Predicted 5k time

on the 735 estimates for 5k were just a lookup table based on VO2Max, very static. As long as your VO2Max had the same numeric value, the estimate was the same. For most people, the result was quite optimistic, because the VO2Max estimates are quite optimistic.

On the Epix the algorithm is more sophisticated and dynamic.

What you describe (you run faster then your estimate suggests you are capable of) suggests one or more of your key physiological metrics (HR Max or LTHR most likely) is incorrectly auto-updated, or you manually set some of them, or they got reset following a FW update.

Best way to reassess your LTHR is to do the LTHR test with a good chest strap, like PolarH10. For me, unless I run with a chest strap at LTHR values, the metric doesn't update.

Garmin metrics and auto-update system are pretty cool, but quite limited and very dependent on HR data. If you want this to stay fine-tuned, learn how it works, what affects the values, how to monitor these values, and if you haven't already, invest in good chest strap to get reliable HR data.

And maybe invest in Stryd, while you're at it ;) to get reliable pace data.

3296047 - thanks for your reply and suggestions.

I do wear a chest strap every time I run, it's a Garmin HRM Dual. Also, my LTHR has updated a couple of times recently.

Regarding the Stryd suggestion, I don't understand how that would help for an accurate *average* pace figure. The GPS accuracy of the Epix is excellent (noticeably better than my previous 945), so the distance recorded is consistent and correct. The overall time aspect is just a stopwatch, therefore the average pace per mile during my runs as shown on the watch and in the Connect app are easily calculated and accurate.

What makes absolutely no sense to me, is if I run at let's say 'x' minutes per mile during my regular runs, why would the Epix predict my 5k race pace to equate to 'x' + 20 to 30 seconds per mile?

Surely, using my usual running pace would be a starting point for calculations for a predicted 5k race time, then reduced (not increased!) taking into account other factors such as heart rate? My feeling is that Garmin have got something wrong here.

3296047 is right. If your pace estimates are off, it is because your VO2 estimate is off, assuming your age and gender data is correct in the app. The main but not only reason the VO2 estimate is off is the HR max is off. A few % change in the HR max can make a big difference. Based on my experience, if you run regularly across the training effects and your pace estimate for the 5k is very close, you can deduct that your Max HR is quite accurate. 

To get there, I started with an average Max HR based on your favorite formula, tested for LTHR, ran for a few weeks on a training plan to get a balance training status, then adjusted slightly my Max HR up and down until my daily suggested workout was showing challenging but achievable paces.

As of today, my 5k pace are still a bit too high, but very close to Stryd's race calculations. I conclude that my LT HR, Max HR, VO2, Critical Power are pretty much in sync and I can feel that in my perceived effort.

In particular, the easy pace is key and the easiest to double check. If the watch is asking to run an easy pace that is more like a tempo pace for you based on perceived effort, it means your VO2 is too high and most likely your Max HR is too low. Max HR too low means your run often closer to your limit for durations longer than average, which means you are in awesome shape, which means your VO2 is great and you can run an awesome 5k.

Vice versa, if your training is balanced, easy pace is very easy and you have no trouble running threshold paces, then the watch underestimates your VO2, your HR max is probably a bit too high and your 5k/suggested workout paces are too low for what you can achieve.

Etupes25 - thanks for the reply and further explanation of how the predicted 5k is calculated. It certainly makes sense for calculating a predicted 5k race time which is *quicker* than general training run times, but surely the watch should never predict a 5k time that is one to one and a half minutes longer than I generally take for 5k when I'm not going flat out? 

Thing is your 5k predictions are not based on your past runs which happens to be 5k, but on your "hardest" 20-30mn runs that you run at about 115% of your LTHR, ie based on your VO2 max level, gender, age. Sure, there is no logic in the watch that says "before I propose a 5k time, let's make sure it is faster that what his guy did on average for 5k", but it is what it is.

"Regarding the Stryd suggestion, I don't understand how that would help for an accurate *average* pace figure. The GPS accuracy of the Epix is excellent (noticeably better than my previous 945), so the distance recorded is consistent and correct."

