Intervals.icu eFTP

¾ of the way through my ramp test my bike decided to part ways with the turbo (will be emailing their support too) so I had a v brief break getting it back on.

This allowed a slight hr drop, pushed on and died with a 15w increase (223 to 237), I expected an bit but am unsure if the lull will have messed up my pacing.

This is actually inline with intervals estimated FTP of 240W, does anyone know how accurate this is? Obviously I will be able to confirm this pretty soon too!

1 Like

My personal experience only, but I’ve found the Intervals.icu FTP to follow along pretty accurately for me. Try 240 for a couple rides and see how it goes. If its doable but a bit to hard drop to the 237 and try that.

Or just wait a day and then ramp test again.

If you are asking what FTP should you train with I would have to say no more than 237. You know that you had some amount of recovery after getting well into the test and so there is no way it could be higher than your test result. I am thinking 235. If it were minute 8 or even 10 of the test then perhaps 237 is accurate. A couple watts is splitting hairs tho.

I probably would not retest personally. If the workouts to follow were too hard, adjust down.

I really wouldn’t use intervals.icu or any other programs for FTP estimates. TR has a testing protocol and their workouts should have proper intensity from that standard test. Adding estimates from other programs just would seem to muddy the waters if you ask me.

1 Like

Less what to train with, more is it accurate enough to validate the interpretation of the ramp test. Which it sounds like it is. Also could be good to let you know when it’s time to test again if you’re on a homebrew plan.

I’ll confirm back after a couple sessions.

Sorry but I am lost on this part.

Fair enough. More data is not a bad thing and confirmation of data is certainly a good thing. I would still argue using a TR FTP test and the standard workouts that have intensities based on that test within the same program should be what is followed. If you test at 237 and another program estimates 240 I would train at 237 every time.

It seems that follow my FTP pretty good
TR FTP -> 345w
Interval.icu eFTP -> 340w
Xert FTP -> 340w

1 Like

When was the estimated FTP set? What effort was it based on? How long ago was that, has it declined at all since, and what training have you been doing since then?

I wasn’t asking what ftp to use per se, more how accurate people had found the estimated ftp or was mine being with 3W fluke, sorry if not clear. I think using 240W would be almost 10% increase which would be awesome but unrealistic! Sorry if not very clear.

I’ll stick with my usual ramps but it’s between tests it’s a good indication it seems if the training is working. Which in just over 6 months I’ve gained over 30W I think it is.

I find that intervals.icu eFTP is about 10W-15W higher than what I can actually hold for an hour.

It’s probably very individual…and might depend on what you expect from an FTP number. A lot of ppl can’t hold their FTP number for an hour & don’t care. FTP is just a metric used to set training plans for many athletes.

Intervals.icu use Morton’s 3-parm model to evaluate eFTP, I think. So depending on where you think you fit on that curve…

not picking on you, but I hear this argument and it makes no sense.

TrainerRoad

  • built workouts designed around Coggan’s classic levels (7 levels or zones)
  • original promoted 8-min or 20-min test for FTP estimation
  • introduced ramp test for FTP estimation
  • didn’t change any workouts after introducing the ramp test

You want an accurate FTP estimate to set training levels/zones for:

  • recovery (level 1)
  • aerobic endurance (level 2)
  • tempo (level 3)
  • sweet spot
  • threshold (level 4)

Doesn’t matter if that comes from ramp test, 8-min test, 20-min test, or some other test. TR workouts are scaled to Coggan classic levels. And those levels assume you have a good estimate of FTP. If the 20-min test gives more accurate results than ramp test, then use the 20-min test to estimate FTP! Its that simple.

And above FTP you may need to personalize the % FTP used in workouts, as there is a lot of individual differences on ability to do work above FTP (vo2max and anaerobic capacity). Its why Coggan updated the classic model and introduced Coggan iLevels (individualized levels). You may fall into the middle of the bell curve and find TR’s 120% for 3-min vo2max intervals is perfect. Or you may fall outside the bell curve and need to increase or decrease that %.

7 Likes

I’d also be interested to understand better how the intervals.icu eFTP is computed (e.g. from which parts of which workouts) … background is that I’m terrible at the ramp test (that’s a whole other topic…) but find that the FTP at which I find (from experience) TR workouts to be consistently achievable is very close to the Intervals.icu eFTP (only just started out using Intervals.icu, so hadn’t seen my eFTP before reading this thread)

I also do poorly at ramp tests and find that Strava, Zwift and Intervals.cu all give very similar results.

I’ll let intervals.icu answer questions about the nuts and bolts of their methodology, but if you are interested in the finer points of the 2 & 3 parm models:

2 Likes

For me I think it can be skewed by shorter efforts, well unless I’m stronger than I think :grinning:.

FTP is 321 (from 60 minute power).

Icu increase to 324 (7m @ 380w)

Icu increase to 335 (3m20s @ 448w)

Icu increase to 337 (3m 10s @ 454w)

I’m happy at the 321 so have left it there.

Really like ICU, but I think my shorter effort power will send the algorithm off.

1 Like

Yes if your short duration power is very good Intervals.icu will overestimate your FTP. You can set the minimum duration that it will use in /settings. Maybe you should go with 7 or 8 minutes.

How it works, short version: It is essentially a generalisation of the 95% of 20m power rule of thumb to any duration from 3 minutes to about 30 minutes. So you just have to do one max effort of any reasonable duration to get an eFTP number.

Long version: Intervals.icu looks for points on your power duration curve of at least 180s (configurable) that indicate that you should be on a higher curve that then one for your current eFTP. If an effort puts you on a higher curve then your eFTP (from the power model parameters for that curve) goes up. If it does not then it wasn’t a max effort and is ignored i.e. your eFTP is left alone. Your eFTP will decay slowly if you reduce your training load and then you need to go do a max effort to get a new number.

6 Likes

I found that the eFTP was quite close to my ramp test TFP, as well as longer Empirical Cycling/Kollie Moore FTP test that I did. The Empirical cycling test was longer than 30 minutes, but I did set a 20 minute PR during it. I don’t have the numbers handy but all of the 3 estimates were within about 2% of each other. I generally don’t do any short all-out efforts, so the model had none of those to work with :slight_smile:

To put my n=1 into the mix, I did a ramp test two weeks ago TR FTP was 351 and icu eFTP was 341 with minimum sample time set at 10 minutes. I have relatively large 1 and 5 minute powers compared to my longer duration performance so I set my minimum time a bit higher. This all being said, as far as I know, icu is not programed to interpret ramp test data so I theoretically could change my minimum time such that my eFTP linde up with my ramp test. In conclusion it doesn’t really bother me, there will never be a computer program that replaces listening to my body and adjusting as needed.

1 Like

Thanks - yes, that’s likely the case with me. One related question: I see 2 slightly different eFTP figures, one in top right of the Fitness screen, and another in bottom left corner of the chart on the Power screen?

I find TR and Intervals.icu to match really close, within 1W right now. But Xert is all over the place. Month ago it had my FTP 23W higher than my TR ramp test. Now it has me 16W lower. Xert did have my FTP within 2W of TR for almost a week, but it oscillates way too much for my liking.

I tend to find it is fairly consistently roughly 5 watts higher then from the ftp test with TR.