A small package of form cues appeared to improve my running overnight. A fifty-three-run archive check would not be kind to my headline.
I uploaded another Garmin activity to an AI model expecting the usual advice: slow down, keep the heart rate under control, recover better.
The analysis pointed somewhere else: at my running form. I was spending too much of every stride moving upward.
The advice was insultingly simple. Shorter steps, quicker steps, land under the body, stop trying to fly between them.
The next day Garmin reported a vertical ratio of 10.1%, down from 11.3%. Minus 1.2 percentage points in twenty-four hours.
My comeback runs had just become an experiment. Had the model found a real form problem, or something else?
These days the experiment has a deadline: the Sofia Marathon, this October, 42.2 kilometres. On 10 June it had none. The marathon entry was still eighteen days away.
From a 207 bpm 10K Finish to a Marathon Entry
Last October I finished the Sofia Marathon 10K in 1:00:20, in bad shape and with the numbers to prove it. Average heart rate 181. Maximum 207.
The raw file shows the finish was not an isolated one-second spike: most of the final minute sat above 195 bpm, twelve seconds of it above 205.
Then running left the centre of my life, first to a snowboarding winter and a finger surgery, then to the renovation that took my lower back when I decided that carrying 120脳60 cm tiles alone was reasonable.
By the time I was running again in 2026 there was no marathon plan. My first AI message about running was meant for my brother. On the evening of 5 May I typed into what I thought was our chat that he could ask ChatGPT how to prepare to run at 5:00 per kilometre, and sent it to ChatGPT. My opening move in this project was informing ChatGPT of the existence of ChatGPT. It took the news well: yes, you can ask like this, then a prompt template and a coach's checklist. That is when it clicked: I could ask for myself. A few messages later I pasted my Garmin numbers into GPT-5.5 and asked for a plan. What came back was an eight-week plan for a 25-minute 5K, and that was the whole ambition.
Translated from Bulgarian. The message meant for my brother: "you can ask chatgpt how to prepare to run it at 5 min per kilometre." ChatGPT: "Yes 馃槄 You can ask like this," then a template and the checklist.
On 28 June, after an 11K trail race, I entered the full Sofia Marathon.
It is the same event where the 207 happened, one year on and four times the distance. I crossed its finish line redlined. This October I want to cross it with a heart that stayed where I put it.
The Bug: Ignoring Garmin's Running Dynamics
I had always understood effort the obvious way. A hard run feels productive. A high heart rate proves commitment. A fast kilometre means the training worked. That logic had produced some respectable single runs and very little consistency.
Garmin was already recording more than pace and heart rate: cadence, stride length, vertical oscillation, ground contact time, power, vertical ratio. I treated those columns as decoration.
Where the numbers come from
The rig is a Garmin MARQ Gen 2 on the wrist and an HRM-Pro Plus chest strap under the shirt. With the strap on, heart rate and the running dynamics come from the chest. Without it the watch estimates them from the wrist. The strap was on for both runs in this story, and for forty-five of the fifty-three runs in the archive check. The 207 bpm earlier is a chest-strap number too.
What is vertical ratio?
Vertical ratio is vertical oscillation divided by stride length: how much the torso bounces against how far each step travels, shown as a percentage. Lower vertical oscillation has been associated with better running economy in group studies, but the ratio itself is not a measure of economy. It is a ratio of two distances, nothing more.
The comparison that matters is me against my own numbers once pace is accounted for.
On 10 June I ran 5.42 km at 7:11/km, average heart rate 152, vertical ratio 11.3%. Nothing catastrophic. The run just looked expensive for the speed it bought.
My old fix would have been to push harder. The analysis said to change the movement instead.
The Patch: Stop Bouncing
What the model saw: one 5.42 km Garmin activity with pace, heart rate and the running dynamics attached. No earlier runs, no baseline, no video.
The cues were not a running-form rebuild. That would have been a terrible idea a few weeks into a comeback. The useful version was five small, general cues, the kind any coach could have handed me: shorter and quicker steps, less reaching forward, nothing exotic. No coach was in the room, and no coach had just read my Garmin file. I did not save the exchange, so even that summary is from memory. Five cues at once is a package, with no single clean variable in it. Then came the next run, and the job of making the change visible in the data.
