Efficiency Factor


Just wondering why TR doesnt calculate or track efficiency factor? (Normalised Power / Av Hr).

I’ve been using TR for about 4 years now so repeat many workouts and a spreadsheet of my efficiency factor for each workout seems to correlate well with my FTP and also recovery. In fact I can normally predict FTP changes by looking at efficiency factor for oft repeated rides (Pettit, Baxter etc done during rest week).

Its also a good metric in itself as max HR probably doesnt improve during training so getting more watts per HR beat sounds like something good to track.


Intervals.icu (my web app) has this and a lot of other interesting power vs HR data. Your rides need to be on Strava.


TrainingPeaks has tracked EF for me since getting a power meter in late 2016. To be honest I usually ignore it, along with aerobic decoupling, except for certain phases (early base, after a season break). I find a more reliable predictor of FTP changes is going out for a sweet spot ride and coming back to see they were threshold intervals. That and doing a hard Wed night world ride and having IF plus or minus 1.0.

When I was using Erg on the trainer, seeing a consistent drop in average HR for sweet spot intervals was a another clue of increasing fitness - for example EF going from 1.38 to 1.33 just doesn’t “pop out” to me, versus average HR dropping from 150 to 145 (and I’m a numbers guy).


Thanks for the inquiry! Knowing what we know about the inconsistency of heart rate, we feel it should be used as a compliment to your power data, but not something we can rely upon as an invariable metric. We have known of the many factors that work against gaining repeatable results from heart rate, including:

– Race nerves
– Level of fatigue
– Riding at different elevations
– Fluctuating stress
– Level of hydration
– Nutrition
– Caffeine intake
– Position on the bike
– Level of recovery
– Lingering illness
– The list goes on …

That being said, I think this article speaks most effectively to why heart rate isn’t a consistent enough unit of measurement for TrainerRoad to invest resources into analyzing that data. Instead, we’ll keep improving upon and building upon the systems we have in place to help athletes better measure improvement! :+1:

HR is a pretty good compliment to power in my opinion, but using it effectively does require some context and knowledge of yourself. So I’m a bit off put by some of TR’s positions on it, but there are free and paid tools out there for anyone interested in learning to use it.


Just out of curiosity, can’t this be said about power too?

Just like having a “bad day” in terms of HR being higher than usual for any number of the reasons you just mentioned. Does this no apply to power too? I am fairly certain that my “FTP” changes from day to day too

Happy to expand upon that! So for example, 200 watts is always 200 watts. Sure, you can have days where your performance varies, but that doesn’t change power as an objective measurement of your effort, whereas HR in response to that effort can change (and we won’t always know if that’s because of some change in fitness or if it’s because of some external factor).

1 Like

Thank you for that. I agree that the 200 watts is always 200 watts. But a persons RPE at 200 watts will differ? Some days that could be a zone 3 effort, another day it could be a threshold effort etc, Which you would feel, and your HR would differ accordingly.

So if you train with power, you will always work out with the same watt/kJ, but your physiological effort will differ. If you workout based on HR, your physiological effort will be “constant” but the watt/kJ would differ?

1 Like

Agreed. Comparing the pure measurements:

  • Power = absolute measurement and consistent data from day to day

  • Heart Rate = absolute measurement but variable value from day to day (at a minimum, not to mention within a day)

  • Rate of Perceived Exertion = relative measurement and potentially variable day to day (at a minimum, not to mention within a day)

On the surface, it would seem Power is superior. But as mentioned, our ability to hit a particular power, and more importantly, the RPE for that effort WILL likely be variable. We all know that hitting numbers in some workouts just plain doesn’t work some days, and then there’s the ones where it feels like we are destroying our prior limits.

Nothing in this area is as simple or cut and dry as we all would like. Power is great, but it is far from perfect when you consider the machine creating it (us) is highly variable in it’s own right. Despite RPE and HR having their known variability, I see real value in both.

Most importantly, I think a practical use of all three data points are worthwhile. Discounting any of them seems like a mistake to me because there are any number of reasons that any of them can be considered junk on their own, while one or the other two may be king on a given day or effort.

Like most things, it’s worth the effort to understand the pro’s and con’s of each, in an effort to take the best from each one and use it to our benefit.

