run-pace-architect

Use when training for anything from a 5K to a marathon — converts one recent race time into every training pace (easy, threshold, interval, rep) using Riegel-equivalent performances, predicts equivalent race times across distances, checks whether a goal-time is realistic, plans 10%-rule mileage ramps with cutback weeks, and prints even-split tables for race day.

Overview

Most runners train in the wrong zone: their "easy" runs are too fast (so their hard days can't be hard enough) and their hard days are too slow (because they are always tired). The fix is a pace table generated from one honest recent race performance — not from a wish, not from a single good interval session.

This skill is that table generator plus the planning math around it:

1. Equivalent performances — Riegel's power-law model (T2 = T1 × (D2/D1)^1.06) converts one race result into predicted times for every common distance.
2. Training zones — easy, long-run, marathon, threshold (defined precisely as the pace you could hold for 60 minutes, solved numerically), interval (VO₂max) and repetition (speed) paces, each derived from the same model.
3. Goal feasibility — is a 22:30 → 21:00 5K in 8 weeks realistic? The tool compares the required improvement against typical per-block gains for trained vs novice runners and says LIKELY / AGGRESSIVE / UNREALISTIC.
4. Mileage ramping — 10%-per-week cap with an automatic down week every 4th week, simulated week by week until the target mileage is reached.
5. Race-day splits — even-split tables (per-km and per-mile) for a goal time.

When to Use

• "I ran a 5K in X — what should my training paces be?"
• "What marathon time does my 10K predict?"
• "Is my goal of sub-1:45 half realistic in N weeks?"
• "I run 25 km/week, I want to get to 50 — how fast can I ramp?"
• "What pace per km do I need for a sub-X race?" → split tables
• Pacing a training plan: which phase (base / build / peak / taper) to run

Don't use for: injury diagnosis, shoe selection, trail/ultra pacing over wildly varying terrain (Riegel assumes even, road-like conditions), or heart-rate zone calculation without field data.

Quick Start

bash

# All training paces from a recent 5K
python3 scripts/run_pace.py zones --race 5k=22:30

# Equivalent race predictions from one performance
python3 scripts/run_pace.py predict --race 10k=48:00 --to hm,marathon

# Feasibility of a goal + phase structure for the block
python3 scripts/run_pace.py plan --race 5k=22:30 --goal 5k=21:00 --weeks 8

# Safe mileage ramp with cutback weeks
python3 scripts/run_pace.py ramp --current 30 --target 50

# Race-day split table
python3 scripts/run_pace.py splits --race hm --goal 1:45

# Everything in miles
python3 scripts/run_pace.py zones --race 5k=22:30 --unit mi

How the Math Works

Riegel equivalent performances

Riegel (1981) modeled distance-running time as a power law:

text

T2 = T1 × (D2 / D1)^1.06

A 22:30 5K therefore predicts a ~47:00 10K, a ~1:44 half and a ~3:38 marathon. The exponent 1.06 is the classic road-running value; it assumes even pace, comparable terrain and adequate training for the target distance. Predictions degrade when you jump more than ~4× in distance (a 5K predicts a marathon poorly for an untrained runner).

Training paces (all derived from the same model)

Zone	Definition used	How it's computed
Easy / aerobic	Comfortable conversational	Marathon pace + 45–75 s/km
Long run	Easy effort, extended	Easy pace (upper half of range)
Marathon	Race pace for 42.2 km	Riegel prediction at 42.195 km
Threshold	Pace holdable for 60 min	Distance solved numerically so Riegel time = 3600 s
Interval (VO₂max)	~3K–5K race effort	Riegel prediction at 3 km
Repetition (speed)	~1500 m race effort	Riegel prediction at 1.5 km

The threshold pace is solved, not guessed: the script binary-searches the distance D where your predicted time equals exactly 60 minutes, then reports that pace. That is the textbook definition of threshold (one-hour race pace).

Feasibility of a goal block

Required improvement = (current − goal) / current. Rules of thumb encoded:

• Trained runners: 1–3 % per 8–12 week focused block is realistic.
• Novice/returning runners: up to 5–8 % is possible (bigger fitness headroom).
• More than that in one block → verdict AGGRESSIVE (extend weeks) or UNREALISTIC (split into two blocks with a race in between).

Block structure output for N weeks: base (~50% of weeks, volume + easy strength), build (~30%, key sessions get race-specific), peak (~20% minus taper, sharpening), taper (final week: −30–40% volume, intensity kept). Every phase lists its purpose and the key session shape.

Mileage ramp safety

• Never increase weekly volume more than 10 % week-over-week.
• Insert a cutback week (−20 %) every 4th week — adaptation happens during recovery, not during accumulation.
• The tool simulates this rule week by week and prints the schedule, including how many weeks the ramp actually needs (usually more than people expect).

Common Pitfalls

1. Feeding it a training-run time. Garbage in, garbage out. Use a real race or an all-out time trial on a measured course, in race conditions.
2. Easy pace creep. If easy pace feels "too slow", that's correct — elite runners do the majority of volume slower than marathon pace.
3. Trusting a 5K → marathon prediction without marathon training. The model assumes you can actually cover the distance; the prediction is "with appropriate training", not "tomorrow".
4. Ramping volume and intensity simultaneously. The 10 % rule applies to volume; add quality only after volume has been stable for 2–3 weeks.
5. Even splits ≠ negative splits. The split table assumes even pacing; going out 10 s/km faster than plan is the classic marathon blow-up recipe.

Verification Checklist

• Input race time parses (5k=22:30, 10k=48:00, marathon=3:15:00)
• Zones table prints all five zones in per-km AND per-mile
• Threshold pace sits between marathon pace and interval pace
• Feasibility verdict matches required-% vs the 1–3 % / 5–8 % bands
• Ramp table includes a −20 % cutback every 4th week and never exceeds +10 %
• Splits table sums exactly to the goal time

Examples

text

$ python3 scripts/run_pace.py zones --race 5k=22:30
Reference performance: 5K in 22:30 (4:30/km)
  Easy          5:26–5:56 /km   (8:45–9:33 /mi)
  Long run      5:41–5:56 /km
  Marathon      4:50 /km        (predicts 3:24:30)
  Threshold     4:40 /km        (your 60-minute pace)
  Interval      4:22 /km        (3K–5K effort)
  Repetition    4:04 /km        (1500 m effort)

$ python3 scripts/run_pace.py plan --race 5k=22:30 --goal 5k=21:00 --weeks 8
Required improvement: 5.6% in 8 weeks → AGGRESSIVE for a trained runner
(likely OK for a novice; consider 10-12 weeks)
  W1-4  BASE    volume +, strides 2×/wk
  W5-6  BUILD   VO₂max intervals + threshold
  W7    PEAK    race-pace specific
  W8    TAPER   −35% volume, keep strides

Data and formulas: Riegel fatigue model; threshold-as-one-hour-pace convention; widely used 10 %-rule and 4th-week-cutback practice. Individual response varies — the numbers are starting points, not prescriptions.