I don't see major improvement vs 945 on pace accuracy, and it still jumps 10s up/down when I try to maintain consistent pace in ie. 5k effort, which I find frustrating. Or when using pace alerts and it shows you're running too slow, when you're not. So you either live with the watch constantly buzzing at you during workouts, or you have to set broader brackets, sacrificing the consistency of the effort...

It also matters if you do short anaerobic efforts, which may affect your VO2Max estimates, as it may affect your pace profile (not sure if garmin use that high end of pace though in their VO2Max calculations).

Can I ask, these moderate 5k efforts you do which are faster than the 5k max estimate, do they show up as anaerobic effort? very high effort numbers?

Thanks for sharing you're using a chest strap, you still may have an errand HR Max recorded (do you get close to that HR Max value in your max efforts, i.e. uphill sprints in the heat?). The LTHR values you get from the strap - do they make sense, when compared to your i.e. 1h long threshold efforts?

Absurd - thanks for your reply.

I came across this thread because I ran a 10K in 44:30 yesterday, while my predicted time was set at 45:10.

So I was curious about which setting to "adjust" to bring the predicted time a little bit closer tomy actual record.

Interesting to know how you can use max HR as a toggle (thank you !) because I set it voluntarily a few BPMs below the real value (I almost never reach my HR, even during intense anaerobic workouts - and I was concerned I would never get a correct anaerobic training load) and maybe this is why my 10K predicted time was a bit overestimated.

I have a question though. Do you know if the LTHR also has an impact on VO2max calculation, and race time prediction ?
Could it also be used as a toggle to adjust your performance estimation closer to reality, as is HRmax ?

drkglb said:

Do you know if the LTHR also has an impact on VO2max calculation, and race time prediction

My understanding is that the VO2Max model is the sole cornerstone of all the race predictions. Within this model, I would expect the watch maintains a pace/duration curve, or equivalent based on big data pattern matching (Garmin mentions AI-based modeling and the watch has a chip dedicated to machine-learning). We know that for the VO2Max model, Garmin looks at linear and non-linear ventilation/work relationship. Ventilation is approximated by HR in sub-threshold intensities, and HRV frequency-domain metrics like HF in threshold and super-threshold intensities.

Microsoft Word - EE-white-paper_1 7 6_modified_corrected 2.doc (firstbeat.com)

The lactate threshold is approximated by the second ventilation threshold (detected by processed HRV-HF) and Max Ventilation is somewhere after that from an HR or pace perspective, when the maximum ~4-5mn steady pace is reached. Using pattern matching, it should be possible for Garmin to fill out the data when the model is partially completed (for example when only sub-threshold efforts are recorded, like walking).

To get a good VO2Max estimate, once HRMax has been fixed. it is important to have a variety of efforts, including 1s-10s very short burst/sprints at maximal pace/power, 1mn anaerobic, 4-5mn VO2Max, 15mn super-threshold, and 30mn threshold best efforts. This will determine how steep the (virtual) pace/duration is, it will focus the VO2Max pace, and the threshold pace (time is horizontal and pace is vertical), therefore the race predictions.

Keep in mind that Firstbeat puts the VO2Max-based predictions in context:

"One of the most interesting application areas for VO2max information is the prediction of race time. In sedentary runners, improvements in VO2max most probably result in improvement in race time, and therefore, the prediction of race time based on VO2max is quite straightforward. In elite endurance athletes, VO2max is not the only determinant of a good race performance, since they all have high VO2max and the margins between the athletes are small. Thus, other physiological, biomechanical, and psychological factors affect the competition results significantly. For example, if a sedentary runner improves his/her VO2max by 2ml/kg/min, the marathon race time could improve an astonishing 15min, while the same absolute improvement in VO2max in an elite athlete could improve the marathon race time by only one and a half minutes! "

I think that the model implemented on the watch is smarter than a look-up in the above table.

On the watch, a way to see whether the VO2Max and threshold estimations are in harmony is by looking at the target for threshold workouts. For example, in my case, my biking FTP is around 250W these days, and the watch is suggesting threshold workouts at 245W: this is a good indication that the metrics are concurring.

very detailed answer ! thanks