Lactate Threshold Pace and Lactate Threshold HR not aligned

This users seems to have a similar issue: https://forums.garmin.com/outdoor-recreation/outdoor-recreation/f/fenix-5-series/128821/lactate-threshold-pace-madness

Not sure neither if this is good or bad...

Same problem here. 
HR LT has been pretty accurate (171-173 BPM for me) but pace is ALWAYS underestimated (4:45 - 4:55).
When my HR is close to 172 pace is usually closer to 4:25 - 4:30.


I don't know why the algorithm is not able to correlate such "basic" values. Maybe there is a factor applied.
It wouldn't be a problem if LT pace ("AnT speed") was not a parameter to calculate training load and training effect (as described in this Firstbeat patent document : US020200261011A120200820 (storage.googleapis.com).

Feels like the more elaborate Firstbeat algorithm get, the more they report wrong data.

I gave up on it :

It doesn't make scientific sens for now based on my data.

drkglb said:

It wouldn't be a problem if LT pace ("AnT speed") was not a parameter to calculate training load and training effect (as described in this Firstbeat patent document : US020200261011A120200820 (storage.googleapis.com).

Thank you for sharing. This is a nice find: finally seeing the details of the rules to calculate the types and levels of TE.

drkglb said:

HR LT has been pretty accurate (171-173 BPM for me) but pace is ALWAYS underestimated (4:45 - 4:55).
When my HR is close to 172 pace is usually closer to 4:25 - 4:30.

You cannot pick a HR/pace data pair, because duration is of paramount importance. This is why the watch models are complex. 

Technically speaking the Threshold pace is your best average pace you can sustain for 1h; it is not the pace your might see when your HR is at LTHR, because duration matter: during a shortish interval, you might get your HR steady around LTHR and manage the lactate for a while, but not necessarily for 1h. Vice versa, you might run a half marathon and have your HR dabble in treshold values or higher, but your average pace will be lower than your LT pace.

The best way to gauge whether they are accurate for you is to try to run 1h at maximum steady pace. If the LT pace of the watch is close to your average pace for this PB, it is accurate by definition.

In my case, I can run the 10k 10s faster than my LT pace, which gives me some confidence I could go 15mn longer if I was at LT pace instead of 10k pace.

Hello ! Yes, I sure tought you would be interested in that document ;)
By the way I also found this one which gives more details about the VO2max algorithm (that you are already quite familiar with) : 1498397218284622862-US20110040193A1 (storage.googleapis.com). You'll notice that latest updates include LTHR detection using respiratory rate determined by HRV measurement...
I'm just wondering if all this data processing is making the VO2max estimation better... Or worse :)

You're right, I know duration matters. But 173 BPM heart rate is very close to my10K HR level, and my PB on 10K is 44:30 (4:28).
I ran a 21K one month ago in 1:39:45 (4:43), which is still faster that what Garmin estimates for my LT pace. Hence the confusion for me :)
Maybe it's just the error margin.

drkglb said:

But 173 BPM heart rate is very close to my10K HR level, and my PB on 10K is 44:30 (4:28).
I ran a 21K one month ago in 1:39:45 (4:43)

Yep, it is pretty clear that your LT pace is slightly underestimated.

drkglb said:

Maybe it's just the error margin.

Maybe, yes, with a stretch. The ~20s/mile error seems to be about 7% in your case. Stryd predictions error range for a 10k are about 1mn for the race in my case, which is a ~2% error range

Maybe the problem is that the LT detection model stands on its own and does not take the training history into account in general.

Along these lines, did you 10k and 21K trigger a LT auto-detection update?

Etupes25 said:

Along these lines, did you 10k and 21K trigger a LT auto-detection update?

No, I don't think so. But my max HR was auto-detected yesterday (187) and it's a bit higher that what I entered (183). Maybe it will adjust in the next threshold runs. We'll see :)