good results on this “deadzone” fastball has to do with the tight triangles concept coined by
@T_Zombro24
. Batters have to defend against the 4-seam, sinker, and cutter all thrown in the zone often.
3
1
15
And finally if we look at the pitcher's perspective of the previous demo we can see that drawing the little black circle stays on the top of the ball and is visible to the pitcher throughout the throw (which is why it works as a drill).
0
0
14
My hypothesis would be that offender number 1 would be the splitter. A pitch that, like the knuckle ball, has significantly less spin and relies heavily on seam orientation to produce its break.
Come to think of it, I am pretty sure Ohtani throws one.
91 at -3IVB/4HB
1
1
10
The goal with the pronated fastball is to get the maximum total movement possible, generally this is chasing vert. It’s possible to pronate the fastball enough to give it “wrong-way” seam effects and create more lift.
Both Strider and Hader see 0:15 of "wrong-way" seam effects
1
0
11
As a pronator your pitch design space is relatively limited. Generally, you will not be able to throw the flashy supinated pitches such as sweepers or hammer curves with significant movement outside of Q1. Don’t let this discourage you.
Below: Alex Faedo's pronator triangle
1
0
10
Good sweepers will still perform, but the field of sweepers are regressing to their mean performance level.
That being said, can't wait to see this again.
0
0
9
It is possible to throw a gyro slider while staying very behind the ball, which is why I recommend pronators try and throw one.
I am a low slot not pronator, but you can see I can get across the horizontal while largely staying behind the ball.
(ball)
2
1
8
SSW sinkers rely on supination not pronation. This is counterintuitive given the arm-side run of the pitch, but to get the most out of the pitch you must cut the ball.
(I will explain physically why this works after giving some tips on throwing one)
1
0
9
There are a few simple ways to harness seam effects on your sinker. My favorite is the black circle:
1
0
9
A good test for a model is to look at my favorite example: Paul Sewald. His fastball at 92 and 13 vert sure seems like a pitch that shouldn’t work, until you realize it creates some of the most shallow angles of a non-submarine pitcher’s fastball in the game.
1
0
9
The “good” changeup generally comes in one of two forms. (1) The 3:00 o'clock sinking changeup, where you chase vertical separation from the fastball.
Consider the pronator king, Devin Williams:
1
0
8
First, the name inspiration. Basically PNWs are sliders with a lot of IVB. This movement places them firmly in the Oregon/Washington area and some even further Northwest.
1
0
8
Many sinker dominant pitchers get around the issue of the opposite platoon by throwing two independent pitch mixes to lefties and righties.
Here we can see both Houck and Graterol use a sinker as their primary pitch against RHH, but as their least used pitch against LHH.
1
0
8
Nevertheless, look at this 90 mph 8.5 IVB/-11.3 PNW.
Might be one of the most insane pitches I have ever seen.
0
0
8
Speas throws his fastball pretty regularly in the "deadzone" with 10-15 vert and 8-10 horz. His spin axis (1:00-1:30) also makes it relatively challenging to make an impressive SSW sinker, as the seam effects push the ball armside rather than down.
2
0
8
The second point: A good changeup.
As a pronator you can throw a changeup with real axes without relying on any of those fancy seam effects. Playing with grips and attempting to pronate the ball can get you decent axis deviation from the fastball, and thus decent separation.
1
0
8
In the data office, we love seeing this. Speas has elite velocity, with a fastball regularly thrown at 100. However, like most supination biased pitchers, creating a shape that plays with the fastball is a challenge.
2
0
7
He may throw majority sliders, but batters will always have to respect that this is in his back pocket.
(submax)
0
0
8
If done correctly, it should look something like this. Here we can see that Holmes goes from a 7:30 tilt to much closer to 9:00 break from the seam effects.
0
2
8
A good way of telling if your favorite pitcher is taking advantage of seam effects is to take a look at the difference in their spin axis (spin based) and observed movement (observed) on baseballsavant.
We can see here Webb has a 2:15 spin axis but a 3:15 movement axis:
1
0
8
This is where seam effects come into play. Sinkers tend to play really well against same-handed hitters, and thus we can develop a pitch that plays well against one platoon.
Here we see Tanner Houck use his sinker which is very effective against righties.
1
0
7
So how can you throw one? I am a big fan of drawing on a baseball. As long as you aren't throwing from a low slot, you should be able to see the dot hold consistently if you throw the slider right.
2
2
7
There is also a practical reason for throwing more supinated pitches when you are a supinator. This year in the minors, Speas set his career low number on walks/9.
For many guys, leaning into motor preferences also helps them command the ball.
1
0
7
It is also easier to throw more strikes when you are confident that your stuff plays in the zone.
It's pretty hard to hit a low 90s slider even when its in the zone. This is especially true when batters have to protect against an upper 80s sweeper and the 100 mph fastball.