The Diff: Garmin Before and After
Session averages from Garmin's activity export. Both runs felt easy, about 5 km. Garmin Connect's own side-by-side agrees, and grade-adjusted pace sits within seconds of raw pace on both days, so elevation was probably not the main explanation.
The headline number was the vertical ratio. The more interesting part is the decomposition: a little less bounce, more distance per step. The stride did most of the work.
Shorter steps producing a longer stride looks like a contradiction. My reading at the time was a story about reaching less in front of the body. Nothing in the file measures that. The reading the file does support is less flattering: stride length is the denominator of the ratio, it grows with speed, and the second run was faster.
Vertical ratio = vertical oscillation 梅 stride length, so both changes push it the same way. Garmin's session averages are rounded, so the components do not recompute the ratio exactly.
Cadence rose two steps per minute. Together with the longer stride, that accounts arithmetically for the faster pace. Ground contact time was effectively unchanged.
Did It Work? What One Pair of Runs Can Prove
These were not controlled runs. The second was 39 seconds per kilometre faster and took place under different conditions, so one pair cannot isolate the effect of the cues.
The sentence I am not allowed to write: AI made me 10% more efficient overnight.
The sentence the data supports: after I changed my form based on the analysis, Garmin recorded a 10.6% lower vertical ratio on the next run, together with a lower heart rate at a faster pace.
That day, everything moved in the intended direction, after a set of cues simple enough to repeat. Whether the change persisted is the question the archive check answers.
The Archive Check: How the Before and After Nearly Fooled Me
Two months of runs later, while editing this piece, I ran the check a good reviewer would demand. Every run in my archive, vertical ratio against pace, before the cue and after.
Every comparable run in the archive. Hollow points are wrist estimates, no chest strap. The two June runs are marked.
The archive was not kind. The 11.3% of 10 June is the highest ratio in my archive up to that day. The before was an outlier, not my baseline. The 10.1% of 11 June is normal for me at that pace. I had touched 9.4% in May, strap on, before the cue existed. And at easy paces since, on long runs and tired legs, the ratio has drifted back above 11.
Fit a straight line through ratio against pace for the forty-five strap runs, twenty-nine before the cue and sixteen after, and the after-cue runs sit slightly higher, about 0.8 percentage points on average. The two periods also differ in distance, fatigue and season, so I read that as a failure to find a breakthrough, not as proof the cue made me worse.
So the honest verdict on the headline: one bad day, then a faster evening run, dressed up as progress. The pattern is consistent with regression to the mean. The cue about not reaching forward may still be good advice. The numbers do not prove it worked, and catching that is the entire reason this series logs everything.
The model deserves its share of the verdict. The run it read as a form problem on 10 June was, the archive now says, my highest-ratio run up to that date. I gave it one run and no baseline, and it read the outlier as my form. The before and after fooled me. The before alone had already fooled the coach. The fix for both of us is the same: the next diagnosis starts from the whole archive.
The model told me to stop bouncing. The archive says the bounce was mostly one bad morning.
The loop held. The conclusion did not.
What survives is the loop. It stays manual: the watch and the strap record, and the file goes to the model with the human context attached, the part Garmin cannot know, the furniture I carried the day before, the bad sleep, the sore skin between two toes. A model's output is a claim, not a fact, and after this check a patch is kept only when the whole history at similar paces agrees. Hand a model one isolated day of your data and it may turn an outlier into a confident story. There is a race on the calendar now. Every patch from here answers to October, and to the runner who actually shows up on Tuesday.
The model can suggest. The legs have veto power.
Appendix
Archive-check method: I included only runs recorded with the chest strap, 29 before the cue and 16 after. I fitted an ordinary least-squares linear regression of vertical ratio against pace. This was a descriptive sensitivity check, not a causal analysis. It did not adjust for distance, terrain, temperature, fatigue, or workout type.
Further reading: Garmin's definition of vertical ratio and a 2024 systematic review and meta-analysis of the relationship between running biomechanics and running economy.