Recovery Logic and ATL/CTL calculation for high-volume athletes
Hi LeCoach Team,
I would like to provide some feedback on the training logic and readiness assessment, as I’ve encountered a persistent discrepancy between my actual training load and the coach’s recommendations.
The Issue: The system continues to prescribe recovery days and warns of high fatigue (ATL), even when I have performed very little meaningful training for over a week. It seems the model is overly sensitive to the number of consecutive training days rather than the actual physiological load of those sessions for an experienced athlete.
Example from my data (Aug 5th – Aug 16th):
Historical context: I came off a very high-volume block in the Alps (CTL 100+).
Recent period: For the last 12 days, I have recorded an activity every day, but most were very short recovery spins or commutes (load < 30).
Actual intensity: In these 12 days, I only had 2 proper workouts (VO2max and Threshold). The rest were effectively "nothing" for my fitness level.
The Conflict: Because I didn't take a full day off (0 load), my ATL remained mathematically high (~127). Consequently, the AI coach kept proposing recovery days, even though my fitness was actually stagnating/decreasing due to a lack of stimulus.
Suggestion for Improvement: For athletes with a high CTL (e.g., >90), could the logic be adjusted so that:
Low-load activities (e.g., <30 load or commutes) are weighted differently or ignored when calculating the "consecutive days" penalty.
Readiness assessment looks more at the nature of the load (Z1/Z2 vs Z4+) rather than just the 7-day ATL average.
The coach recognizes when an athlete is "under-trained" despite a high ATL, especially when the TSB is negative purely due to a high frequency of very low-intensity activities.
I feel like I'm losing my "Alpine form" because the system is too protective based on a mathematical model that doesn't fit the reality of my training intensity.
Best regards,
Gianni
Comments1
An Anonymous User
Aug 18
@Gianni Riemslagh i thanks for raising this.
Low-load activities and commutes should, as of this week, be excluded from the consecutive-days penalty. If you still encounter that, could you email us at info@lecoach.app so we can look at the specific case?
On ATL/CTL, though, it’s more nuanced.
ATL is not determined by how many consecutive days you trained. In Intervals.icu it is a 7-day exponentially weighted average of your training load. CTL works the same way, but with a 42-day time constant.
There is no special mathematical distinction between a "rest day" and an "activity day". The calculation only sees the load for that day. A day with load 0 reduces ATL somewhat faster than a day with load 20 or 30, but simply recording an activity every day cannot by itself keep ATL high.
For example, suppose you start with ATL 127 and CTL 100.
If you then do 7 consecutive days at load 30/day:
ATL falls from 127 → ~66
CTL falls from 100 → ~89
Form moves to roughly +24
Even at load 50 every single day, ATL would fall to ~78 after one week.
After 12 straight days at load 30/day, ATL would be ~47 and CTL ~83.
So coming off a large Alpine block, doing substantially less work should cause ATL to fall quite quickly even if you continue doing a short recovery ride or commute every day. CTL falls more slowly, which is exactly what the different 7-day and 42-day time constants are designed to represent.
The two proper workouts you mentioned will of course increase ATL according to their actual training load, but the low-load days between them do not keep ATL artificially high simply because they are consecutive.
I do agree with the broader point that ATL/CTL are mathematical load models rather than a complete measure of physiological readiness. Workout type, intensity and recovery signals can add useful context. Training load itself already accounts substantially for intensity, but two sessions with the same load are not necessarily physiologically identical.
However, an ATL of ~127 after the 12-day period you describe cannot be explained by the absence of a zero-load day. If it genuinely remained around 127 while most of those days had load <30, I'd actually like to look at that specific data, because either the higher-load workouts materially explain it or there is something else worth investigating. Feel free to email us on info@lecoach.app.