Half Marathon — Oct 4, 2026

Plan opens after the May 30 5K. May 30 result: 21:35 on a short course (~3.05–3.06 mi, equivalent ~21:55–22:00), avg HR 164 (sub-max effort, well below threshold). Borderline VDOT 44/45 but not a clean benchmark. Holding VDOT 44, plan as written. Re-test at the Aug 22 5K (Wk 12) on a measured course at full effort. Both lifts are light because legs are still recovering from the 5K. Aerobic base only — no Wed quality for 5 weeks.

Scaled up from the original 28 to match true baseline (~33 mpw recent rolling average). Lifts at full load. Sat→Sun mile shift moves long to 11 (Wk 3's target arriving early) — fine given the April 13-miler in your history.

Strides return — short, fast, full recovery. Neuromuscular only, not a workout.

12-mile long run. Sleep is the leading indicator — if two nights of poor sleep, drop a mile from Wed or swap Sat to rest.

Tune-up race week. First quality session shifts to Wk 7 (800m intervals) — racing replaces this week's Wed quality; first T session is Wk 8. Mini-taper only (Wk 5 cutback + light Mon lift); not a full taper. Race by effort, not pace — don't go out at 7:00/mi to bank time. Result feeds into early calibration but the Aug 22 5K (Wk 12) still happens — it captures post-T-block fitness, which is what sets Phase 3 paces. Skip Thu Lift B (race day).

✅ RESULT (Jul 9): 21:47 official on a true 5K (~7:01/mi), avg HR 169 / max 183. Even splits (7:07 / 6:57 / 7:08) with a 6:40 kick — well-executed pacing. Vs May 30 (21:35 on a short course, ~22:00 equiv) this is a clean flat-5K benchmark under summer heat + mid-build fatigue. VDOT 45 (was 44). Note: 1K–1200 workout reps run 6:46–6:53 — faster than 5K race pace — so 21:47 likely still undersells the ceiling; Aug 22 tests whether there's a 46 in there. Paces bumped plan-wide (T 7:25, I 6:52). HMP/goal held at 7:47 / 1:42 — not moving the race goal off a mid-build tune-up.

First I session. 800s at 3K–5K effort (3:26 = 6:52/mi). 3-mi total at I — Daniels' ceiling. Hit 3:26 per rep; if rep 5 or 6 slows by 5+ sec, stop the workout.

✅ RESULT (Jul 15): 6:50 / 6:59 / 6:56 / 6:55 / 6:57 / 6:49 — all reps within ~4 sec of target, no decay, rep 6 the fastest. HR climbed 150→166 avg across reps, max 177 (vs flat-road functional max ~183) — controlled, finished with room. 6:52/mi I-pace confirmed honest, possibly conservative; hold VDOT 45 until Aug 22 per calibration rules. Note for Wk 10 4×1000: recovery jogs drifted long (2:52→4:14 for ~400m, some near-walk) — cap them at ~rep time (~3:30) and keep them a true jog. Thu 5 E skipped for air quality — lift/core/Copenhagen done indoors.

Full sub-T week: ~50 min in the LT zone vs ~25–30 under the old plan, at lower strain per session. Sat sub-T is fine the day before an easy long — it's recoverable by design; if Sunday's HR won't stay under 140, that's the overreach signal and Sat drops to easy next week. RDL adds resume now that Copenhagens have settled.

Lift loads freeze after this week. The 1000s are for turnover and confidence, not fitness — 4 reps, done right, ease into rep 1. By now Wed/Sat sub-T paces at 165 should be visibly faster than Wk 8's; that trend is the block working.

The load-bearing calibration. Course runs long (GPS ~3.24 mi last yr), uncertified — judge by GPS avg pace / best-5K split, never gun time or HR (flat-5K max ~183 is normal for you). Go out at 6:55/mi and find out. Execution fix from Jul 9: spend the kick early — mile 2 is where you leave time; start squeezing at the 2-mi mark; finish with a 15–20s sprint, not a 60s one. Triggers: ≤6:52 held → VDOT 46 (T 7:17, I 6:46, A-goal armed) · 6:53–7:10 → hold 45 · 7:10–7:25 → hold 45, flag heat · >7:25 on a true max grind → 44 (T 7:33) and C-goal conversation. Update once, don't tinker.

✅ RESULT (Aug 22): 21:29 official. Reconciles exactly with the watch — started 6 sec before the mat, line-to-line 5023 m (a true 5K; 2025's was 5.24 km) = 6:53.0/mi, and normalized to 5000 m ≈ 21:23 vs a rolling best-5K of 21:24. Daniels VDOT 46 at 5 km is 21:25, so this clears the table standard on raw pace before any grade adjustment. 21:24 for the 5K split — 6:53/mi raw, 6:49/mi GAP, both inside the gate (≤6:55 raw / ≤6:52 GAP). Avg HR 169 / max 181. Splits 6:56 (HR 157) · 7:01 raw but 6:43 GAP on the climb (HR 164) · 6:51 (HR 168), then 7:10 up the final 0.14. GAP halves 6:55 → 6:51 — negative split on effort, and the fastest 400 m of the race was the last one (6:18 pace). Executed as written: attacked at 2.5, not 2.0.

→ VDOT 46. A tier (1:40:30) ARMED. Paces bumped plan-wide: T 7:17 · I 6:46 · R 6:10 · E 8:31–9:34 · M 7:49 · HMP 7:40. Opening mile at HR 157 against a 183 ceiling means 46 is a floor, not a ceiling; it does not mean run Wk 13 at 47 paces.