3
0
7
Another, but slightly less scientific way is to try and cut your 2-seam a little. For some guys, it is as simple as telling them to try and cut it and rip it.
There is a fine balance here however, because if you cut the ball too much you will not get the right effect.
1
0
7
The number of pitchers who throw this pitch has been increasing, and I suspect that we will see more as we head into 2023.
But why will we see more of these? Lawrence generates a whiff under his PNW.
1
1
7
Or (2) a running changeup where you allow the speed difference to create significant drop from your fastball, but on a movement plot you primarily see a difference in horizontal movement.
Think of Chase Anderson CH: (From 2022 because Rockies data scares me)
1
0
7
The final point, and my favorite to talk about: the hard gyro.
I would be remiss to not mention throwing something on the zero line. As a supinator, it is likely that it will be challenging to throw anything with serious gloveside movement, but you do not need to.
1
0
7
He was the biggest stuff gainer from our old iteration of wStuff+ to our newest model which added this interaction term.
0
0
7
The issues with compressing performance into one number cannot be understated. Furthermore, the sweeper's dominance in this stat has been waining.
The reason this affects Stuff models is because if you are not updating your training set, you are missing changes in the meta.
1
3
7
At Tread, we use a bucketing system to separate pitchers into different groups based on release points. The buckets are defined by difference from average. For example, most pitchers fall into the 0 height,0 side bucket and relatively few fall into the 2,-2 bucket.
1
1
6
Let’s first consider Mitch Keller. Mitch Keller has a stock release height, and stock vert on his fastball. This creates stock angles at the plate, angles I would largely consider to be in the “deadzone”. So how is he able to produce better whiff numbers than Sewald?
1
0
6
Sometimes the most effective pitch design for a fastball is just to work on some velo.
0
0
5
Refer back to the R.A. Dickey knuckleball, clearly this does not take a clean parabolic arc. Therefore--to some extent--we know that late break does exist, but do any other pitches mimic this?
1
0
4
Shoutout
@ArowThrows
for being my competition that day and helping guide my training.
@cory_wilder
,
@Alexkachler10
, and
@DTreatss
were also pivotal in developing a lot of the training philosophies I currently use.
1
0
6
I'm very thankful to be working with such a great group of coaches and support staff here at Tread.
@ purbs
0
0
6
These heavily pronated fastballs are some of the most effective in the game. These are also largely out of reach for heavily supination dependent pitchers.
1
0
6
Seam Shifted Wake (SSW) works by manipulating the seems to restrict airflow to 3rd base and bottom side of the ball. Consider this demo from texasleaguers:
1
0
6
Therefore, SSW sinkers more closely resemble a cutter at ball release than a traditional change-up or high efficiency 2-seam.
1
0
5
If we look at each bucket as a group we can aggregate fastball metrics and find the average break for a fastball from each release point bucket.
1
1
5
Also gotta mention
@TurnerGivens
and his lifting squad for putting me onto waves.
I hit a belted 405 for a 40 pound belted pr and later hit 405 unbelted (no video :( ) for a 60 pound pr
1
0
5
The same IVB from a foot higher release height will generate far more “sink.” However, it is easy for the model to miss this relationship between variables, and easier still for an analyst to be unaware that the model is not mapping this relationship either.
1
0
5
Let's consider the "9-parameter" philosophy that I mentioned. This refers to calculate induced break by the 9 parameters:
starting positions of x,y,z
starting velocities of x,y,z
and constant accelerations of x,y,z
1
0
3
On the other end of the spectrum, we can see guys with "normal" release points generate significantly less vertical break by utilizing seam effects. Logan Webb is a well known example of this, averaging -2.2 vert on his SSW sinker.
1
0
5
He grips his CH deeper into the horse-shoe than he had been for his sinker, so we made the adjustment from a traditional 2-seam grip to one deeper in the horse shoe:
1
1
4
Bench went from a calculated 205 to a real (but swissbar) 235. I'm spending a lot of training economy there right now to hopefully improve the weakest part of my lifting.
@imstilllucas
hit me up with some meathead lifting protocols.
1
0
5
Webb has also started utilizing offspeed pitches more than fastballs. Depending on platoon we can see he favors sliders vs RHH and changes vs LHH over using either fastball.
1
0
5
Are sweepers worse now than in 2017? This does not seem to be the case.
Size and velo have been stable or improving since 2017.
Rather, the market is saturated. In 2017 there were 57 pitchers who threw a sweeping pitch at least 25 times. There were 271 in 2023.
1
1
4
Oftentimes, it takes a lot of tuning of the model to verify how the relationships between variables pan out in xgBoost algorithms. This is largely due to the fact these models are black box models; you put something in, you get something out without a clear reason why.
