Low HRV During a Training Block: A Diagnostic Guide for Ultra Runners¶

Week 9 of a 50-miler build. Training is going fine — long runs are solid, vert is accumulating, nothing is injured. But every morning the wearable comes back orange. You backed off a session two weeks ago. Nothing changed. You do not know if you need more recovery, more sleep, or if the device is just reporting noise from a hard block.

Here is the problem: one number is doing two jobs. Your readiness score captures both residual training stress and sleep quality degradation. Most mornings during a heavy build, both are present. The score cannot tell you which one is louder — and that distinction determines the right response.

This post builds the framework for splitting those two signals apart. If you are not yet familiar with the four-state HRV decision matrix — what RMSSD is, how to read a trend against your personal baseline, and when to act on suppression versus ignore it — read the foundational guide first. That post covers the basics. This one picks up where it leaves off, specifically for runners running distances where the stakes of getting the diagnosis wrong are meaningfully higher.

Ultra Training Loads Do Something Specific to Sleep¶

The reason this diagnostic problem is harder for ultra runners than for marathoners is not volume alone. It is a measurable population-level difference in how ultra-distance training affects sleep architecture.

A 2025 study of 490 athletes published in the International Journal of Environmental Research and Public Health found that ultra-endurance runners showed significantly worse sleep quality and higher night-to-night sleep variability than endurance runners — controlling for training volume and race distance. The key finding was not just that ultra runners slept worse on average; it was the variability. Their sleep quality fluctuated sharply from night to night in ways that endurance runners' did not. That variability matters because HRV is calibrated against a recent baseline. When your sleep quality swings 30% night-to-night, your "normal range" for morning HRV becomes a moving target during a build, not a stable reference point.

The physiological mechanism is straightforward. High-volume training — especially sessions generating significant muscle damage, heat stress, or back-to-back loading — elevates cortisol and core body temperature well into the overnight window. Elevated core temperature directly fragments sleep architecture: the body needs to drop core temperature to enter and maintain deep sleep stages. The result is that the same training that accumulates HRV-suppressing fatigue during the day also degrades the overnight recovery that would otherwise dissipate some of that suppression. The two sources reinforce each other. And on the wearable's screen, they look identical.

This is the structural problem ultra runners face that marathon runners face less acutely: sleep quality degradation is not a confound to HRV — it is a direct consequence of the training load, running on a parallel track with the load-fatigue itself. Treating the readiness score as a single signal when it is carrying two distinct problems produces diagnostic errors. You back off training when the problem is sleep. You protect sleep when the problem is load. Neither response is wrong in isolation; making the wrong call on which applies is the error.

The Two Signals Underneath One Number¶

Before the diagnostic table, name the two sources precisely.

Training load suppression is accumulated parasympathetic downregulation from repeated high-effort sessions. This is the signal the four-state matrix was designed to catch. It builds over days, correlates with elevated resting heart rate, and requires rest or reduced intensity to dissipate. It is directional — it gets worse as a block progresses and improves with rest days. It is what you are measuring when you see HRV trend steadily downward through weeks three through five of a build.

Sleep quality suppression is a distinct short-term HRV reduction caused by fragmented, shortened, or architecturally disrupted sleep. It can appear on any single morning — not necessarily correlated with a hard session the day before. Resting heart rate may be completely normal. It recovers quickly when sleep normalizes, often within one night. It is not directional in the same way training load is; it reflects the sleep event that just ended, not cumulative fatigue.

The practical distinction: training load suppression requires a training response (modify the session). Sleep quality suppression requires a sleep response (protect the upcoming sleep window, do not necessarily reduce the session). Conflating them produces the wrong prescription.

The Diagnostic: How to Tell Which Signal Is Driving Your Score¶

Four observable inputs, cross-referenced against the two hypotheses. Check all four before modifying a session.

How to use the table: if three or more indicators align in one column, that is the working hypothesis. Act on it. If the indicators split — say, HRV has been down three days but resting HR is normal and the subjective feel is groggy rather than heavy — you are likely in the overlap zone described in the next section.

Two common clean cases:

Case A — HRV suppressed four days running, resting HR up 5 bpm over baseline, feel physically flat but motivated, felt better halfway through yesterday's easy run, currently in week four of a build. All four indicators point left. This is training load suppression. Modify today's quality session.

Case B — HRV suppressed this morning only, resting HR normal, feel groggy and foggy with low motivation, no recent training spike but flew across time zones yesterday. All four indicators point right. This is sleep quality suppression. Proceed with the planned session; prioritize tonight's sleep window.

