Lower Threshold Power, what is this?

modeled LTP is always 20+% higher than my LT1 estimate.

how are you estimating LT1? % of FTP? How are you determining that percentage?

Breath test and talk test to give HR (about 120-125bpm). The Xert LTP is almost always in low tempo as % ftp, and riding at Xert LTP usually results in a HR of 135-140bpm. While I can ride all day (and climb for 8 hours) at HR of 135+ it’s definitely not LT1 as I need to consume a lot of carbs to keep it going.


At LT1 you should be producing 1-2 mmol/l lactate so a significant amount of glycogen is still being utilized, I think ~30% from cho if IIRC. I think depending on if youre using power at LTP/LT1 vs HR for long durations will affect your rpe there since cardiac drift will be an issue. I havent really tested this however and it’s mostly from looking back at older data with stochastic efforts. Never tried to pace directly to lt1 for long durations so I could see LTP being slightly above LT1

IMHO its like ftp, often overestimated because nobody really enjoys riding slow (and then posting to Strava).


Dr Seiler said in a recent interview that having minimal hr drift was due to staying below lt1. I had already come to tr with a training history of using hr as my main metric, using my promotional code from my stages for joining. Conversational pace was the old rule for most rides and runs so pace was usually set my rpe and measured with hr. I haven’t taken the ftp challenge yet, but i can routinely ride in traditional zone 1 hr up to about 75% of my manually estimated ftp. Doing 1 hour at that ftp setting seems quite hard but it might be possible so my lt1 might be higher percent of my actual 1 hour power.

And when my aerobic fitness is high, I can have minimal HR drift well above LT1. As in doing Galena 3x20-minute intervals and seeing less than 3.5% drift (aerobic decoupling) for the entire 90 minute workout. If I select 70 minutes for the three 20-min intervals and two 5-min recovery valleys, my aerobic decoupling is 2.3%. If I select 45 minutes for 2nd and 3rd intervals (plus recovery valley), my aerobic decoupling is -0.5%. There is absolutely no way my LT1 is at 90% ftp or 94% ftp or 92% ftp.

That Galena was done early week 5 of SSB-1 high volume plan. My ftp from week 1, using 8-min test was 236W. On the day of Galena, Xert modeling had ftp of 231W and LTP at 192W. That LTP is in the middle of my Tempo zone.

I’d argue that your 8 hour pace is closer to LT1 than Tempo

Agree, but that wasn’t my point.

What I’m saying is that Xert’s LTP estimates are all over the place, from upper aerobic endurance to the middle of tempo. I’ve even seen Xert freak out after my best ramp test ever (Sept 2018), which in TR gave me 250W ftp but Xert modeled ftp at 208W (from 198W) and LTP of 202W (jumping from 140W). Seen some wacky Xert stuff from time to time, and find WKO4 does a much better job at modeling (everything).

1 Like

Yep, all of this. Pretty much my experience too.

All this talk about LTP and MAF training has got me interested in actually testing LT1 and LT2


If you’re seeing anything “freak out”, check your data. If you have consistent data and performance, you should see your signature and LTP vary only slightly from ride to ride / breakthrough to breakthrough. It’s when data is either contains errors or you haven’t expressed any fitness parameters for an extended period that errors can accumulate. Once you know what’s happening though, these are easily corrected. Contact support and someone will help you if needed.


To be fair and a bit less curmudgeonly, I have to do similar things in WKO4 when data becomes too old or is bad in the first place

I don’t. Same data with power starting Oct 2017. Haven’t seen any blatant modeling issues in WKO4.

My mFTP just went down about 30W day before yesterday utterly out of the blue. I’m like: “so can I actually use this metric or it is yet another petty data thing to trigger my nerd rage?!”

Running list of input from ppl I’ve asked/reported:
“Bro, feed the model”
“Relax, mFTP is just a guide.”
“Hey dude just spend some of that time-crunch you have to track down bad data”
“You’re taking the model to literally”
“What’s WKO4”
“Your data is old”

Good times!


All models have assumptions, and conditions when the model is accurate or not. If you don’t take the time to explore that, my suggestion is to use mFTP to triangulate your ftp along with other data. Or perhaps as a prompt to do whatever test (20-min, 8-min, ramp, or TTE) that does the best job of estimating your ftp. Or decide you don’t want to feed the model and move on.

I like nerding out on the Xert MPA graphs, great stuff but I wasn’t able to put it to practical use (my judgement call on that, as I could find a way if motivated). After paying for a year of Xert, I joined TR and did a block of SSB high volume, and watched my Xert modeled FTP drop and drop and drop. Support pointed me to a multi step blog post to fix the modeling issue, and that post did properly explain the conditions under which Xert modeling does a good job. But it was just too fiddly for me. And I don’t get implying LTP is equivalent LT1, and yes I can parse the blog post and understand that mirroring is not the same as equivalent. Just keeping it real, not trying to be curmudgeonly.

1 Like

There was a post on their FB page recently that said their algorithm sometimes throws a wobbler when there is a big jump in TP. 250 from 198. On the activity you have to manually change the TP to something closer to what you believe it is before doing a recalc on it. TP in their model is also dependent on an accurate PP number

My LTP1 sits at 78% of my FTP. Which pretty much ties in exactly with Seillers numbers.

From the horse’s mouth there is no intended physiological alignment with LT1/VT1/AeT…any such agreement is purely a happy coincidence