1 Like

Hi Ivy,

Your comments and article are accepted but I am not sure if you grasped my point.

More specifically, I am interested in looking at my efficiency factor when -

  • I am doing an 0.5 - 1.5 hr TR indoor workout (which I have repeated many times)
  • On my bike which has a static set up

I also tend to eat & drink the same things during those workouts so most of the variables above arent really variables, and the ones that are variables are most definitely of interest (eg level of recovery and lingering illness) to my performance on the bike.

My point was that efficency factor is a good metric to bear in mind considering the other variables you mentioned as it is a “measure of aerobic efficiency specifically on low intensity aerobic rides” Joe Friel also provides more detailed reasons for this metric in the “Power Meter Handbook” which is worth a read.

So personally after 4 years of TR rides training with power and just under 1400 rides on trainerroad, I have started looking at power efficiency on workouts I have repeated many times over several seasons. Please note this isnt a big data exercise its just me logging and looking at my rides on excel. From my data, efficiency factor appears to give me a good indication of my condition at that point in time which work with the FTP tests as these are only done every 4-6 weeks on most TR plans. I assume Joe Friel also found it of use during coaching as he has included it with training peaks as a metric.

Normalised power isnt a good metric to measure your condition on a base level ride if heart rate isnt included because of the variables you mentioned above. For example, caffeine and my motivation level will increase normalised power but they also increase heart rate accordingly so dont seem to impact efficiency factor. Recovery from illness or too much TSS will drop my normalised power relative to heart rate so will reduce efficiency factor.

So for example Nates efficiency factor on Baxter (a workout he repeats many times) would give him a good measure of his condition at that point in time and an indication of the success of any experiment he was running to try and improve his performance. In my experience it would also give him an indication of whether he was due an uplift or reduction in FTP at his next test.

Anyway that was my point on efficiency factor but if you want more details on the metric have a look at Friels book or Training peaks etc.

One final comment linked in with efficiency factor is that although the TR software is excellent, one issue is that it has no ongoing measure of an athletes current (daily) condition. A TR rider may need to back off training if they are coming down or recovering from illness, have too much TSS or an experiment on nutrition isnt working etc. Therefore the athlete has to find a way to evaluate their condition (perhaps by measure resting heart rate, heart rate variability, qualitative feeling or efficiency factor on a ride) to know when to back off or ramp up training. Effectively the TR rider has to self coach in these areas as there is no TR metric for guidance.

However, I did notice Training peaks has started to link with heart rate variability software. My experience to date measuring heart rate variability is that it is a bit hit and miss as it often appears to me to be a lagging rather than a leading indicator. HRV and the best ways to self evaluate condition during training between FTP tests could probably make a good podcast topic by the way.

Anyway keep up the good work.


Thank you for digging a bit deeper into this idea, Neil.

We certainly agree that TrainerRoad can improve in this area, and have some things in the works that we think will help! :wink:

Regarding heart rate and Efficiency Factor, I think its important to speak to our goal of keeping our training tools as simple, yet effective as possible. When we created performance analytics, we wanted to focus on providing simple, yet powerful metrics in a manner that’s clean and easy to understand for the average cyclist. We try to focus on and develop tools we know will help athletes get faster, and while this isn’t to say that Efficiency Factor isn’t insightful/useful, we just don’t have plans to incorporate this metric to TrainerRoad this time.

That being said, we’re always trying to improve, and we really do appreciate this feedback on the tools you’re finding useful to get faster with. We’ll certainly pass this feedback along to the team!


I wanted to share my own experience with the efficiency factor idea as applied only to my endurance riding…

for the past year, I’ve been riding my endurance based rides an average of 2 hours with a standard deviation of 1 hour… with my shortest endurance rides being 1 hour, and my longer usual rides being 4 hours. my record longest was 6 hours.
I have seen basically no correlation between EF of my endurance rides and duration…
this graph is a scatter plot of rides from July 2020 till now… (27 rides)

i make a 4 workout running average/max/min graph running… i’m wondering if anyone else has a similar graph to share? I have 80 endurance workouts in my excel table… and they are plotted here.