1
0
5
It is ultimately a very complex system but you can measure it on radar-based ball tracking systems like Trackman. You know it has worked if you see a significant decrease in vertical break after slightly cutting your 2-seam.
1
0
4
One cue that has worked for me is to think “cutter” and walk down the pitch to an acceptable vert and horz. Staying behind the ball as much as possible is the name of the game here. This will create the most velocity on the pitch.
1
0
4
Generally the movement goals for this pitch are just to be close-ish to 0,0. The second goal of this pitch is to throw it as hard as possible. As a pronator do NOT try to significantly supinate this pitch, you will lose substantial velocity moving against your motor preferences.
1
0
4
From 2019 to 2022 Logan Webb underwent a dramatic fastball (and mechanical) overhaul. We can see from 2019 he favored a four-seam fastball and a higher slot. This fastball averaged 13.5 vb and 9 hb. It was not particularly effective in either platoon.
1
0
4
In search of better results, he has completely changed his fastball. He now throws a sinker at -2/14.8 as his primary fastball. He also changed his mechanics a lot from 2019 to 2022, but this is not always required to access effective SSW.
2
0
4
These are all plus pitches and batters defending against them are forced to identify and make swing decisions on 3 separate offerings all traveling 90+ mph. Forcing batters to have to defend the zone with multiple offerings can help protect your 4-seam.
Cutter at 6/-4 at 95:
1
0
4
This is very similar (though mirrored) to the effects seen on sinkers.
My last thread predicted an increase in SSW sinkers in the next few years. This thread will be about how to throw one and physically how this effect is achieved in the context of fastballs.
1
12
223
1
0
4
This is also how you can have a negative induced vertical break on a fastball while the spin axis is above 3 o'clock. (This is pretty cool)
As a note, without taking advantage of these seam effects you cannot achieve negative IVB through Magnus effect while above 3:00 spin axis.
1
0
4
In the most extreme cases, we can see very low release points and very high IVB sliders generate ridiculous profiles. Consider this -2.4 VAA PNW from Adam Cimber:
1
0
4
There is a lot of room to experiment with this pitch and a surprising amount of shape you can push with it.
Consider this Tread favorite, and a pronator exclusive:
1
0
4
So we can give our Stuff+ model an interaction term, we just have to create the interaction term ourselves. By doing so, we can more accurately represent the value of each of these pitches, where a model without the interaction may have trouble picking up on this relationship.
1
0
4
When the air kicks off this top seam, it imparts a force to the top of the ball (Newton's 3rd Law). Thus the ball is actually pushed downward as a result of the seam orientation during ball flight.
2
0
4
This puts you in the funny situation that cutting the ball more will actually increase the vertical break on your fastball. The ideal range of efficiency is around 75-90%.
Logan Webb: 76% eff
Tanner Houck: 80% eff
Brusdar Graterol: 91% eff
1
0
4
Besides, it's cooler to strike people out when they swing over your fastball (as shown by Tanner Houck below).
0
0
4
Sewald massively outperforms his "deadzone" fastball's movement because it produces an extremely shallow VAA at the top of the zone. Consider this comparison with Kyle Wright and the "average" FF in the same zone.
1
0
4
Likewise, we can see this group has significantly lower vertical break than their cohort:
1
1
4
This is reason enough to throw one if it comes naturally because sweepers are well known to be one of the best pitches in the game.
However, I will argue that the PNW is even more effective than the traditional LR Sweeper.
1
0
4
When thrown up in the zone, we can see the PNW continue to perform. Whiffs remain high for this pitch at the top of the zone, even when they decrease for the common sweeper.
1
0
3
Like in the video I opened the thread with, you want to throw your 2-seam with the circle stable and rotating on the top of the ball, favoring the first base side slightly.
1
0
3
Thus, when we control for plate location and compare Holmes and Lawrence we see the big difference.
Despite averaging almost 5 inches more depth than Holmes, Lawrence's fastball cross the plate at a much more shallow angle.
1
0
3
It is also a good time to note that both Keller and López have one thing in common. They are starters that throw fuel. Consider this 97 mph heater for a K from López.
1
0
3
So why do I highlight these examples? A lot of time is spent trying to pitch design fastballs. A pitch that’s shape is often predetermined by the biomechanical factors of the pitcher throwing. Generally, a fastball’s shape is determined by whatever profile achieves maximal velo.
1
1
3
If we were working with a linear model (I love linear models), the obvious answer would be creating an interaction effect!
1
0
3
Common sweeper usage would have the pitcher throw low and gloveside.
When compared to more traditional sweepers (lower IVB/Low Ride), we can see that the PNW performs nearly identically.
1
0
2
This also forces the batter to defend the plate like in the Keller example, because they have to be prepared to hit a fastball thrown to all parts of the plate and distinguish it from both his changeup and sweeper thrown to ideal locations.
1
0
2