HRV dipped overnight. The week average slid too.

Last night's overnight HRV read 72.0 ms, a reading 13.2% under its own 14-day average of 82.9 ms. The sleep score fell to 49, 25.6% below its own average of 65.9. Today's plan already called for rest, and no session was logged.

A reading like that invites a reaction: skip tomorrow's plan, call it a bad week, change something. The 14-day window changes the picture. The spread, or standard deviation, across those two weeks was 12.1 ms, and the gap between last night's reading and the mean is 0.9 of that spread, a z-score of -0.9. A reading inside one standard deviation of its own mean sits within the range that window produces on most nights.

What the 7-day average showed

The last 7 nights averaged 80.7 ms, against 85.1 ms over the 7 nights before that, a 5.2% drop week to week. HRV-guided training protocols lean on that week-to-week comparison to decide whether a plan needs to change. Today, with the dip sitting inside one standard deviation and the scheduled day already a rest day, no change was called for.

What similar HRV-guided studies found

A 4-week study of 26 healthy, moderately fit men tested this kind of baseline directly. One group trained on a fixed schedule. The other had easier sessions or rest substituted whenever HRV fell below a reference range built from the preceding 10-day mean and spread, or showed a falling trend across 2 days¹. The HRV-guided group's VO2peak rose from 56 to 60 ml/kg/min. The fixed-schedule group's did not change significantly, and the difference in VO2peak change between the two groups was not itself significant¹.

A later 8-week study of 40 recreational runners found the HRV-guided group trained at moderate or high intensity less often, 13.2 sessions compared with 17.7. Their 3000 m time still improved by 2.1%, while the fixed-plan group's 1.1% improvement was not significant. The authors describe the difference between the two groups as small².

A meta-analysis pooling randomized trials in trained endurance athletes found that both HRV-guided and fixed training raised VO2max, with a significantly higher effect in the HRV-guided group, an effect size of 0.187³. Amateur-level athletes and women showed the better results in that analysis, and the authors describe the overall effect, 0.402, as small³.

Why a wider baseline helps here

A separate meta-analysis of 8 studies and 199 participants compared HRV-guided with fixed training directly. HRV-guided training was better for vagal-related HRV measures. It was not significantly better for resting heart rate, maximal aerobic capacity or endurance performance itself⁴. Read together, the four studies describe a baseline-and-trend method that can shift physiology and sometimes performance, without promising a change from any one night's number.

None of these studies tested this rider specifically, and none of them turn a z-score of -0.9 into a verdict. What they share is a method: set a baseline from a window of days, require more than one reading to call it a trend, and reserve the bigger adjustments for a baseline that has genuinely moved. Applied here, last night's number and this week's slightly lower average both sit close enough to their own history that the scheduled rest day needed no second look.

What this means before the next hard session

For a rider tracking FTP, the useful habit is to keep watching two numbers: a baseline built from roughly the past 10 to 14 days, and whether the most recent week is trending away from the week before it. A single night inside its own range, on a day the plan already marked as rest, is routine. A week sliding the same direction for several days in a row is the signal worth planning around.

The same two numbers read differently on a hard training day. A planned threshold or interval session sitting on top of a 14-day HRV reading that is still near its own mean carries less risk than the same session stacked on a week that has been sliding for several days straight. Today carried neither case: the plan called for rest before the numbers were even checked, so the overnight reading and the week's average were read after the fact, as context for a decision that had already been made.

• Read HRV against its own 14-day mean and spread before reacting to one low night.
• Compare the last 7-day average with the 7 days before it to see whether the week itself moved.
• Keep a rest day that was already on the plan when a single reading still sits inside its usual range.

Does one low HRV reading mean a session should change?

A single low reading rarely calls for a change on its own. A reading within one standard deviation of its own 14-day mean sits inside normal night-to-night swing. The HRV-guided training studies cited here changed sessions using a longer baseline or a 2-day falling trend¹.

Why use a multi-day window at all?

Single-night noise is common. The review behind daily HRV training built its decision rule from a multi-day mean and spread to filter that noise out, adjusting training only when the signal crossed a longer baseline or trended down across 2 days¹.