Room for Improvement
No methodology is perfect. Here's what we know needs work and where each improvement stands. We'd rather be honest about our limitations than pretend they don't exist.
- Planned
Bodyweight Tonnage Integration
Factor in athlete body mass for bodyweight exercises (push-ups, pull-ups, dips, pistol squats) so they contribute to tonnage and load metrics. Will use the weight from your profile or a manual override.
- Shipped
Universal Internal-Load Currency (sRPE → AU)
Move from tonnage-as-primary to a universal session-load number using Foster's sRPE × duration method. AU is now the headline load metric on every session card and rolls up weekly with a 3-bucket modality split (Strength / Endurance / Mixed). Tonnage stays visible as a strength-specific secondary.
- Future
Effective Sets (Hard Sets) Tracking
Track sets taken within 0-3 RIR of failure separately from total sets. Research increasingly suggests that "hard sets" are a better predictor of hypertrophy stimulus than total volume. This would give a more nuanced view of training dose.
- Future
Time-Based Volume Equivalents
Convert time-based exercises (planks, carries, isometric holds) into a volume-equivalent metric so they stop being invisible in session totals. Likely a time-under-tension proxy.
- Future
Individualized Alert Thresholds
Replace static delta thresholds (0.7 / 1.2 RPE, 10% / 25% volume/load) with personalized thresholds derived from each athlete's historical variance.
- Planned
Multi-Formula e1RM with Confidence Intervals
Run Epley, Brzycki, and Lombardi formulas in parallel and report the median with a confidence band, rather than relying on a single formula.
- Planned
Partial Session Completion Tracking
Move beyond binary "completed / not completed" to track what percentage of a session was actually finished.
- Future
Per-Muscle-Group Weekly Volume
Aggregate weekly set volume by muscle group using our exercise library classifications. This would let us flag under- or over-trained muscle groups relative to evidence-based guidelines (10-20 sets/week).
- Future
Power / HR Endurance External Load (TSS, TRIMP)
Today, endurance external load falls back to sRPE × duration because we don't capture power data or HR thresholds. Adding a power-meter ingest pipeline (TSS) and a stored HRmax / HRrest profile (HR-zone TRIMP) would let endurance sessions report objective external load alongside the subjective AU number. Out of scope for the load-monitoring v1 ship.
- Planned
Hooper Drop-off Mitigation
If an athlete dismisses the daily readiness prompt 3 days in a row, suppress it for a 2-week cooldown. The current 2-week guess is unvalidated — once we have real submission data we'll tune the threshold and cooldown length to maximise submission rate without becoming a nag.
- Future
Prilepin's Table as LLM Strength-Prescription Guardrail
Validate LLM-generated strength sessions against Prilepin's reps-per-set / total-reps ranges for the implied %1RM (derived from targetRpe → RIR → e1RM). Two surfaces: a plan-generation guardrail that flags or corrects out-of-range prescriptions, and a post-session signal that compares actual sets × reps × estimated %1RM against the optimal range. Deferred until we have 20+ real LLM-generated weeks audited; a ~8% high-intensity overshoot rate on a small fixture set was borderline-but-too-small to act on.