5k Race Predictor

Mistake - I meant 5k, not 6k.

Garmin uses Firstbeat to calculate VO2 Max and predictions. 1). however, my garmin vo2 rating does match what Firstbeat  charts show. It's clear that garmin is NOT factoring age. It might be in the code and someone "commented it out" (maybe for testing) years ago and forgot to reset. 2.) Race predict is wrong due to the vo2 is off.  You'll need to search for Firstbeat and look at their charts. there are a lot online...

(I know this as I have seen the errors myself and I actually spoke to someone that is working on a case against Garmin do to cardiac issues from overtraining due to Garmins faulty data.)  So be cautious. 

Take this with a grain of salt, as I'm not a subject matter expert. A FirstBeat employee used to post here, but I'm not sure if he's around anymore to confirm or deny any of this stuff. cc: HermanB

It's always been my impression that Garmin's VO2 Max estimator (*) is based only on:

1) your max heart rate

2) how fast you can run at a given heart rate (relative to your max HR)

(* Some would argue it should more properly be called "VDOT", but that's a tangent.)

I think the way it's supposed to work is that it takes data from your submaximal efforts and extrapolates that to a maximal / race day effort, based on comparing the HR of your effort to your max HR. As a very simplified example, say I run a 10k easy run in 1 hour (10k per hour), at 70% of max HR. And suppose a maximal 10k effort is defined as 90% max HR. Then the algorithm will guess that my 10k race effort would be .9 / .7 * 10 kph = 12.86 kph, which roughly translates to a 46:40 10k.

(Here I'm assuming a very simplified linear relationship between heart rate and speed.)

Then they take that 10k estimation, look up the corresponding VDOT value using the tables / Jack Daniels formula, and spit out all the other estimated race times. At least, that's how I believe older Garmin watches worked -- this was born out by the fact that any two runners who had the same Garmin VO2 Max would always have the same race predictions.

Newer watches, like 945 LTE, also take training volume into account. So they might adjust half marathon and marathon predictions to be slower if you don't have enough mileage.

So if you plug your Garmin VO2 Max into one of VDOT tables/calculators, you may get similar race predictions, but the Garmin predictions for longer distances may be significantly slower than what's in the tables.

SUPRNOVA said:

It's clear that garmin is NOT factoring age.

My educated guess is that age doesn't matter for the estimation.

People like to say that physical characteristics like weight should be factored in, and maybe even age, but I disagree, because things like sex, weight and age already affect how fast you can run at a given heart rate (relative to your max HR), all other things being equal.

As a not-so-great analogy, it's known that height decreases after a certain age, yet I don't need to know your age in order to measure your height. (It's not a great analogy because Garmin is estimating your VO2 Max, not measuring it.)

Then again, I've never tried editing my age or weight on my watch and seeing if it resulted in drastic changes in Garmin's VO2 Max estimate. My guess is that it wouldn't.

(Of course if you were actually measuring VO2 Max in a lab, then obviously weight would be a factor, since (relative) VO2 Max is a measure of oxygen consumption per unit time, per unit body weight, and a lab test directly measures oxygen consumption. But I don't think it's necessary for the Garmin's estimation.)

My other anecdotal / educated guess for this is that the VO2 Max / VDOT tables for equivalent race predictions make no reference to age, weight, height, or any other physical factors -- for each VO2 Max value, they just list race distances and times. Same with the Jack Daniels' VDOT calculator (which those tables are based on): https://runsmartproject.com/calculator/

If I enter a 10k race result of 40 minutes, it gives me a VDOT value of 52 -- no other information needed.

Anyway, here's the Firstbeat white paper on its VO2 Max estimation method. It talks about *logging* age, but never says that age is used in the estimation process itself.

https://assets.firstbeat.com/firstbeat/uploads/2017/06/white_paper_VO2max_30.6.2017.pdf

(Page 2)

Methods used to estimate aerobic fitness
The only direct way to actually measure maximal oxygen
consumption is to use open-circuit spirometry in a laboratory. In
this procedure, pulmonary ventilation and expired fractions of
oxygen and carbon dioxide are measured during a controlled
exercise protocol. When the direct measurement of VO2max is
not feasible or desirable, a variety of submaximal or maximal
indirect tests with a controlled exercise protocol can be used to
estimate VO2max [1]. Also, non-exercise methods have been
developed to estimate a person's VO2max from individual

characteristics such as age, sex, anthropometrics, history of
physical activity, or resting-level physiological measurements
[13]. However, these non-exercise based assessment methods
are often very inaccurate

The non-exercise methods blurb is very interesting to me because it explicitly mentions that other methods use age and sex, but then when you look at the method they actually use, it basically focuses on heart rate and speed.