⚠️ Course was rerouted vs 2025. Today's track diverges from last year's line from ~0.87 to ~1.45 mi (up to 163 m off); the rest matches within 1–5 m. Same model, same smoothing, both tracks: 2025 = 3.254 mi / 188 ft / cost 1.0150 (+6.2 s/mi) · today = 3.146 mi / 136 ft / cost 1.0062 (+2.6 s/mi). The reroute removed ~0.11 mi and ~50 ft of climb, right on the 1.25–1.75 main climb — which is why mile 2 came in at 7:01 raw / 6:43 GAP instead of grinding worse. So the 6:55 raw gate was priced for a course that wasn't run; recalibrated it tightens to ~6:52–6:55 and 6:53 sits at the edge.

The pass rests on GAP and HR, not raw pace. GAP 6:49 vs ≤6:52 is method-consistent (same Minetti model as grade_pace.exs), and GAP uses the elevation actually run, so a reroute can't corrupt it. Independently: today carried 141 ft of gain vs Jul 9's 92 ft over the same ~3.1 mi, run 11 sec/mi faster at identical avg HR 169 — that comparison never touches the 2025 file. Verdict holds, margin is thinner than it first read: Wk 13's 6 @ 7:40 is a real gate, not a formality.

Block 2 opens; this Sunday is the A-goal test. Aug 22 armed the A tier, so the 6 HMP miles run at 7:40 — controlled and repeatable means the gate passes and Wk 15 rehearses at 7:40; a grind means B (7:47) stands, no shame in the data. Judge it on miles 5–6 of the block, not mile 1. HMP now always rides late: race pace on tired legs is the skill.

✅ RESULT (Aug 30): 6 mi in 44:40 moving = 7:26/mi. Gate passed with room. 14.02 mi total, 2:01:22 (raw block time 45:40 = 7:37/mi — not the number to read; see below).

Rain broke the watch mid-block. At 9.65 mi the wet touchscreen auto-advanced the workout to the final cooldown interval, and switching to the touchscreen-disabled mode took 41 seconds standing still (t=5003–5044 s, HR 165→137). The device logged none of it as a pause — total elapsed 2:01:28 vs moving 2:01:22, just 6 s recorded — so that dead time sits inside mile 4 of the block. Watch laps also ran long, so the block was re-cut from the record stream at t=3328 s / 6.00 mi and stopped time (<1.0 m/s) subtracted per mile. The 16 s at mile 3 is the prescribed water stop (≤20 s per the lap card), legitimate and correctly out of running time:

mi	elapsed	stopped	moving	HR
1	7:40	3 s	7:36	148
2	7:34	—	7:34	156
3	7:45	16 s (water)	7:28	159
4	7:49	41 s (watch)	7:07	162
5	7:18	—	7:17	167
6	7:34	—	7:34	167

GAP ≈ raw on every mile — flat, no downhill assist. Then 2 E. The block beat 7:40 by 14 sec/mi.

The overshoot is modest and converging — not a pacing problem. Decomposed: 7:31/mi over the 3.646 mi with a working pace field (+9 sec/mi), then a 6:36 burst over the 0.33 mi straight off the 41 s standstill (HR had fallen to 137 — a restart artifact, not evidence about feel), then ~7:26 over the remaining 2 mi by feel, closing with a 7:34 final mile — 6 sec/mi off target. Feel converged toward the number as the block went on. Nothing here says perceived HMP is miscalibrated.

The whole block topped out at HR 170 and closed at 167, under LTHR (170–173). Running a goal-pace block slightly fast and never reaching threshold HR is a straightforwardly good result.

Cost, honestly small: ~2 miles landed at T rather than HMP, on top of Wednesday's 4×1 mi T. Wk 14 is a cutback and absorbs it.

Two notes for reading the numbers. (1) The 41 s stop dropped HR to 137 — a partial reset, so the closing 167 is somewhat flattered; six continuous miles at 7:26 would have finished higher. (2) Miles 2 and 6 were both 7:34, at HR 156 vs 167; steady-state to steady-state (mile 3 → mile 6) ≈ −6% decoupling. Fine for a block run 14 sec/mi hot, but it does not establish 7:26 for 13.1. It establishes that 7:40 for 13.1 is comfortably inside current fitness.

→ A tier (1:40:30) holds, more convincingly than the raw 7:37 read. No pace change — VDOT doesn't move off a training run, and Aug 22 already set 46 as a floor. If Wk 15's 8 @ 7:40 also comes in this easy, that is when a more aggressive race-day plan gets discussed (Wk 15 or the Wk 17 taper session). Not before.

Peak week — and the goal locks Sunday. 8 mi at goal pace, late in a 14-miler: pass at 7:40 → A locked (1:40:30). Pass at 7:47 but not 7:40 → B locked. Struggle at 7:47 → C conversation (7:56), honestly. Sunday outranks everything else this week — protect it.

Hold 7:40 — for the drift curve, not to re-prove the pace. Aug 30 settled whether 7:40 is holdable for 6 miles late in a long run: it is, emphatically (7:26, HR topping out at 170). The open question was never 6 miles — it's the extrapolation to 13.1, and what extrapolates is HR cost per mile at goal pace over a longer block. Sunday can't supply that: HR ran 156 → 167 while the pace was also getting faster and absorbing a 41 s stop. Eight steady miles at 7:40 gives a clean drift curve to project from. Program the watch by pace at 7:40, every mile inside 7:35–7:45, even if 7:26 feels free at mile 3. HR is guardrail only — sustained above 175 before mile 6 of the block means the pace is wrong, not the effort. Judge on miles 6–8. Disable the touchscreen before starting, rain or not: a wet screen auto-advanced the Aug 30 workout at 9.65 mi and cost 41 s standing still.

Last quality long run. Everything at the locked goal pace now — no more testing, just grooving. Lifts drop to maintenance permanently from here.