Stickd report · 2026
The Most Common Gymnastics Deductions, by Level and Event (2026 Report)
Across 359 practice routines analyzed by Stickd in the 2025–26 and 2026–27 seasons, bent knees appeared in 72% of routines, more than any other deduction, followed by step on landing at 70%.
These are deductions detected by Stickd's AI in practice videos, not official judging data. The judges' scores at a meet are the official ones. Figures are the share of routines in which a deduction appeared, so they do not depend on deduction values.
Key findings
- Bent knees is the most common deduction overall: it appeared in 72% of routines, an average of 1.3 times per routine. Step on landing was next at 70%.
- The top deduction differs by event. Vault: bent arms (77%); uneven bars: step on landing (81%); balance beam: bent knees (90%); floor exercise: bent knees (74%).
- The highest rate for any event: bent knees in 90% of balance beam routines, with balance checks in 79%.
- On bars, a low cast angle appeared in 71% of routines.
- Bent knees become less common at higher levels: 82% at Levels 1–3 and 81% at Levels 4–5, against 53% at Levels 6–10, where step on landing (66%) leads.
- Falls were rare in these practice videos: 7% of routines. Most detections are small form faults that repeat within a routine.
The 10 most common deductions
Share of all 359 routines in which each deduction appeared at least once. All levels and events.
Every deduction in the data (26)
| Deduction | Routines with it | Avg. per routine |
|---|---|---|
| Bent Knees | 72% | 1.33 |
| Step on Landing | 70% | 0.80 |
| Leg Separation | 50% | 0.76 |
| Flexed Feet | 49% | 0.64 |
| Bent Arms | 48% | 0.67 |
| Artistry | 43% | 0.43 |
| Insufficient Amplitude | 39% | 0.52 |
| Insufficient Split | 34% | 0.52 |
| Arch in Back | 34% | 0.47 |
| Pike in Body | 30% | 0.50 |
| Body Alignment | 28% | 0.34 |
| Poor Posture | 26% | 0.29 |
| Balance Check | 24% | 0.39 |
| Under-rotation | 22% | 0.24 |
| Rhythm Issues | 20% | 0.25 |
| Low Cast Angle | 18% | 0.27 |
| Deep Squat on Landing | 17% | 0.18 |
| Heel Drop During Turn | 13% | 0.13 |
| Low Leap Height | 11% | 0.14 |
| Shoulder Angle | 10% | 0.11 |
| Fall | 7% | 0.07 |
| Incomplete Turn | 5% | 0.05 |
| Pause or Hesitation | 4% | 0.06 |
| Missing Element | 4% | 0.06 |
| Composition | 4% | 0.04 |
| Hop on Landing | 3% | 0.03 |
By event
“Routines with it” is the share of routines on that event in which the deduction appeared. “Avg. per routine” counts repeats, so a routine with three bent-knee calls adds three.
Vault (50+ routines)
| Deduction | Routines with it | Avg. per routine |
|---|---|---|
| Bent Arms | 77% | 0.79 |
| Step on Landing | 71% | 0.77 |
| Leg Separation | 42% | 0.48 |
| Arch in Back | 40% | 0.44 |
| Bent Knees | 37% | 0.37 |
| Insufficient Amplitude | 33% | 0.46 |
| Body Alignment | 31% | 0.31 |
| Shoulder Angle | 29% | 0.29 |
| Flexed Feet | 27% | 0.27 |
| Pike in Body | 27% | 0.29 |
Uneven bars (80+ routines)
| Deduction | Routines with it | Avg. per routine |
|---|---|---|
| Step on Landing | 81% | 0.83 |
| Bent Knees | 79% | 1.40 |
| Low Cast Angle | 71% | 1.09 |
| Leg Separation | 69% | 1.31 |
| Flexed Feet | 51% | 0.76 |
| Pike in Body | 44% | 1.14 |
| Bent Arms | 43% | 0.70 |
| Insufficient Amplitude | 34% | 0.41 |
| Arch in Back | 31% | 0.58 |
| Rhythm Issues | 30% | 0.37 |
Balance beam (40+ routines)
| Deduction | Routines with it | Avg. per routine |
|---|---|---|
| Bent Knees | 90% | 1.90 |
| Balance Check | 79% | 1.69 |
| Step on Landing | 79% | 0.81 |
| Artistry | 76% | 0.76 |
| Insufficient Split | 71% | 1.05 |
| Insufficient Amplitude | 50% | 0.81 |
| Flexed Feet | 45% | 0.55 |
| Leg Separation | 45% | 0.50 |
| Rhythm Issues | 38% | 0.50 |
| Bent Arms | 33% | 0.45 |
Floor exercise (170+ routines)
| Deduction | Routines with it | Avg. per routine |
|---|---|---|
| Bent Knees | 74% | 1.45 |
| Artistry | 67% | 0.67 |
| Step on Landing | 61% | 0.79 |
| Flexed Feet | 56% | 0.72 |
| Insufficient Split | 52% | 0.80 |
| Bent Arms | 46% | 0.68 |
| Leg Separation | 45% | 0.63 |
| Insufficient Amplitude | 40% | 0.53 |
| Arch in Back | 35% | 0.46 |
| Poor Posture | 28% | 0.32 |
By level group
All events together. Where a level group had too few routines or gymnasts to publish on its own, it is merged with its neighbor (see methodology).
