Keep the ball out in the fingers.
Held like a fastball: out in the fingers, not choked into the palm.
The four-seam, tilted.
Two-seam fastball (sinker)
Same arm, the axis rolled toward the hand. Ride becomes run, and the ball finishes at the knees. This is how the pitch sinks.
Specimen 01
seam-informed schematic
Sourced, not corrected
01Grip Lab
View
Hand
Drag to inspect, or use the View buttons
The runner / 01
Two-seam fastball (sinker)
Almost as hard as the four-seam, but it tails toward your arm and sinks. A ground-ball pitch.
Ball depth
Out in the fingers
Spacing
Slight spread
Thumb
Thumb supports underneath and can sit slightly off-center.
Release feel
Throw fastball effort and let the along-seam contact tilt the axis.
Held like a fastball: out in the fingers, not choked into the palm.
Palm gap cue
- Index seam / Starts the arm-side biasRuns along one narrow seam. Along the seam, not across it.
- Middle seam / Pairs with index to guide runRuns along the paired narrow seam. Same width as fastball.
- Thumb / Balances control under the seam pairUnder leather, slightly off-center. Do not pinch it tight.
See it live / via @PitchingNinja
The grip is the cause. This is the effect, the same pitch shot in a real bullpen. Sourced, not corrected.
- 1Set your index and middle fingers along the two narrow seams where they run close together, not across the horseshoe.
- 2Keep the same width and softness as your four-seam: a finger-width apart, held out in the hand.
- 3Rest your thumb underneath on the leather; some pitchers ride the side of the thumb against a seam.
- 4Throw it like a fastball. The grip makes the movement. You do not have to.
Same effort as the four-seam. Stay behind it and let the seams do the work.
The sourced grip, in full
Index and middle fingers run along the pair of narrow seams that sit close together, rather than across the horseshoe. That orientation tilts the spin off pure backspin and trades ride for arm-side run and sink.
Reputable analysisDriveline Baseball, How to Throw a Sinker or Two-Seam Fastball
Paraphrased from Driveline, not quoted.
The standard Driveline two-seam grip lays the index and middle fingers on the two close seams, thumb underneath or slightly off-center, ring finger steadying the side. It is the baseline they reach for about three-quarters of the time.
Reputable analysisDriveline Baseball, How to Throw a Sinker or Two-Seam Fastball
Paraphrased. The FT 1 label and the roughly 75 percent usage come from Driveline.
For a sinker specifically, the index finger often shifts slightly inward while keeping seam contact, biasing the release toward side-spin and diving action over balanced backspin-and-run.
Reputable analysisDriveline Baseball, How to Throw a Sinker or Two-Seam Fastball
Paraphrased.
The most extreme movers ride a single seam: a one-seam grip that orients the leather to bend the airflow itself, adding run and sink beyond what spin alone predicts. Driveline found 12 of the 20 largest seam-shifted-wake movers threw a one-seam sinker.
Reputable analysisDriveline Baseball, Seam-Shifted Wakes and their Effect on Sinkers/ 2020
Paraphrased. The 12-of-20 figure is from the Driveline 2020 analysis of the largest axis-deviation pitchers.
Grip geometry is schematic and shows the standard along-seam family, not every one-seam or sinker variant.
02Release Room
Translate the hold into a release.
Grip shape only matters if the release makes sense. This room keeps the player reading pressure, thumb support, and ball depth before movement numbers show up.
Same effort as the four-seam. Stay behind it and let the seams do the work.
index finger owns the shape.
Starts the arm-side bias
Release is a feel, not a formula.
Throw fastball effort and let the along-seam contact tilt the axis.
Film Room / watch a master
The schematic shows the release path. This is the release: Gibson walking through the delivery himself. Real hand, real ball, real finish, no model can fake it.
The release rolls the fingers over the outer half of the ball, adding side-spin to the backspin and tilting the axis toward the arm. A lower slot adds more of that tilt; on one-seam grips the seam orientation adds late sink without changing the slot. The cue is a downhill plane attacked at the knees, a pitch built to be hit on the ground.
