PACER's training methodology: turning running science into an executable plan
PACER doesn't hand out plans by gut feel, and it isn't a chatbot wrapper. It's an AI agent that makes its own decisions: three proven bodies of running science — Daniels VDOT, Hansons cumulative fatigue, 80/20 polarized training — are handed to a self-built kinematics engine and scheduling orchestrator, which reason over your real training data to produce the best personalized race plan for right now.
In one sentence: PACER replaces guesswork with sports-science models. On every data sync the model re-reasons — your plan is always the best answer for this moment.
1. Daniels VDOT and E/M/T/I/R pacing
VDOT, introduced by running coach Jack Daniels, is a measure of your current effective running fitness derived from a recent race or all-out test. With a VDOT value, training intensity can be translated into five precise pace zones instead of vague instructions like "take it easy" or "go faster":
| Zone | Name | Purpose |
|---|---|---|
| E | Easy | Aerobic base, capillary and mitochondrial development, recovery |
| M | Marathon pace | Adaptation to goal race pace |
| T | Threshold | Improve lactate clearance, delay fatigue |
| I | Interval | Develop maximal oxygen uptake (VO₂max) |
| R | Repetition | Improve running economy and speed |
PACER estimates your VDOT from multiple sources in priority order (manual anchor > real all-out test/race > device race prediction > pace-plus-heart-rate estimate > VO₂max fallback), and maps each training day's target pace directly onto E/M/T/I/R.
2. Hansons cumulative fatigue
The core idea of the Hansons method is "cumulative fatigue": rather than relying on single monster long runs, you keep training on legs that are never fully recovered, so you can resist fatigue in the late stages of a race. In practice this means a comparatively restrained long-run cap, volume spread across the week, and careful control of single-session intensity. PACER applies this constraint when designing weekly load and long-run share — avoiding the classic mistake of "blow up on a weekend long run, do nothing all week".
3. 80/20 polarized training
A large body of endurance research and elite practice points to the same distribution: about 80% of training volume at low intensity (easy aerobic) and only about 20% at moderate-to-high intensity. The low-intensity base builds your aerobic engine; a small dose of high-quality stimulus drives the breakthroughs. The most common amateur mistake is exactly "too much moderate intensity" — not easy enough to recover, not hard enough to improve. PACER enforces this ratio across periodization and guarantees hard days are never adjacent, leaving room to recover.
4. Supporting models: fatigue quantification, progression and nutrition
- Banister TRIMP: converts each session's duration and heart-rate intensity into a "training impulse" to quantify fatigue and fitness, preventing overtraining.
- Higdon's 10% rule: weekly mileage generally increases by no more than ~10%, keeping injury risk in check.
- ACSM nutrition & hydration: combines sweat rate and session duration into post-run hydration and fueling guidance.
5. The three-layer decision model: composing theory into daily training
PACER wraps the science above in a three-layer decision pipeline that re-reasons end to end after every sync:
- ① Multi-dimensional state sensing — reads body and environment signals: COROS training data sync, temperature and environment adjustments, sleep and HRV recovery monitoring.
- ② Core computation engines — makes decisions with scientific models: Daniels VDOT for current fitness, Banister TRIMP for fatigue, runner injury medicine for risk alerts.
- ③ Dynamic plan output — produces executable daily training: weekly adaptive plans, pre-race taper, precise hydration and nutrition.
Because decisions are built on your real data, PACER holds one hard line: when data is missing it says so and asks you to sync — it never fabricates mileage, pace, heart rate or medical diagnoses.