Levels 1–3 (90+ routines)
| Deduction | Routines with it | Avg. per routine |
|---|---|---|
| Bent Knees | 82% | 1.73 |
| Flexed Feet | 66% | 0.92 |
| Step on Landing | 61% | 0.63 |
| Artistry | 53% | 0.53 |
| Bent Arms | 52% | 0.71 |
| Leg Separation | 51% | 0.63 |
| Insufficient Amplitude | 35% | 0.45 |
| Insufficient Split | 32% | 0.44 |
| Body Alignment | 31% | 0.37 |
| Arch in Back | 28% | 0.38 |
Levels 4–5 (100+ routines)
| Deduction | Routines with it | Avg. per routine |
|---|---|---|
| Bent Knees | 81% | 1.38 |
| Step on Landing | 75% | 0.81 |
| Bent Arms | 52% | 0.75 |
| Insufficient Amplitude | 52% | 0.68 |
| Flexed Feet | 48% | 0.58 |
| Leg Separation | 46% | 0.68 |
| Artistry | 44% | 0.44 |
| Insufficient Split | 38% | 0.59 |
| Balance Check | 30% | 0.51 |
| Arch in Back | 29% | 0.36 |
Levels 6–10 (optional) (90+ routines)
Merged: Levels 6–7 and Levels 8–10.
| Deduction | Routines with it | Avg. per routine |
|---|---|---|
| Step on Landing | 66% | 0.87 |
| Bent Knees | 53% | 0.90 |
| Leg Separation | 52% | 0.86 |
| Arch in Back | 43% | 0.70 |
| Bent Arms | 43% | 0.60 |
| Under-rotation | 38% | 0.42 |
| Flexed Feet | 34% | 0.41 |
| Pike in Body | 34% | 0.51 |
| Artistry | 33% | 0.33 |
| Insufficient Split | 30% | 0.53 |
Xcel (all divisions) (40+ routines)
Merged: Xcel Bronze–Silver and Xcel Gold–Diamond.
| Deduction | Routines with it | Avg. per routine |
|---|---|---|
| Step on Landing | 88% | 1.02 |
| Bent Knees | 76% | 1.67 |
| Bent Arms | 52% | 0.69 |
| Insufficient Amplitude | 52% | 0.81 |
| Leg Separation | 52% | 0.95 |
| Balance Check | 50% | 0.86 |
| Arch in Back | 48% | 0.55 |
| Insufficient Split | 48% | 0.64 |
| Pike in Body | 48% | 0.90 |
| Artistry | 45% | 0.45 |
By level group and event
The top five for each combination with enough data. Combinations that fall below the privacy thresholds are left out.
Vault
Levels 1–5 (compulsory) (20+ routines)
| Deduction | Routines with it | Avg. per routine |
|---|---|---|
| Bent Arms | 76% | 0.81 |
| Step on Landing | 71% | 0.81 |
| Arch in Back | 48% | 0.52 |
| Bent Knees | 48% | 0.48 |
| Insufficient Amplitude | 48% | 0.67 |
Uneven bars
Levels 1–3 (20+ routines)
| Deduction | Routines with it | Avg. per routine |
|---|---|---|
| Bent Knees | 92% | 1.92 |
| Step on Landing | 88% | 0.88 |
| Low Cast Angle | 84% | 1.04 |
| Flexed Feet | 68% | 1.00 |
| Leg Separation | 64% | 0.76 |
Levels 4–5 (30+ routines)
| Deduction | Routines with it | Avg. per routine |
|---|---|---|
| Low Cast Angle | 87% | 1.30 |
| Bent Knees | 83% | 1.27 |
| Step on Landing | 77% | 0.77 |
| Leg Separation | 73% | 1.27 |
| Flexed Feet | 53% | 0.77 |
Balance beam
Levels 1–5 (compulsory) (20+ routines)
| Deduction | Routines with it | Avg. per routine |
|---|---|---|
| Bent Knees | 96% | 2.24 |
| Artistry | 88% | 0.88 |
| Balance Check | 76% | 1.84 |
| Step on Landing | 76% | 0.80 |
| Insufficient Split | 72% | 1.08 |
Floor exercise
Levels 1–3 (50+ routines)
| Deduction | Routines with it | Avg. per routine |
|---|---|---|
| Bent Knees | 80% | 1.63 |
| Artistry | 73% | 0.73 |
| Flexed Feet | 68% | 0.93 |
| Bent Arms | 51% | 0.73 |
| Step on Landing | 46% | 0.47 |
Levels 4–5 (30+ routines)
| Deduction | Routines with it | Avg. per routine |
|---|---|---|
| Bent Knees | 85% | 1.51 |
| Artistry | 77% | 0.77 |
| Step on Landing | 77% | 0.87 |
| Insufficient Split | 72% | 1.05 |
| Flexed Feet | 67% | 0.72 |
Levels 6–10 (optional) (40+ routines)
| Deduction | Routines with it | Avg. per routine |
|---|---|---|
| Bent Knees | 63% | 1.29 |
| Step on Landing | 57% | 0.98 |
| Artistry | 55% | 0.55 |
| Insufficient Split | 49% | 0.86 |
| Leg Separation | 49% | 0.67 |
How to use this
- Parents: the top rows for your gymnast's level and event are the faults most gymnasts at that stage are still working on. Seeing one on a score sheet is normal, not a sign that something is wrong.
