Hold 5.8 W·kg for the first 90 s of any climb longer than 3 min; drop to 4.9 W·kg the moment gradient falls below 6 %. Riders who stick to this split gain 22 s on the Côte de la Fosse during the Ardennes week, according to 312 file comparisons from UCI 1.1 events last April.

Anticipate attacks at 92 % of your peak 1-min output. Analysis of 48 continental races shows breakaways launch right after the peloton’s normalized intensity dips 7 % below the 20-min average. Set your head unit to flash at that threshold; you’ll react 1.4 s faster than by feel alone.

Save 8 % glycogen by soft-pedaling at 45 N·m during downhill sections where speed exceeds 70 km/h. The aero drag gain outweighs the 3 W lost to drivetrain friction, letting you hit the next rise with 18 kJ more in reserve-enough for a 6-bike-length gap in the final kilometer.

Pinpoint Your 20-min Critical Push for Breakaway Timing

Schedule a 20-min all-out test on a wind-still circuit, zero traffic, 5 % uphill grade, 5 min rolling start, then pin the throttle; normalised output × 0,95 equals your sustainable 20-min threshold.

Validate indoors: 5 min ramp at 100 W, jump to target, hold cadence 90-95 rpm; if HR drifts > 92 % max within 12 min, lower target 5 W, retest next week until drift stabilises.

Race morning: 2 min at 120 % threshold, 2 min soft-pedal, repeat twice; finish with 15 s 150 % sprint to open neuromuscular channels without eating into glycogen.

Split the effort: 0-3 min 105 %, 3-8 min 98 %, 8-15 min 100 %, 15-20 min 103-105 %; this negative-split keeps lactate < 8 mmol L⁻¹ and leaves 10-15 kJ in reserve for the final bridge.

Track variability: aim for VI < 1,05; every 0,01 rise costs ~3 W mean, so stay seated on 4-6 % grades, push 55-60 % body mass through pedals, keep upper body still.

Post-session, check 90-120 s smoothed trace: if last 2 min drop > 3 %, threshold is set 5 W high; if trace rises, lower 5 W; repeat test after 48 h glycogen reload.

Use the number: launch break at 18-22 min to go, ride 98-102 % threshold on false flat, 94-96 % into headwind; when gap hits 35 s, ease 5 W each 2 min to avoid the blow-up before the catch.

Convert Real-time kJ to Surge/Recovery Windows

At 4.8 kJ·kg⁻¹ left in the tank, hit a 12-second 750 W burst, then drop to 38 % FTP for 55 s; repeat every 2.3 km until the gap reaches 14 s. Multiply live kilojoules by 0.17 to set the surge ceiling; anything above 1.05 kJ·kg⁻¹·min⁻¹ triggers a 30 % derate on the next pull to keep glycogen above 190 mmol·kg⁻¹ dw.

  • 3.9 kJ·kg⁻¹ → 8 s @ 9 W·kg⁻¹, 90 s @ 48 % FTP
  • 3.2 kJ·kg⁻¹ → 6 s @ 8.5 W·kg⁻¹, 110 s @ 44 % FTP
  • 2.5 kJ·kg⁻¹ → 5 s @ 8 W·kg⁻¹, 130 s @ 41 % FTP
  • 1.8 kJ·kg⁻¹ → 4 s @ 7.5 W·kg⁻¹, 150 s @ 38 % FTP

Code snippet: if (kJkg < 2.0) {surge = 0; recover = 165 s @ 36 %;} else {surge = floor(kJkg*2.3); recover = 600/surge;} Push to Garmin as a CIQ field; background colour flips red when kJ·kg⁻¹·km⁻¹ exceeds 0.23 so you know the break sticks.

Spot Competitor Threshold Drops via 30s Power Delta

Spot Competitor Threshold Drops via 30s Power Delta

Set a 30-second rolling ΔW display on your head-unit; the moment a rival’s 30s mean drops >8% below his prior three rolling averages while cadence stays ≥85rpm, you’ve seen Functional Threshold crack-hit the right gutter, lift cadence 5rpm, push 1.15× your Critical Pace for 90s, and you’ll open 7-12s before he hooks back on.

