Your Body Lies to You Under Stress. Here's When to Override It.

“I felt great” is sometimes the most dangerous thing an athlete can say.

A Whoop user posted something that stuck with me. Green recovery. 82%. HRV above baseline. Sleep score was solid. Every metric said go. He laced up for a threshold run. By kilometre three, his legs were dead. His pace was 20 seconds slower than it should have been at the prescribed heart rate. He cut the session at 5 kilometres instead of 10 and spent the rest of the day wondering what happened.

His Whoop was right. His body was wrong. Or more accurately, his body was lying to him, and the Whoop was faithfully reporting the lie.

The disconnect between feel and physiology

There’s a well-documented phenomenon in fatigue research called “fatigue masking.” Under certain conditions, your subjective perception of readiness doesn’t match your actual physiological state.

The most common trigger is cumulative stress. Not the acute kind where you train hard and feel tired the next day. That’s obvious and self-correcting. The dangerous kind is the slow buildup over two to three weeks of elevated training load, poor sleep, work stress, or caloric deficit. Each day, you feel approximately fine. Your body is adapting, compensating, redistributing resources. You train well. Your metrics look normal.

Then one day, without warning, you collapse. Performance craters. Energy vanishes. Motivation disappears. Everything was fine until it wasn’t.

This happens because your body’s stress compensation mechanisms are designed for short-term survival, not long-term athletic performance. Your HPA axis (hypothalamic-pituitary-adrenal axis) can maintain elevated cortisol output and sympathetic nervous system activation for weeks before the system fatigues and crashes. During that compensation phase, your wearable metrics may look normal or even good. HRV can appear stable or slightly elevated during sympathetic overdrive. Resting heart rate might stay flat because your body is masking the accumulating debt.

The metrics aren’t wrong. They’re reporting what’s happening at the autonomic level. But what’s happening at the autonomic level doesn’t tell the whole story.

When green doesn’t mean go

There are specific scenarios where recovery scores consistently mislead athletes.

Sympathetic overdrive. In early overreaching, the sympathetic nervous system ramps up. This can temporarily elevate HRV (parasympathetic withdrawal makes the heart rate more variable in some measurement contexts) and mask fatigue. The athlete feels wired rather than tired. Energy is high. Sleep might even be disrupted because of elevated sympathetic tone, but the morning HRV reading looks fine because it was captured during a brief parasympathetic window.

Emotional stress masking. Athletes under significant emotional stress (relationship problems, work crises, family issues) often report feeling “fine” physically while carrying enormous allostatic load. The body routes resources toward managing the psychological stressor, leaving fewer resources for physical recovery. But because the stressor is non-physical, wearable metrics may not capture it until the system is already depleted.

The taper paradox. Athletes reducing training load before a race often feel worse before they feel better. The transition from high training stress to rest can actually unmask fatigue that was being suppressed by exercise-induced endorphins and routine. Recovery scores improve because training load is down, but the athlete feels terrible. This isn’t a data error. It’s the body finally expressing the fatigue it was holding.

Illness incubation. In the 24 to 48 hours before symptoms of a cold or virus appear, many athletes report feeling normal or even unusually good. The immune system is mounting a response that hasn’t yet produced symptoms. Some athletes report their best training sessions happening the day before they get sick, likely due to elevated inflammatory markers temporarily boosting perceived energy.

RPE: the oldest metric, still the most honest

Rate of perceived exertion (RPE) is a 1 to 10 subjective scale that’s been used in exercise science since the 1960s (originally Borg’s 6-20 scale, now commonly simplified). It measures one thing: how hard does this feel right now?

RPE is imprecise. It’s subjective. It’s influenced by mood, sleep, caffeine, and a dozen other variables. And for all of that, it catches things that wearable metrics miss, specifically because it integrates your total state rather than measuring one physiological channel.

When a seasoned athlete says “my legs feel like concrete today even though I slept well and my recovery score is green,” that’s RPE overriding the data. And in most cases, the athlete is right to listen.

