Ramp rate (CTL ramp)
Ramp rate is how fast your Chronic Training Load (CTL) is climbing, expressed in CTL points per week. It's the single best summary of how aggressively you're building fitness. A common rule of thumb keeps sustained ramp rate in the range of about 3–8 CTL per week; push it much higher for too long and you tip from productive overload into overreaching and injury risk.
Ramp rate = ΔCTL over a week (CTL_today − CTL_7-days-ago)
rule-of-thumb sustainable band ≈ 3–8 CTL/week
(highly individual — depends on age, history, and current CTL)Because CTL is a 42-day average of daily TSS, raising it requires each week's load to exceed that trailing average — and that surplus is the ramp. A ramp of +5/week means you're adding fitness quickly; flat or negative means maintenance or detraining. Early in a base phase or after a layoff, ramp can safely run high because you're rebuilding lost fitness; near your peak CTL, the same absolute ramp is far harder to sustain.
The 3–8/week band is a heuristic, not a law. Tolerance depends on age, training history, current fitness, life stress, and sleep — a young rider rebuilding may absorb 8+/week for a few weeks, while a masters athlete already at high CTL may find 3–4 is plenty. The right way to use ramp rate is as a guardrail with individual calibration, watched alongside TSB and subjective markers, not as a target to be maximized.
Sustained high ramp is the classic setup for non-functional overreaching: TSB stays deeply negative week after week, form fails to rebound on easy days, and performance stalls or drops. The remedy is periodization — recovery weeks that deliberately let ramp go negative so CTL consolidates — rather than an ever-climbing line. A ramp that only ever goes up is a plan with no off-ramp.
How AdaptCycling uses Ramp rate
AdaptCycling's solver enforces a weekly CTL ramp-rate bound, so a plan can't quietly stack 40% more load into a single week, and it schedules recovery weeks off the CTL trajectory rather than by calendar alone. That bound is one of the hard invariants the deterministic solver owns, keeping the LLM's plan shaping inside a safe build rate.