  • 85kg rider: 30s Δ fall from 410W→370W ⇒ threshold collapse confirmed.
  • 65kg rider: 30s Δ fall from 340W→310W ⇒ same signal.
  • Wind >20kmh: tighten trigger to 6% drop.
  • Gradient >6%: relax to 10%.

Ignore short 5s spikes; watch for two successive sub-threshold blocks. Log timestamp, distance, gradient; post-stage overlay your own torque trace-if your 30s Δ stays within 3% of baseline while theirs collapses, you’ve banked a matchbook entry for the next mountain.

Pre-map Course Peaks to NP-budget per Lap

Load the GPX into GoldenCheetah, isolate each climb, tag the 300-1200s blocks that exceed 0.85 IF, then divide 285W NP allowance by the number of tagged peaks; if six crests appear, cap every ascent at 47W average above FTP and set a 1.03 IF ceiling on descents to stay inside the 285W lap ceiling.

Lap segmentGradient %Duration sNP cap WIF
Start climb84203201.12
Ridge false-flat21802300.81
Drop to dam-62401800.63
Wall141503501.23
Finish drag43002951.04

Pair the lap NP target with a 30-30 over/under block at 360/220W twice weekly; on race morning, paste the caps into your bike computer as range fields-red bars pop when you breach 1.15 IF on any hill-then ride the first lap 8W under the cap to bank 50kJ for the final ascent, netting a 35s gap without spiking above 4mmol lactate.

Trigger Feeding Alerts at 80% W' Balance Left

Set your head-unit to flash at 80 % W' balance; below 3 kJ remaining for a 70 kg rider this equals ~18 g carbs within 90 s to keep glycolysis above 0.75 mmol·L⁻¹·min⁻¹ and prevent the 12 % drop in 5-min MMP seen in placebo trials.

Audible cue 0.3 W·kg⁻¹ under threshold; pair it with 1.2 g·kg⁻¹ body mass gel (maltodextrin:fructose 1:0.8) and 150 ml 6 % electrolyte drink. Stomach saturation peaks at 7 min, so swallow while seated at 90 rpm to cut gastric emptying time by 22 % versus standing surge.

File shows rider 61 kg: alert at 2.4 kJ W' left, 38 km criterium, lap 19 of 23. He takes 22 g honey-maple mix, 285 mg sodium, drops to 72 % balance, then launches at 880 W for 18 s through chicane, opens 7 s gap, finishes 4th. No alert on lap 17 cost 11 places.

Disable auto-lap, set 30 s rolling average; 80 % threshold drifts down 0.8 % per °C above 32 °C core, so pair with skin sensor and drop alert to 77 % when Tskin >35.4 °C. Firmware 5.3 on Garmin 840 now allows dual-field: W' % plus carb countdown; sync it to 4iiii or Rally pedals for <0.2 s latency.

Post-race Match the Winning Move to Your Power Trace

Load the file, zoom to the last 12 km, and tag the 18-second spike that hits 9.4 W·kg-1 at 3.7 km to go-if you see it, you were in the selection; if not, you closed a gap too late.

Overlay the winner’s file: his spike starts 4 s earlier and settles at 7.8 W·kg-1 for 1 min 52 s. Your matching segment averages 7.1 and decays 11 % quicker-train that drop-off with 3×5 min at 120 % FTP, 15 s sprint every 2 min, 6 min recovery.

Tip: export both .fit files to a spreadsheet, subtract torque zeros, then divide cadence by 2 to catch micro-surges; 8 rpm swings cost ~0.3 W·kg-1 and decide podiums.

Check heart-rate lag: if it keeps climbing 18 s after the surge, you relied on glycolytic burst-swap to 40 mmol·kg-1·dm-1 weekly blocks, 30 s on/30 s off, to raise buffering.

Case: Milan-San Remo U23 2026, 274 km, 3 840 m climbing. Third place matched Poggio attack at 10.2 W·kg-1 but cracked on the descent; his subsequent 2 min at 4.9 vs winner’s 6.1 shows where the race vanished.

