Run more efficiently — see where your energy goes.
Record a run. Pose Coach measures your running efficiency across four dimensions — cadence, vertical oscillation, symmetry, and stride — and tells you the one thing to optimise next. This scores how economically you run, not how fast.
One efficiency score, four weighted dimensions.
Pose Coach reads your gait data, scores four efficiency dimensions 0–100 each, and blends them by the weights below into a 0–100 overall score and an S/A/B/C/D grade. The weights reflect each dimension's known contribution to overall efficiency.
Vertical Ratio carries the most weight (30%) because wasted up-and-down bounce is the most direct and common source of energy loss. The same S–D scale powers our other scores, so grades stay comparable.
| Dimension | Weight |
|---|---|
| Cadence Efficiency | 25% |
| Vertical Ratio | 30% |
| Time Symmetry | 25% |
| Stride Efficiency | 20% |
| Grade | Efficiency | Score |
|---|---|---|
| S | Near elite | ≥ 90 |
| A | Solid | ≥ 80 |
| B | Core in place | ≥ 70 |
| C | Significant gap | ≥ 60 |
| D | Foundation needs work | < 60 |
The four dimensions
Each answers three questions: what it measures, what's ideal, and how to improve it. The order never changes — it matches the score card.
Cadence Efficiency
Measures — whether your cadence (steps per minute) falls in the elite range. Too low means excessive air time and hard landings; too high means overly choppy steps.
Ideal — target range 175–185 spm (full marks in range, zeroing out at 30 spm deviation). In-app: {v} spm · on target / low / high.
- If low: add metronome drills targeting 175–185 spm; short intervals speed up adaptation
Vertical Ratio
Measures — the ratio of vertical oscillation to stride length (%). A higher ratio means more wasted bounce per stride — energy going up instead of forward.
Ideal — elite standard < 6%; < 8% good; < 10% high; ≥ 10% too high (zero score). In-app: Vertical ratio {v}% · elite / good / high / too high.
- If high: strengthen core and hip flexors; focus on horizontal propulsion, not bounce
Child note (age < 14): ARKit's adult-skeleton estimate inflates a child's foot oscillation, so adult thresholds don't apply. For children this dimension falls back to a neutral score of 50, marked "Not calibrated · child." It's a tool limitation, not the child's form.
Time Symmetry
Measures — whether your left and right legs split stance time evenly (each 50%). Asymmetry suggests a bilateral strength or flexibility imbalance.
Ideal — closest to 50%: < 1.5% deviation is ideal, < 4% slight, ≥ 4% marked (zeroing out at 10%). In-app: Time symmetry {v}% (ideal 50%) · near ideal / slight / marked.
- If asymmetric: add single-leg drills (lunges, hops) to balance both sides
Stride Efficiency
Measures — the ratio of your actual stride length to the theoretical optimum for your pace. Both too short (under-propulsion) and too long (overstriding) lose points.
Ideal — optimal stride = pace (m/s) ÷ 3.0; an actual/optimal ratio of 1.0 = full marks (zeroing out at 0.3 deviation). In-app: Stride {v} m, optimal {opt} m (pace {spd} m/s), ratio {r}.
- If mismatched: let pace and cadence set a natural stride — don't deliberately overstride
References: Struzik et al. 2016 (175–185 spm elite cadence) · Gómez-Molina et al. 2017 (< 6% elite vertical ratio). These are literature thresholds, not yet calibrated to a Pose runner sample — see below.
Your efficiency score card, element by element
- Score ring + grade badge — the weighted overall 0–100 and its S/A/B/C/D letter.
- Grade verdict — a one-line read from the grade; at B/C/D it names your weakest dimension.
- Weakest-dimension drill — at B/C/D, a specific drill for the dimension costing you the most.
- Four dimension bars — Cadence / Vertical Ratio / Time Symmetry / Stride, each with its score and a one-line detail; the weakest is highlighted.
- Running Power (estimate) — shown only if you enter your body weight; a physics estimate, flagged as such.
- Engine version + status — the engine version and a note that reference values are preliminary (literature-based).
Each score comes with a verdict — and the one thing to drill
The score gives a grade-based verdict; at grade B / C / D it appends a drill for your weakest dimension — what to practice, not just "keep at it."
| S | Running efficiency near elite level. Keep it up. |
| A | Solid efficiency. Focus on the weakest dimension to advance. |
| B | Core efficiency in place. Targeting your weakest dimension will yield clear gains. |
| C | Your weakest dimension shows a significant gap; targeted training recommended. |
| D | Efficiency foundation needs work. Start with your weakest dimension. |
Tap Generate Training Plan to hand the efficiency score to the AI coach and get a targeted plan.
Honest about the numbers.
Efficiency reference values come from sports-science literature, not yet calibrated to a real-runner sample. Scores are directional reads — how far you deviate from published elite references — not a validated efficiency scale.
It can
- Quantify cadence, vertical ratio, symmetry and stride efficiency
- Point to your weakest dimension and a drill for it
- Estimate Running Power (if you enter your weight)
- Flag recordings whose data isn't reliable enough to score
Keep in mind
- Reference values are literature thresholds, not a Pose runner sample (preliminary_ref_values)
- It doesn't replace lab testing or a coach's periodic assessment
- Running Power is a Minetti-2002 estimate — it deviates from sensors like Stryd
- For children (age < 14), Vertical Ratio is neutralised — adult thresholds don't apply
Preliminary calibration
Thresholds from Struzik 2016 / Gómez-Molina 2017, not yet statistically calibrated to Pose runners.
Children (age < 14)
ARKit's adult skeleton overstates a child's vertical oscillation, so Vertical Ratio falls back to a neutral 50.
Running Power
A physics estimate from your weight (Minetti 2002), flagged "estimate" — reference only, not a load prescription.
If a complete run isn't captured, the app asks you to record again. If the data isn't reliable enough (e.g. cadence < 100 or > 240 spm, oscillation > 30 cm, stride > 4 m), it flags "data isn't reliable enough" — re-record with the camera held level and steady, in good lighting.
Efficiency is measurable — see where yours stands.
Record a run and see your four efficiency dimensions, your grade, and the one thing to optimise next.
Scoring engine v0.1.0 · Reference values are preliminary (literature-based). A directional read of efficiency, unrelated to speed.