The research on this is consistent. When HRV-based recovery and RPE disagree, RPE is a better predictor of actual workout quality. A 2019 study by Saw et al. in the British Journal of Sports Medicine found that subjective wellness measures (including RPE, mood, sleep quality, and fatigue ratings) were more sensitive to impending overtraining than any objective physiological marker alone.

This doesn’t mean RPE is always right. Anxious athletes tend to overrate fatigue. Over-motivated athletes tend to underrate it. But the general principle holds: if your body is telling you something that contradicts your wearable, your body is usually the more complete sensor.

The two-signal rule

The most useful framework for managing this disconnect is simple: never make a training decision based on one signal.

If your recovery score says green and your legs feel great, train hard. Both signals agree.

If your recovery score says red and you feel terrible, rest. Both signals agree.

The interesting cases are the disagreements:

Green score, bad feel. This is the dangerous one. The data says go, but something is off. This is where athletes get hurt or dig themselves into overtraining holes. The right move is to start the session at low intensity and see what happens. If you warm up and the legs come around after 15 minutes, it was just sluggishness. If the heaviness persists or worsens, cut the session. Trust the feel over the score.

Red score, good feel. Less dangerous but worth understanding. This often happens after a hard training day when HRV is suppressed but the body feels recovered because of adequate sleep and nutrition. The red score is reflecting yesterday’s training load, not today’s readiness. An easy session or a modified version of the planned workout is usually fine.

Yellow score, uncertain feel. The majority of days for most athletes. The safe default is to do the planned workout but stay flexible. If it’s going well, complete it. If something feels wrong, scale back without guilt.

This two-signal approach catches the scenarios where either data or feel alone would steer you wrong. It’s not sophisticated. It doesn’t require a dashboard or an algorithm. It just requires the athlete to check both sources and take the disagreements seriously.

Why this is hard to automate

RPE is the one metric that can’t be passively collected. Every other input in an athlete’s data ecosystem flows in automatically: HRV from the wearable, training load from the watch, sleep stages from the ring, nutrition from the food log. RPE requires the athlete to stop and answer a question.

This is a design challenge, not a data challenge. The information is valuable. The friction of capturing it is the barrier. Most athletes won’t log RPE consistently unless the system makes it effortless and demonstrates clear value from the input.

A well-designed system would ask for a 1 to 5 subjective check-in each morning (how do you feel?) and before each workout (how do your legs feel?). It would then correlate those inputs with the objective metrics and flag disagreements. “Your recovery score is 78% (green) but your subjective rating is 2/5 for the third consecutive day. Consider reducing intensity today.”

That correlation between subjective and objective data is where the real insight lives. Not in either signal alone, but in the pattern of agreement and disagreement between them over time. An athlete whose RPE consistently diverges from their recovery score is either heading toward overtraining, dealing with non-training stress that the wearable can’t see, or has a recovery algorithm that doesn’t match their physiology.

All three of those are important to know. None of them are visible from a single data source.

The cost of always trusting the number

The athlete from the Reddit post did the right thing. He started the run, realised the data was wrong about his readiness, and cut the session. That’s good self-awareness developed over years of training.

Newer athletes often don’t have that calibration. They’ve been told to trust the data. They bought the wearable specifically to remove subjectivity from their training decisions. When the number says green, they go green. When the number says train, they train.

For 80% of days, this works fine. The recovery score is a solid general indicator and following it is better than guessing. But the 20% of days when it’s wrong are disproportionately important because they’re the days when accumulated fatigue finally exceeds the body’s compensation capacity. Training hard on those days doesn’t just waste a session. It extends the recovery timeline and can push an athlete from functional overreaching into non-functional overreaching, which takes weeks to months to recover from.

The number is a tool, not an authority. The best athletes use it as one input among several. The athletes who treat it as gospel are the ones who end up writing confused Reddit posts about how everything was green right up until they fell apart.

Where P247 fits

A single recovery score is a first-generation tool. Useful, widely adopted, and fundamentally limited. The next generation needs to integrate subjective and objective data, flag disagreements between them, and help athletes understand when their body is masking fatigue that the metrics can’t see.

Not replacing the wearable. Not dismissing the data. Adding the context that makes the data trustworthy. Including the one data point that no sensor can capture automatically: how the athlete actually feels.