The conversation

What's working. First, your legs are changing, and this is the most important thing in the diary. On Tuesday's 7 km at 8:10/km your cadence was 166 spm and ground contact 246 ms. On 21 October, at the same pace, it was 158 spm and 308 ms. Your feet now spend 60 ms less on the ground per step, which is the first hard evidence that the Monday gym is fixing the thing that failed in both marathons. Second, your engine has room to spare: the 5 km gate went in 31:07 at 146 bpm, which is tempo effort rather than flat out, and the 20 km went at 7:57/km in 25 °C and 93% humidity at about 135 bpm. Third, you hold your pace when it's set for you. Wednesday's threshold reps all landed within 10 s of the band, the last one fastest, and your easy days are now actually easy. Fourth, you need 7:56/km to make the 5:35 cutoff. You already run long runs at that pace in Bangkok heat, and the same effort should come out 20–35 s/km quicker in a Kyoto February.

What worries me. Consistency, not fitness. Weeks 3 and 6 dropped to about 16 km, right after your two biggest weeks, which is the same crash pattern your history shows. The repeat rule has fired three times, so the plan has lost about three weeks of progression. That is why your longest run is 20 km in week 9 when it was meant to be 24. Next, you start every long run too fast: weeks 7, 8 and 9 all went out quicker than the 8:15 I asked for. It cost nothing at 20 km. At 32 km and on race day it's how Osaka faded, with pace dropping while heart rate stayed flat. Nothing past 30 km has been tested yet, and both marathons fell apart between 30 and 35 km. Your ground contact still drifts up late in long runs (272 to 283 ms yesterday). The drift is smaller, but it's the same signature, and weeks 15, 18 and 21 (30, 32 and 32 km) are where we find out. Last, gym quality is hard to see. Week 9's gym shows 31 minutes against 55 planned with no sets logged, and that is the one session I'd never cut.

One thing in the plan I'd fix. The plan has week 10 at 45 km and week 11 at 57 km. That's a 27% jump, and it breaks your own 10% rule. I'd hold week 11 nearer 50 km and keep the 24 km long run.

My view: sub-5:00 is alive on paper but not proven, and it stays that way until the gates in weeks 13–19. Finishing under the cutoff is the realistic goal right now, and it's very achievable if two things hold. No more 16 km weeks, because the fitness is there and missed weeks are now the biggest risk. And your 30+ km long runs start at 8:15 or slower, with a gel every 40 minutes from this week on. Do those and I think you'll cross the line in Kyoto this time.

What it probably is. Pain just below the kneecap that comes on under heavy leg press is most often the patellar tendon complaining about load. The same tendon takes every downhill in Kyoto. It has now happened twice on the same machine, and the second time came with a jump in weight, so the weight is the likely trigger, not the exercise itself. Your shin was also sore in week 2, so check whether it was the same leg and the same spot.

Tomorrow's 7 km. Before you run, test the knee: go down a flight of stairs, then do 5 slow squats to chair height.

• No pain, or 1–2/10: run it easy, flat route, heart rate under 140.
• 3/10 or more, or it hurts when you walk: skip the run and rest. One missed easy run costs you nothing.
• During the run: stop if it goes past 3/10, or if it gets worse as you go rather than settling.

Leg press from now on. 1. Drop back to the weight you used before today, then go 20–30% lighter than that for the next two sessions. 2. Put your feet higher on the platform and stop at about 90° at the knee. Going deeper with feet low puts the most load on that tendon. 3. Add weight by no more than about 5–10% a session, and only when the last session caused no pain that day or the next morning. The 10% rule applies to the gym too. 4. If it still hurts at light weight, swap leg press for a wall sit (30–45 s, held at a comfortable angle) until it settles.

When to see a physio. See one if there is swelling, if stairs keep hurting, or if it is not clearly better within a week. The 16 km on Sunday depends on how the knee feels on Thursday and Friday, so tell me after tomorrow's run.

What the data says.

• Your load jumped. Week 3 was 15.7 km and week 4 was 36.5 km. Your 7-day load against your 28-day average reached 1.58 on 3 Oct and is 1.45 today. Above about 1.5 is where injury risk starts to climb. That jump came from coming back after the rained-out week, not from being fit for 36 km.
• Your shin was sore in week 2. That is the spot I'm watching.
• Your legs tire late in long runs. On Saturday's 14 km, ground contact rose from 282 to 305 ms. This is the Osaka problem starting to show.
• There are good signs too. Cadence has risen from 151 to 163, and ground contact on shorter runs dropped from 270 to 255 ms. Today you ran 7:45/km with heart rate between 128 and 142. Your form is getting better.

About the pace. The pace bands only stand in for effort. The rule that matters is heart rate under 140 on easy days. If you run 7:45 and stay under 140, that is an easy run for you right now. Don't make yourself shuffle into an unnatural stride to hit 8:15. Shuffling changes how your legs take the load, and that can hurt more than the extra 20–30 s/km.