(Page 3)

The Firstbeat method for accurate assessment of a person’s
aerobic fitness level (VO2max) during uncontrolled exercise is
described next in detail.
Physiological basis of the method
It is well known that there is a linear relationship between
oxygen consumption and running speed. The oxygen cost of
running increases when running speed increases. At identical
submaximal speeds, an endurance athlete runs at a lower
percentage of his or her VO2max than an untrained person,
although both maintain similar VO2 [6]

...

Calculation steps
The following calculation steps are used for VO2max estimation:
1) The personal background info (at least age) is logged
2) The person starts to exercise with a device that measures
heart rate and speed
3) The collected data is segmented to different heart rate
ranges and the reliability of different data segments is
calculated
4) The most reliable data segments are used for estimating the
person's aerobic fitness level (VO2max) by utilizing either
linear or nonlinear dependency between the person's heart
rate and speed data

To me the key is where it says that heart rate and speed are used to estimate VO2 Max. That's it. Nothing about age, weight, sex, height, etc.

(Page 4)

So the short answer here is that if my race predictor results were wildly optimistic, I would check to make sure that my max HR was set properly. If your max HR is set too high, then your race predictions would be too fast.

But anecdotally, I'll say that Garmin's race predictor has always been a little optimistic for me (and many others).

Some would say that it tends to overestimate your VO2 Max by maybe 4-5 units. I'm not sure why your VO2 Max value of 36 predicts a 31:17 5k, though, when the standard tables would predict 26:21, unless there actually is a correction for age in the newer race predictor. Or maybe the correction is purely based on mileage.

Or maybe because they saw my daily runs at 14:00 are all aerobic, easy runs.

allan570 said:

Or maybe because they saw my daily runs at 14:00 are all aerobic, easy runs.

Haha that could be it.

Anyway, if you want a better race predictor, you can sync your garmin account with the free website https://www.runalyze.com.

It predicts its own VO2 Max from your run data, but it's similar to what I described above (what I believe Garmin is doing):

Runalyze estimates the Effective VO₂max for every run with heart rate data based on the relation of heart rate and pace. Optionally, elevation can be taken into account as well.

But the great thing about it is that it applies a correction factor based on your *actual* race results, unlike Garmin. All you have to do is tag your activities as races in runalyze, where applicable.

So say runalyze predicts a 20:00 5k race for me, but I actually run 22:00 (10% slower). Then its future VO2 Max predictions will be 10% lower and so will its race predictions.

The race predictions I get from runalyze are much more realistic (slower) than the Garmin predictions. The other kinda neat thing is that it (optionally) shows your Garmin VO2 max with 2 decimal places, so you can see the small changes in your VO2 Max after each run (which can also explain why Garmin says that your fitness went up or down after a run, even though your whole number VO2 Max didn't change.)

www8.garmin.com/.../GUID-31B2458A-859A-4A34-AB83-224E4A29387A.html

Your device uses the VO2 max. estimate (About VO2 Max. Estimates) and your training history to provide a target race time. The device analyzes several weeks of your training data to refine the race time estimates.

Yeah so any deviation from the "standard" race prediction based on VO2 Max / VDOT and the Garmin race prediction seems to be based only on training history and not any physiological factors like age (unless they just don't mention it.)

I don't think they ever disclosed exactly how they use your training history, though. I assume volume (miles per week) is part of it, especially for longer distances. Could also be based on training load and maybe the load focus breakdown.

first, everyone has provided some helpful insights already. i second the recommendation to consider Runalyze and their race predictions being more reasonable. in my experience, my 945LTE way over predicted what i could do in races. i'm 53, my max HR is correctly set at 192, and it was telling me i could outrace my late 20s self at all distances. (eg, 19:40 min 5k)

beyond the FirstBeat analytics, something is amiss with the 945LTE's algorithm. last month, i purchased a 955 (and sold my 945LTE at discount to my friend) and it immediately adjusted my race predictions at all distances to more reasonable levels while retaining my V02max numbers.

right now, it says my 5k estimate is 22:28 which while seemingly optimistic, seems remotely doable although even then, at my age, i don't think i'd push myself to that extent in fear of injury. (i don't do any interval work these days to minimize injury risk) [i ran a 24min 5k a couple months ago on a dirt trail and wasn't completely gassed so i would assume that it might be possible with some speedwork and pushing harder]

it will be interesting to see if your race predictions change when some of the 955 firmware changes make their way to the