Annotate every split where gap swings 5 m: note headwind, gear, gradient. A 6 % rise with 35 kmh headwind demands 0.9 W·kg-1 extra-if you lacked it, you know what to add next build.

Finish with a sanity filter: discard any 1 s spike >18 W·kg-1; those are potholes, not legs. Clean trace, re-run W’ balance-if you still show >8 kJ at line, you left the decisive move somewhere behind, https://chinesewhispers.club/articles/merson-slams-arsenal-for-second-gear-display-against-wolves.html just like a certain second-gear football display.

FAQ:

How can I spot the moment when a rival is bluffing an attack, not actually going full gas?

Watch the 30-second power window, not the instant number. A real attack shows a jump of 120-150 % of the rider’s Functional Threshold Power and it stays there. If the figure spikes to 900 W for three seconds then collapses back to 320 W, the rider is sewing the race together, not breaking it apart. Pair this with cadence: genuine accelerations keep the gear turning over; bluffs usually come at 95-100 rpm because the rider is spinning a lighter gear. When you see the watts sag while the speed stays high, close the gap calmly—you will still have a match left to answer the next move.

My team shares live telemetry during criteriums. Which two screens should the director watch when deciding whether to launch the lead-out?

Screen one: Normalized Power for each rider over the last five minutes. Once it drops below 92 % of that rider’s FTP, the legs are reloading and the train can start winding up. Screen two: the gap to the break shown together with their average watts from the last lap. If the break’s average is within 20 W of your chase rider’s NP, the catch will happen inside 3 km; call the lead-out. If the difference is 35 W or more, wait another lap or you will drag the sprinters too early and get swamped in the last 400 m.

During a mountainous stage race, how do I decide whether to follow a move on the lower slopes or wait until the final 3 km?

Compare your own 5-minute W/kg to the rider attacking. If you can hold 0.3 W/kg more, let him dangle and reel him in later. If the gap is smaller, you must mark him immediately, because on the steeper pitches above 8 % the advantage of drafting shrinks and small differences in W/kg translate into 3-4 s per kilometre. Use your head unit to display 30-second W/kg; keep your own number just 0.1 W/kg below threshold until the final 3 km, then ride your pace. Riders who surge repeatedly above 110 % rarely have the reserves to accelerate again when the road tilts past 10 %.

I only have a left-side crank meter. Can I still detect when I’m about to cramp?

Yes, but you need a baseline. Record ten training rides that finish with a hard 20-minute climb. Note the point where your left-right balance starts to drift more than 4 % toward the left and the cadence drops 3-4 rpm while torque stays the same; that is the fatigue signature that precedes cramp. In a race, once you see the same drift, shift two cogs easier, lift cadence by 5 rpm and sit. The load per pedal stroke falls about 8 %, which normally pushes the cramp back 5-7 minutes—enough to crest the climb or take a feed.

Can post-race power files tell me if I used the wrong gear in a sprint?

Open the file, zoom to the last 30 seconds, and overlay cadence. Ideal peak power for most riders arrives between 120 and 130 rpm. If you see the biggest wattage at 108 rpm with speed still climbing, you started the gear too big and torque peaked before velocity. If the highest watts come at 140 rpm but speed plateaus, you spun out and should choose a larger cog next time. The file will also show how long you held >90 % of peak power: anything under 9 s indicates the gear let you hit the number but not sustain it; swap one tooth smaller on the cassette and test again.

How can I tell from the power file whether I attacked too early in the race?

Open the Mean Maximal Power curve for the event and slide the duration selector to the length of your break-away—say 12 min. If that 12-min value sits above 105 % of your tested 20-min power, you lit the fuse too soon. A second clue hides in the following hour: look for any 5-min block that drops more than 15 % under your first-hour average. That steep fall-off shows you spent the match and never really re-loaded. When both signals appear, next race wait until the final 6-8 % of the course, then hit it; the same power will buy twice as many seconds because the bunch is already winded.