What I recommend. 1. Go by heart rate on easy days, not pace. Run with your natural stride. If heart rate passes 140, slow down or walk until it drops to 130, then run again. 2. Use walk breaks to slow down. If you can't run slower, run normally and walk 1 minute every 8–10 minutes. Your average effort drops and your form stays the same. 3. Long runs are the exception. Keep heart rate under 135 for the first two-thirds of Saturday's 16 km, and use walk breaks to do it. The long run is where pace matters, because that is where your legs failed in both marathons. 4. Don't miss tomorrow's gym session. Only three strength sessions are logged since September. Gym work is what protects your shins and makes your legs last. 5. Listen to your shin. Stop the run if it goes past 3 out of 10. If it hurts when you walk the next morning, take a rest day. 6. Week 7 jumps too much. It goes from 44 km to 51 km, which is +16% and breaks the 10% rule. I'd hold it at 48 km. Tell me and I'll change it.

In short: your engine is ready for the training. Your legs and tendons need more time, so how much you add each week matters more than how fast you run.

What went right.

• You ran the whole 14 km and stopped there. You didn't chase 15, and after last week's 8.1 km in the rain, this is the long run that counts.
• Heart rate stayed aerobic. It averaged 123 bpm for the first third, 137 for the middle and 139 for the last, under 140 until km 12, and only touched 145 in the final kilometre. That's a normal climb, not a run that got away from you.
• Cadence held. While you were running it was 161, 160 and 160 spm across the three thirds, right in the 160–165 long-run target. It didn't fade at the end, which is what I was watching for.

The problem: the pace. Strava's 7:51/km average includes about 16 minutes of walking and stopping (1.3 km of the 14). The 12.8 km you actually ran went at 7:37/km. That's 18 seconds faster than the 7:55–8:25 band, the same as last week's 7:34. The breaks made the average look right while the running stayed too fast. I wrote it up too kindly in the week 4 note at first and have corrected it.

The signal worth watching: ground contact. Your foot stayed on the ground about 282 ms per step in the first third, then 301 and 305 for the rest. Cadence and pace didn't change, so the extra 20 ms is your legs getting tired: they push off the ground more slowly. That's how Osaka and Kyoto 2024 went in the last 10 km. Here it showed up after 25 minutes, and part of the reason is that the running was faster than it needed to be. Running in the band is what lets the legs last. The Monday gym is the other half.

For next Saturday's 16 km: 1. Judge pace by the running, not the average. If the watch shows faster than 7:55 while you're moving, ease off. In Bangkok heat, 8:00–8:10 is right. 2. Walk breaks are fine. Take them for water or a gel. Just don't run faster in between to make up the time. 3. Rehearse the gel at 45 minutes, if you didn't today. From week 9 it's a gel every 40 minutes on every long run, and your stomach needs practice before then. 4. Send the screenshots again. If ground contact rises later in the run, or by less, at a slower pace, that's the legs getting stronger. That's the number I want to see move.

Overall: the effort and the cadence are right. The pace while running and the legs late in the run are what this block is for.

Why not 180? That number came from elite racers running at about 3:00/km. Cadence rises with speed, so at your easy pace of 8:10–8:45/km, 180 would mean tiny shuffling steps. That tires you out and raises your heart rate without helping your legs.

Where you are. Your last four readings were 151, 155, 159 and today's 164. You already have a sensible target in reach. The usual advice is to raise cadence about 5% above where you started (about 155), which gives roughly 163.

The targets. 1. Easy runs (8:10–8:45/km): 160–166. Today's 164 is right in range. Just stop it pulling your pace down to 7:38. 2. Long runs (7:55–8:25/km): 160–165. A drop of 2–3 spm in the last few km is normal. A drop below about 155 means the legs are going. 3. Strides and faster work: 175+. Cadence climbs naturally with speed. Don't hold it down.

Why the long run matters most. At Osaka and at Kyoto 2024, your pace fell apart in the last 10 km while your heart rate stayed steady. Legs that are giving out usually show it as a falling cadence and a longer, heavier stride. If you can still run 160+ at 30 km, the gym work is paying off, and that matters more than any easy-day number.

How to practise it.

• Take shorter steps; don't just move your feet faster. If the pace goes under 8:10, shorten your stride again.
• Give it two or three weeks. Your calves and shins do more work at a higher cadence, which is part of why km 5 and 6 felt hard today. It gets easier.
• Don't push it on the day after gym. If your legs are heavy on a Tuesday, 158–160 is fine.
• Ignore how you feel in the first km. Today's first km averaged 9:21 including a stop, so judge cadence from km 2 onward.

Keep sending the Apple Health screenshots. With a few more weeks of readings we'll be able to see whether the long runs hold 160+ to the end.

Why the plan holds. 37 km against the 34.0 km you ran is +8.8%, inside the 10% rule. You hit every week 2 session and none were missed, so this is a progression week and not a repeat. Heart rate stayed under 140 all week; the long run averaged 127 bpm and drifted from 119 to 133 across the run, so the engine isn't the constraint. The shape is four running days, Friday and Saturday off, Sunday 14 km, which matches your rules. The 14 km is +2 km on the long run, and a gel at 45 minutes fits it.

