Measurement log
Four of Garmin's five running dynamics are just your pace again
26 runs from my own export. Ground contact time is 97% pace. Vertical ratio is 93% pace. Only one of the five tells you something pace does not — and it barely moves.
Garmin gives a Forerunner 265 owner five running form numbers without asking for a single accessory: cadence, stride length, ground contact time, vertical oscillation and vertical ratio. They arrive in a section called Running Dynamics, which is the sort of name that suggests they are five separate things about how you run.
I plotted all five against pace across 26 runs of my own. Four of them are pace with a different unit on the axis.
The two ends of the result
Source: my own Garmin Connect account export, summarizedActivities, all 26 running activities between 17 June and 21 August 2026. Lines are least-squares fits on pace.
Left panel: ground contact time. Every run sits on the line. Knowing only how fast I was going explains 96.6% of how long my foot stayed on the ground.
Right panel: vertical oscillation, the same 26 runs, same x-axis. A cloud, and a narrow one — the whole two-month range is 8.63 cm to 9.95 cm. Just over a centimetre separates my easiest jog from my hardest interval session.
All five, ranked by how little they add
| Metric | R² against pacemeasured | Range across 26 runs |
|---|---|---|
| Ground contact time | 0.966 | 253–296 ms |
| Vertical ratio | 0.925 | 7.60–9.06 % |
| Stride length | 0.852 | 100–133 cm |
| Cadence | 0.714 | 153–166 spm |
| Vertical oscillation | 0.399 | 8.63–9.95 cm |
Two of these are arithmetic rather than physiology and should be read that way. Stride length is speed divided by cadence, so it cannot be independent of pace. Vertical ratio is vertical oscillation divided by stride length, so it inherits stride length’s dependence — which is exactly why the ratio scores 0.93 while the oscillation it is built from scores 0.40.
That leaves two measurements that are genuinely about your gait: ground contact time, which turns out to be almost perfectly determined by how fast you are going, and vertical oscillation, which is not.
So is vertical oscillation the useful one?
It is the only one carrying information pace does not. Whether that information is worth anything is a different question, and the range is the reason to be careful.
My 26 runs span 4:29/km to 6:25/km — a genuinely wide spread of efforts, from recovery jogs to threshold work. Across all of that, vertical oscillation moved 1.32 cm. Garmin publishes colour zones for this metric that are several centimetres wide. In other words, my entire two-month variation would sit comfortably inside one band.
That is what you would expect from a number that mostly describes you rather than today. Your bounce is a property of your body and your gait, and it does not swing about because you ran a tempo. Watching it day to day is watching noise around a personal constant. Watching it across a year of deliberate form work might be different — I do not have a year of data, and I am not going to pretend otherwise.
What this means before you buy an accessory
Garmin sells the HRM-Pro chest strap and the Running Dynamics Pod partly on these metrics. I own neither, so I cannot tell you whether their versions of these numbers are more accurate than the wrist’s.
What I can tell you is the part that does not depend on accuracy: four of the five metrics track pace so closely that a more accurate version of them would still be telling you your pace. An accessory that measures ground contact time better does not hand you a new training signal; it hands you a better-measured version of something your watch already knew from your GPS.
The honest reason to buy a chest strap is heart rate during intervals, where optical wrist sensors are weakest. Running dynamics is the wrong reason.
The same pattern, twice
This is the second time I have plotted a Garmin metric against pace and watched it collapse into a line. The first was running power, where pace alone explained 94.5% of the watts.
The pattern is not that Garmin is making numbers up. It is that a watch strapped to one wrist has a small number of real inputs — GPS speed, an accelerometer, an optical heart rate sensor — and presents them back to you as a larger number of named metrics. When two of those metrics are computed from the same input, they will agree, and their agreement is not confirmation.
It is worth knowing which of your screens are independent measurements and which are the same measurement wearing a different label. On the 265, four of the five are mostly the same measurement.
Method, so you can check it
Every number above came out of summarizedActivities.json in a Garmin account export — the file
that holds all 97 fields per activity, including these five.
Here is what else is in that archive, and how to
get one.
Pace was computed as duration ÷ distance from the same records rather than taken from Garmin’s
avgSpeed field, and R² is an ordinary least-squares fit of each metric on pace. Filtering: all 26
activities of type running in the period, no exclusions.
The caveat that matters: this is one runner, one watch, 26 runs, on flat ground, across a pace range of about two minutes per kilometre. A hillier data set would loosen the ground contact time fit, because pace stops being a good proxy for effort as soon as the ground tilts. If you have your own export, the code is four lines and I would rather you checked than believed me.
Sources
- My own Garmin Connect account export,
summarizedActivities, 26 running activities between 17 June and 21 August 2026 - Device records in the same export’s FIT files, confirming no external sensor was paired: every entry is the watch, its GNSS chip, or an internal component