Reputable analysisDriveline Baseball, How to Throw a Sinker or Two-Seam Fastball
Paraphrased from Driveline; the seam-shifted-wake point is cross-referenced to the Driveline SSW article.
03Movement translation
It runs toward your arm and settles at the knees.
Running along the seams tilts the spin, so the ball tails toward your throwing hand and rides far less than a four-seam, which is why it sinks by comparison. You trade a little pure life for movement that buys ground balls and weak contact.
The measured movement, if you want it
A four-seam rides because near-pure backspin points the Magnus force straight up. A sinker tilts that axis toward the arm and throws in less efficient spin, so the force points up and sideways instead. The ball rides less, runs toward the arm, and finishes low, which is how it turns barrels into ground balls.
Official dataMLB.com Statcast Glossary, Sinker (SI)
Synthesized from the MLB glossary (drops, forces grounders, thrown lower with less spin) and the Driveline axis and seam-shifted-wake mechanics.
Spin axis and force direction, drawn in render space
Spin rate
≈ 1,900 to 2,250 rpm for elite arms. Webb spins about 1,939, Valdez about 2,150. The point is not high spin; sinkers are often thrown with less spin to keep the ball down.approx
Official dataBaseball Savant (Statcast), Logan Webb/ 2024
Webb 1,939 rpm and Valdez about 2,150 rpm from Savant; the band is a rounded characterization, not a rule.
Active spin
≈ Lower effective lift, by design. Valdez sat near 79 percent active spin in 2024; the axis tilt and any gyro or seam component point less of the force straight up.approx
Reputable analysisPitcher List, How Lower Is Better for Framber Valdez/ 2024
Lowering it in 2024 dropped his induced vertical break to about 6.3 in and restored the ground-ball bite.
Arm-side run
≈ Strong arm-side run is the signature. Framber Valdez runs about 15.5 to 16 inches toward his arm side, well beyond a typical sinker.approx
Official dataBaseball Savant (Statcast), Framber Valdez/ 2024
Savant arm-side horizontal break across 2024 to 2026. The MLB glossary defines the sinker qualitatively by hard downward and arm-side movement; it sets no inch benchmark.
Induced vertical break
≈ Far below a four-seam. The Webb sinker induces only about 1 inch of ride; it does not drop more than gravity, it just rides far less than a backspinning four-seam, so it sinks by comparison.approx
Official dataBaseball Savant (Statcast), Logan Webb/ 2024
Webb 2024 induced vertical break is about 1.1 in (his ~15 in figure is arm-side run, not ride). Total drop including gravity is about 31 in.
Spin axis
≈ Backspin tilted toward the throwing arm, around a 1:30 to 2:30 clock for a right-hander versus near 12:00 for a four-seam. Less of the spin converts to lift, so the ball rides less and sinks.approx
Reputable analysisDriveline Baseball, How to Throw a Sinker or Two-Seam Fastball
General range; mirror it for a left-hander. Valdez re-oriented his to about 10:45 in 2024 by dropping his arm angle near 43 degrees.
It rides far less than a four-seam and runs toward the arm. Thrown low, it produces ground balls.
02Master files
The verified baseline.
Three documented sinkers, three ways down. The visual is our own seam schematic. Every figure is season-stamped and links to its source.
Master file · 01Verified · Attributed
Framber Valdez
The textbook modern sinker as a ground-ball engine, a whole elite starter built around one arm-side-running, diving fastball thrown at the knees.
Velocity
≈ 94.1 mphapprox
Reputable analysisPitcher List, How Lower Is Better for Framber Valdez/ 2024
2024; Savant shows about 94.2 in 2025.
Induced vertical break
≈ ~6.3 inapprox
Reputable analysisPitcher List, How Lower Is Better for Framber Valdez/ 2024
2024, down from 9.9 the prior year. Low for a fastball, which is the point.
Arm-side run
≈ ~15.5 inapprox
Official dataBaseball Savant (Statcast), Framber Valdez/ 2024
2024 Savant horizontal break; elite, well above the league sinker.
Spin rate
≈ ~2,150 rpmapprox
Official dataBaseball Savant (Statcast), Framber Valdez/ 2024
A single tracked 2024 sinker read 2,153 rpm.