The two conditions. 1. The shin decides this whole week. Keep the plan only if it was still 3/10 or less after Sunday's 12 km and quiet the next morning. If it was worse, hold the week at 34 km, as the coach note already says. 2. Wednesday's 8 strides should be gated on the shin. Week 2 dropped strides because of the ache, and this is the first week they come back. If the shin is anything above a faint 1-2/10 by Wednesday, skip them and run the 8.5 km easy. They're the one sharp element near a niggle, and they're optional.

One thing to watch. Monday's gym is your first session on the machines, and Tuesday's run follows it. Hold the pace band on heavy legs. Keep the first session conservative on weight, and especially on calf raises given the shin. You want soreness that fades by Tuesday, not a spike. If it doesn't, Tuesday becomes a slower easy run, not a skipped one.

So no change to the numbers, only the shin-based fallbacks.

Why the gap is that big. Bangkok evenings run ~30°C and humid; Kyoto in late February is ~3–10°C. At the same effort (same heart rate, same physiological stress), heat slows you down — more blood gets diverted to your skin to dump heat instead of to your legs, plasma volume drops, core temperature climbs, and perceived effort goes up for identical pace. race.json's own note puts that at 20–35 s/km, and 7:35 (Bangkok MP midpoint) → 7:07 (Kyoto) is a 28 s/km gap — right in that band. So 7:30–7:40 isn't a different, slower goal from 7:07 — it's the same target, same effort, just measured in a hotter room. That's also exactly why your easy days are capped at 140 bpm rather than a pace: heart rate is the honest number when the thermometer keeps moving on you.

It's not just a tax you pay, either. Training in real heat also does something real to your body — it expands plasma volume and improves how you dump heat, and that adaptation carries over even when you race somewhere cool. So the Bangkok pace isn't only "slower because it's hot," part of it is banking fitness a Kyoto-based runner training in Kyoto-like temperatures wouldn't get for free.

How you actually get to be sure, rather than just trusting the formula: you don't — not from a spreadsheet. That's the entire point of the four decision gates (weeks 7, 13, 15, 19). They're run in the same Bangkok heat but graded against fixed clock times (5 km ≤ 31:00, 10 km ≤ 65:00, etc.) that already assume you'll get the Kyoto discount on race day — so hitting them in the heat is a harder ask than race day itself, which is your margin. Week 15 is the one that matters most for this exact question: 30 km with the last 10 km held at marathon pace. That's not a formula, that's you finding out in training whether 7:30–7:40 actually holds late in a long run, which is precisely where Osaka and the 2024 DNF both fell apart. If it doesn't hold there, the plan steps the goal down rather than pretending the number is real.

Rain vs. snow are different problems, and worth not lumping together:

• Rain, even fairly heavy, costs you very little pace directly — a handful of seconds for wet footing, more if there's wind on top of it. Cool and wet is honestly not far off ideal marathon weather physiologically.
• Cold combined with wet is a hypothermia and kit problem before it's a pace problem — that's exactly why "Cold-weather kit" is on your race checklist already: you've never raced in single digits, so gloves, sleeves and a throwaway top need testing beforehand, not discovering on the day.
• Snow/ice is the one that actually should change your pace, deliberately — the course climbs at Arashiyama in the first 10 km and again near 38 km, and on ice the right move is to protect the finish over the split. On a bad-footing day, the FLOOR pace (7:56/km, the 5:35 cutoff) is the number that matters, not 7:07 — and there's about 49 seconds per kilometre of room between your marathon-pace goal and that floor. That gap is your actual weather insurance.

So: trust the 7:30–7:40 as the real target for now, but treat it as a hypothesis the gates are there to test — not something the plan (or I) can promise from a temperature chart alone.

Two reasons it's reading lower than usual: you're on a treadmill under AC rather than in Bangkok heat, which is the exact load the pace bands are built to account for; and a flat belt has no air resistance, so the same pace costs less effort indoors than out. I wouldn't chase 140 by pushing the pace — the shin niggle already has strides off this week, and closing the gap by running faster works against the thing the ceiling is protecting. If it bothers you that the session doesn't feel like enough, add a 1% treadmill incline instead — the standard fix for the missing wind resistance, and it loads the calves rather than the shin. Keep the 140 ceiling regardless of what gets you there. If the low reading persists once the shin has settled and you're back outside in the heat, that's worth a note in the week's write-up as a fitness signal — for now it's the plan working, not you falling short of it.

Bread is a good choice — simple carbs, easy to digest, low fat and fibre. Two things to tune next time: eat it 60-90 minutes before you start rather than right before, so it clears your stomach before pace; and this is separate from the long-run fuelling rule (a gel every 40 minutes) which only kicks in on long runs from week 9, where you're depleting glycogen over 90+ minutes — a 7-8 km easy day just needs enough in the tank going in. A small carb snack an hour or so before an evening session, especially after skipping a meal, is a sensible habit, not a one-off fix.

The treadmill's number comes from belt rotations, a mechanical count independent of how you move, so it's the one to believe. Heart rate is a separate sensor, so today's 125 bpm average — well under your 140 easy ceiling — holds regardless. Clean easy day. Read the treadmill console next time and give me those numbers when I do the weekly refresh, so the log uses the real distance instead of the inflated one.