- Coaches: the frequent, repeated faults (bent knees, flexed feet, leg separation) cost a little each time and show up in most routines. Short daily drills on them tend to pay back across every event. The drill library has drills for each.
- Gymnasts: pick the top one or two for your event and practice the fix in the skills where it shows up. The skills hub breaks each skill into progressions, and each deduction name above links to what it is and how it is judged.
Methodology
Data source. Completed Stickd analyses of practice videos uploaded on the web and iPhone app by gymnasts, parents and coaches in the 2025–26 and 2026–27 seasons (USA Gymnastics competitive year, August to July): 171 routines in 2025–26 and 188 routines in 2026–27. In total 359 routines from 172 gymnast profiles, with 3,356 detected deductions. Women's artistic gymnastics only: vault, uneven bars, balance beam and floor.
What “appeared in” means. A deduction appeared in a routine when Stickd's analysis of that video detected it at least once. The percentage is the share of routines in the group with at least one detection. The average per routine counts every detection, including repeats, divided by all routines in the group.
Exclusions. Deleted analyses and routines, and analyses that failed, are not counted. Also removed:
- Internal, staff, App Store review, demo and test accounts: 20 routines.
- Gymnasts recorded as male (men's gymnastics is scored differently): fewer than 10 routines.
- Clips not tagged to vault, bars, beam or floor: 19 routines.
Routines with no level recorded count toward the overall and per-event tables only.
Privacy thresholds. A group (an event, a level group, or a level group on one event) is published only when it has at least 20 routines from at least 10 different gymnasts, and no single gymnast supplies more than 25% of its routines. A level group that falls short is merged into Levels 1–5, Levels 6–10 or Xcel; if the merged group also falls short, it is left out. Within a group, a deduction is listed only when it appeared in at least 10 routines from at least 5 gymnasts. Group sizes are rounded down to the nearest 10. No gym names, gymnast names or dates finer than the season are published.
Left out for too little data: Levels 6–10 (optional), vault; Xcel (all divisions), vault; Levels 6–10 (optional), uneven bars; Xcel (all divisions), uneven bars; Levels 6–10 (optional), balance beam; Xcel (all divisions), balance beam; Xcel (all divisions), floor exercise.
AI detection caveats. Deductions were detected by Stickd's AI from video. It can miss a fault or call one a judge would not, and camera angle, distance and lighting affect what it sees. Artistry and rhythm are judgment calls even for people. The videos are practice routines that users chose to upload, not a random sample of gymnasts or meets, and a few gymnasts uploaded many routines.
Scoring values. On September 29, 2026 Stickd updated the value it assigns to each deduction. Every routine in this edition was scored before that change, so this report shows how often deductions appear and does not average their values. For typical values, see each deduction's glossary page.
License. The aggregate tables on this page are licensed under CC BY 4.0. Cite Stickd and link to this page. Data as of 2026-09-29.
Frequently asked questions
- What is the most common deduction in gymnastics?
- In Stickd's data it is bent knees. Across 359 practice routines analyzed by Stickd, bent knees appeared in 72% of routines, followed by step on landing at 70%. These are AI-detected deductions on practice videos, not official meet results.
- What is the most common deduction on balance beam?
- Bent knees, in 90% of beam routines, followed by balance check (79%) and step on landing (79%).
- Are these official judging statistics?
- No. The figures count deductions that Stickd's AI detected in practice videos uploaded by gymnasts, parents and coaches. Judges at a meet may see a routine differently, and the judges' scores are the official ones. Use the figures as a guide to what to practice.
- Can I use these figures in an article or presentation?
- Yes. The aggregate tables are licensed under CC BY 4.0. Cite Stickd and link to this page.