1. Per-run metrics
- Max HR. Your input, otherwise the 95th percentile of the highest heart rate of each run (190 bpm if there is no data). Rare hard efforts make it too low.
- Resting HR. Your input, otherwise 55 bpm.
- Averages. Time-weighted. Each sample holds until the next one, with gaps capped at 30 s.
- HR zones. Upper bounds at 60, 70, 80 and 90% of max HR. Zone 5 is above 90%.
- TRIMP (Banister). Sum over time of minutes × HRR × 0.64 × e1.92 × HRR, with HRR = (HR − rest) / (max − rest), limited to 0–1. Without HR samples the workout average is used.
- Efficiency factor (EF). Speed in m/min divided by average HR. Higher means fitter.
- Aerobic drift (decoupling). (EF first half − EF second half) / EF first half × 100. Above about 5% means the aerobic base is not yet stable.
- Cadence. Steps per minute from step intervals of 5 s to 5 min. Values outside 120–230 are dropped.
2. Training load
- CTL (fitness). Daily TRIMP averaged exponentially over 42 days. It builds slowly with steady training and fades slowly with rest. See Training impulse.
- ATL (fatigue). The same over 7 days. It jumps after a few hard days and drops within a week of rest.
- TSB (form). CTL − ATL after the latest day's load, as shown in the Assessment. Negative means fatigued. Positive means fresh, for example after a taper (tapering).
- Example: one run of load 100 on a day when CTL and ATL are 0 adds about 14 to ATL (100/7) but only about 2.4 to CTL (100/42).
- Days and weeks use UTC. Weeks start on Monday.
3. Fitness estimates
- Pool. Runs in the selected data range (default 12 weeks) with 1 km or more. Hard efforts are runs of 15+ min with average HR at 82% of max or more. With none, all runs of 15+ min are used and confidence is low.
- Riegel prediction. T10k = T × (10000 / distance)1.06 (Peter Riegel). The 5 lowest predictions in the pool are used, and their median is the 10 km estimate (adjusted Riegel). Plain Riegel scales one run, usually your best, and one lucky or short run makes it too fast. The median of 5 ignores one or two outliers and still follows your fitness.
- Threshold pace. Predicted 10 km pace × 1.03, roughly the lactate threshold pace. The goal field overrides it.
- Easy pace. Median pace of runs with average HR at or below the easy HR ceiling (70% of max). With fewer than 3 such runs: threshold pace × 1.25.
- Easy HR ceiling. 70% of max HR.
- Weekly volume. Distance of the last 4 weeks of the data range divided by 4. The data range is at least 28 days.
- Predicted race time. The same Riegel formula applied to the race distance. It is compared with your target time: more than 3% faster is ambitious, more than 8% faster is likely out of reach.
4. Adaptive rules
| Rule | Trigger | Effect |
|---|
| Fatigue | ATL > 1.5 × CTL | Volume × 0.8. The next quality session becomes an easy run. |
| Easy run heart rate | 2 or more runs in 14 days that are not hard efforts (average HR below 82% of max) and at least 15% slower than threshold pace, with average HR above easy ceiling + 3 bpm | Easy pace +10 s/km |
| Aerobic drift | Median drift of 60+ min runs (28 days, at least 2) above 5% | Weekly ramp 0% (hold volume) |
| Efficiency trend | Median EF of the last 28 days at least 2% above the 28 days before (3+ runs in each) | Weekly ramp 10% instead of 8% |
| Recent activity | No runs in 7 days | Message only. The plan restarts from the 4-week average. |
| Pace confidence | No hard effort in the data range, or average hard-run pace more than 1.12 × threshold pace | Message only. Confidence is low. |
5. Plan generation
- Length. From the current week to the week of the race.
- Volume. Starts at max(weekly volume, 10 km) × volume factor. Grows 8% per week (see rules). Every fourth week is a cutback at 80%. The cap is 35, 50, 65 or 90 km per week for races up to 5 km, 10 km, half marathon and longer.