Ground-ball rate
≈ 60.6% (MLB leader)approx
Reputable analysisFanGraphs, Framber Valdez/ 2024
2024, about 60.6 to 61.7 percent by cut; led MLB. Whole-arsenal, driven by the sinker.
Master file · 02Verified · Attributed
Logan Webb
The right-handed answer to the same idea: a sinker-first starter whose wide arm-side spread rolls bats over into weak contact.
Velocity
Induced vertical break
≈ ~1.1 inapprox
Official dataBaseball Savant (Statcast), Logan Webb/ 2024
2024. Almost no ride, the defining trait of a sinker. Total drop with gravity is about 31 in.
Arm-side run
≈ ~15.1 inapprox
Official dataBaseball Savant (Statcast), Logan Webb/ 2024
2024 Savant arm-side horizontal break.
Spin rate
≈ ~1,939 rpmapprox
Official dataBaseball Savant (Statcast), Logan Webb/ 2024
A low-spin sinker, about 13.2 revolutions hand to plate, the less-spin-more-sink design.
Ground-ball rate
≈ 57.2% (4th in MLB)approx
Reputable analysisFanGraphs, Logan Webb/ 2024
2024; led MLB with 353 ground balls produced.
Master file · 03Verified · Attributed
Clay Holmes
The seam-shifted-wake case: a power sinker whose movement bends well past what its spin axis predicts, the clearest live example of a ball running on its seams.
Velocity
≈ 93.7 mphapprox
Official dataBaseball Savant (Statcast), Clay Holmes
2025 to 2026 as a Mets starter; he sat 96.6 mph as a 2024 Yankees closer.
Axis deviation (seam-shifted wake)
≈ ~28 degapprox
Reputable analysisColin Hofmeister, Reverse-Engineering the Perfect Sinker/ 2026
A 28.2 degree gap between inferred and observed movement, among the most deceptive sinkers in the league.
Spin rate
≈ ~2,130 rpmapprox
Official dataBaseball Savant (Statcast), Clay Holmes
Current Mets sinker, about 14 revolutions hand to plate.
Filed only when the bar is met. A real figure from the wrong tracking system, or a great arm whose signature pitch sits in a different category, gets left off rather than dressed up. The gap is the honesty.
Tier 03 / Field Notes
Field notes from the bullpen.
Every pitcher fiddles. A thumb creeps lower, a seam catches more leather, a cue from a coach suddenly makes the pitch move. Pitch Atlas keeps those experiments visible, labeled, and debated, so the small discoveries stop disappearing into group chats and comment sections.
How notes rank
Evidence and context, never who shouts loudest.
There is no single correct way to throw a pitch, but there are better and worse claims. A funny line should never outrank a tested grip. So notes do not rise on raw votes. They rise on weighted signals - provenance and adoption first.
- 35%Provenance
Source tier and evidence. A coach-observed note with a link outranks an unverified hunch with none.
- 20%Adoption
How many other pitchers independently tried the same tweak. Real replication, not passive clicks.
- 20%Usefulness
Marked useful, measured against views so a small loud crowd cannot stuff the rank.
- 15%Context match
How close a note is to your level, slot, and velocity. Computed in your session, never stored.
- 10%Community confidence
Considered practice: a real sample size behind the claim, not one good bullpen.
The vocabulary
- Field Note
- One pitcher's report on a grip variant.
- Variant
- A specific, named change from the canonical grip.
- Tried This
- A second contributor replicating the tweak.
- Coach Note
- A coach reporting on an arm they work with.
- Source Challenge
- A standing request for the evidence behind a claim.
- Provenance
- The source tier and evidence that set a note’s rank.
- Adoption
- How many others independently tried it.
- Needs Evidence
- A note flagged for missing support, kept visible.
The living layer
Field notes open soon.
Soon you will log your own grip tweak, mark the ones you have tried, and flag anything off, anonymously or under a handle you keep. When the community layer opens, every note will carry a source and confidence label, a content filter will block abusive language, and any note can be flagged. A note hides automatically once enough people report it.
When they open, every community variant will carry the same source and confidence labels as the records above. Nothing appears here unsourced, and no count is shown until it is real.