The framework is probabilistic, not binary. Three of four is a working diagnosis. Two of four is ambiguous — proceed to the next section.

The Feedback Loop: When Both Signals Are Running at Once¶

This is the mechanism that makes ultra training diagnostics structurally harder than the single-signal problem. Heavy training load directly degrades sleep quality through the cortisol and core temperature pathway described earlier. That degraded sleep produces additional HRV suppression on top of the residual training stress. To the wearable, the compound event looks like more training load. The runner backs off. The training adaptation slows. The sleep quality may or may not improve.

Here is what this looks like in practice at week eight of a 100K block. Training load has been accumulating for six weeks. ACWR is in a healthy range — you have been building carefully. HRV has been suppressed for five days. Resting HR is elevated. But when you look at your sleep data — not just the readiness score, but the actual sleep stage breakdown — you see fragmented nights: two to three wakeups, reduced deep sleep percentage, total sleep time down 45–60 minutes from your norm. The wearable is reading a compound problem as a unitary one.

The feedback loop is active when all three of these are present simultaneously: HRV suppressed five or more days, resting HR elevated, and the sleep data itself — duration or quality scores, not just the readiness rollup — has visibly declined from your base phase norms. That combination is not "train less." It is "train less AND fix the sleep," and the sleep fix has to happen first, because reducing load without addressing the sleep degradation will produce only partial HRV recovery.

Ultra runners doing back-to-back long runs are particularly exposed to this pattern. The second day of a back-to-back typically produces the worst overnight cortisol elevation. If the sleep that follows is fragmented, the subsequent week's HRV readings reflect both the accumulated load and the compounded sleep deficit — and neither recovers on the schedule the runner is expecting.

One practical signal that the feedback loop is running: HRV has not meaningfully bounced after a rest day. A single rest day typically produces at least a partial HRV recovery when training load is the primary driver. If you take a full rest day and the next morning's reading is still flat, sleep quality disruption is almost certainly the secondary signal — the rest day did not address the sleep problem.

Your Device's Accuracy Affects the Threshold, Not the Framework¶

The diagnostic table above applies regardless of which wearable you own. The accuracy of the overnight HRV measurement, however, affects how much weight to give individual readings versus multi-day trends.

A 2025 comparative validation study found that Oura Ring Gen 3 achieved a concordance correlation coefficient of 0.99 for overnight HRV measurement — making it the most reliable device for the single-night readings this diagnostic depends on. WHOOP 5.0 achieved 0.94, which is meaningful but introduces slightly more overnight variance. Garmin and Apple Watch show wider variance still in overnight passive readings, though both perform better for exercise heart rate.

This is not a buying guide. If you own a Garmin and the diagnostic points toward sleep quality suppression on a single morning, the right response is not to buy an Oura. The right response is to require two or three confirming indicators from the diagnostic table before acting on that one reading. The framework is the same; the evidence threshold shifts slightly by device.

Session Modification After You Have Diagnosed¶

Two modification menus — one per diagnosis. This is where ACE Fitness's May 2026 piece on HRV-gated training decisions correctly identified when to modify but deliberately stopped short of prescribing how. The distinction by cause matters because the prescriptions differ.

The critical distinction in the sleep quality suppression rows: a poor night of sleep does not automatically mean reduce training load. It means sleep is the variable to fix. Cutting training when the underlying problem is sleep delays adaptation without addressing the root cause. Unless the diagnostic clearly shows a sleep quality suppression pattern with training load indicators in the normal range, hold the session and address the sleep.

The feedback loop row is the exception. When both signals are confirmed active — multi-day HRV suppression, elevated resting HR, and visibly degraded sleep data — a full quality session is unlikely to be absorbed productively regardless. Twenty-four hours, an easy run to maintain the aerobic pattern, a deliberate sleep extension, and a reassessment the following morning is the appropriate response.

Automating the Diagnostic¶

The framework above requires cross-referencing three to four signals — HRV trend, resting HR trend, subjective feel, sleep quality data — across multiple days. Done manually with a spreadsheet and a wearable's app, it is feasible but friction-heavy. The pattern is easy to miss when you are mid-build and not looking at all four inputs simultaneously.

NavRun's weekly AI training report does this cross-referencing automatically. It pulls your Strava training load, surfaces your week-over-week pattern, and flags suppression events with context — noting whether the timing aligns with load spikes, back-to-back sessions, or travel that would implicate sleep quality. The output is not a score; it is a plain-language pattern description that tells you which diagnostic column your last week most closely resembles.

Connect your Strava and see what your last training block looks like through this lens. The number on your wrist is the starting point. The pattern across your training history is the diagnosis.