- Taper. 1 week for races under 15 km, 2 weeks up to 30 km, 3 weeks above. At most one third of the plan. Taper weeks run at 75%, 60% and 45% of peak volume.
- Phases. The last 20% before the taper is peak. The first half of what remains is base, the rest is build.
- Days. Quality on Tuesday, long run on Sunday. 3 to 6 days per week fill the other slots with easy runs.
- Long run. 30% of weekly volume, at least 3 km, capped at 12, 16, 22 or 32 km by race distance. With few days per week it grows beyond 30% (up to the cap) so the week reaches its volume. Pace is easy pace + 15 s/km.
- Quality session. 20% of weekly volume. Base: easy run with 6 × 20 s strides. Build: tempo and intervals alternate. Peak: intervals, or tempo for races of 30 km and more. Tempo is half the distance at threshold pace. Intervals are 1000 m repeats (about half the distance) at 95% of threshold pace with 90 s jog.
- Target time. Race pace = target / distance. In peak weeks, every other quality session is tempo work at race pace, but never faster than 3% below the predicted pace. Race day shows the pace and the target.
- Volume check. Easy runs are at most 75% of the long run, and the long run is capped by race distance. These limits set the most weekly volume a number of days can hold (for a marathon: 70 km on 3 days, 100 km on 4). The weekly target is limited to that value and the plan shows a warning.
- Easy runs. The remaining volume divided over the easy days, at least 3 km, at most 75% of the long run. The week total shown is the sum of its sessions.
- Race week. Sessions stop two days before the race, with no long run. The race is the last session.
- Safety check. A plan is rejected if a session is more than 5% above the long-run cap.
6. Recommendations
Shown at the top of the Assessment. Runs of 1 km or more only. Data range = runs between the two borders of the range slider in the sidebar. The newer border is "now" for all numbers, so you can look at your fitness as it was on an earlier day. The plan still starts today. All numbers on every tab use only these runs. Last 28 days = runs in the 28 days before the end of the data range. With fewer than 3 runs in 28 days only a notice is shown.
| Recommendation | Trigger | Value shown |
|---|
| Recover first | ATL > 1.3 × CTL, or TSB < −20. Replaces the pace target and the freshness note. | TSB, ATL, CTL |
| Faster pace target | 2 or more hard runs in the data range. Hard run = 15+ min with average HR at 82% of max HR or more. | Average pace = total time / total distance of the hard runs. Target = average pace × 0.98 (2% faster), for one hard run per week. |
| Add a hard effort | Fewer than 2 hard runs in the data range | One steady 20 to 30 minute effort per week |
| Many long runs, little intensity | More than 4 runs of 75+ min in 28 days, and less than 10% of time in zones 4 and 5 | VO2max intervals: 5 × 3 min at threshold pace × 0.95 |
| Almost no high intensity | Less than 5% of 28-day time in zones 4 and 5 (and the rule above did not apply) | One interval session per week, same pace |
| Too much moderate and hard running | Less than 70% of 28-day time in zones 1 and 2 (and neither rule above applied). The target is about 80%. | Share of time in zones 1 and 2 |
| Long run too large | Over the weeks of the last 28 days with runs (at least 3): average of longest run / weekly distance is above 45% | That average. The plan uses 30%. |
| No long run | No run of 75+ min in 28 days and at least 8 runs in 28 days (and the rule above did not apply) | Run count |
| Run more often | Runs in 28 days / 4 is below 3 | Runs per week |
| Rebuild volume | Last 4-week average km per week is below 75% of the range average (range distance / range weeks) | Both averages. Rebuild about 10% per week. |
| You are fresh | TSB above +10, with CTL above 0 and no fatigue trigger | TSB |
Time in zones = sum of time per zone over the runs. Zone bounds are in section 1. The percentage thresholds are fixed rules of thumb, not personalised values.
The plan is a pure function of goal, data and date. It is recalculated whenever the data or the goal changes. This is a rule-based tool, not medical advice.