lately my EF has plummeted and i attribute it to my shift in focus to knocking out two lifting sessions per week with nothing but very light endurance and recovery rides on bike. starting inmarch… i got more regular with my endurance riding and i switched to outdoors as well as trying to drive up my FTP interval durations… i kinda suspect these were fatiguing… IDK…
Ultimately at this time I really feel similarly to what IvyAudrain said… there are just too many factors always fluctuating… keeping them all straight is impossible. and if you can’t keep them straight then it makes it impossible to glean any valuable insight from overall trends…

I’m going to keep adding data and see if there might be some year-by-year trends but i won’t hold my breath…
but one thing I do know for a fact is that in the very long term (100 year trend), my EF will go to infinite because my HR will go to 0…

I track this for the Lento rides indoor, I’ve been targeting 75% of FTP. Here’s the visual:


I don’t spend much time looking at EF.

WKO has a lot of graphs, lately I’ve been looking at the Performance dashboard as it shows trend lines (along with peak power) at various durations. For example showing only the trend lines:

for this season that started mid/late August. It is easy to see increasing average power @ durations of 3 hours, 2 hours, 90 minutes, and 60 minutes. I haven’t done any long sweet spot or threshold intervals, so those trends represent a focus on increasing aerobic endurance by ‘pushing up from below.’

Back to your point on EF, this is an older article on the topic:

Source: https://www.trainingpeaks.com/blog/efficiency-factor-and-decoupling/

Note the use of “few weeks” and for what it is worth, I find it harder to see trends in the EF graphs unless I add a trend line. And even then I don’t spend much time looking at EF. Might want to trial WKO5 software if you like looking at data, and want a large collection of graphs automatically generated from your data.

1 Like

I’m confused… the x axis is duration of workout?
I take it that all of the rides at the 2 hour mark are the “Lento” rides? I don’t know what the protocol is for that TR workout and can’t access them…

yah… i have wko5 too… But I find it frustrating to use… It is a little bit of a “black box” for me.
if only i didn’t have a full time job, I’d watch all of Tim Cusick’s webinars then maybe I’d finally feel competent to use it!!!

takes time to learn, but you do have access to the formulas in graphs/charts/reports. So less black box and more learning curve IMHO.

No, all the blue dots are “Lento” rides of different duration. I don’t use TR.

1 Like

I don’t completely agree, because you need to measure the 200 watts… and power meters can also be more or less accurate, depending on type. Eg loss in power train (dirty bike) or those you need to calibrate etc etc

Hi All,

This topic as it appears to be live again. So I thought I would provide a few more details of why I look at EF for rides & what I use EF for on TR as I didnt provide much of this background in prev post.

Anyway some background on me. I’m and engineer and also interested in cycling coaching and physiology. I’m currently getting my coachinng qualifiications and I’m a level 2 British Cycling Coach (low level).

As an engineer I have also been involved in testing for many years . One recent project involved a hardware in the loop simulation for vehicle suspension testing. For this project we involved some very clever including people from an F1 background, Prof’s who had worked on electronic mtn bike suspension (basis for Fox stuff now), consultants etc and upgraded our test rigs for hardware in loop simulations.

One of the inputs into the rig was terrrain so we could input various terrains to see how the suspension performed. These terrains included metalled roads, pave or at the more extreme end tank proving grounds. We could then look at the performance of the system with these inputs. However, if we thought the system / test rig was behaving strangely we could input a known and often repeated input (like a simple sine wave) at a known speed, vehicle model etc. We had then controlled many of the variables using the standard terrain input and were then looking at the system under test and test rig to try and understand why it was behaving strangely.

So my analagy with TR is that we can go back and do an often repeated workout as a test ride. This is what we do with a ramp test but a ramp test puts a lot of strain on the rider. So an easy ride is better to do if we are recovering or dont want or need the training stress. Perhaps we can then pick one we do many times so we have lots of data. For me thats pettit as I’ve done it nearly 90 times and I can easily hit the Pettit power numbers every time as its not too taxing for me.

So after looking at my EF data for Pettit against FTP (and my general fitness / wellness) there is a strong correlation. I also tend to do pettit on recovery weeks, as a filler ride or when recovering from illness. So the big bonus for me when I look at the EF on Pettit is I can get an idea whether -

  • I am due a bump up or down in FTP
  • I’m still ill or recovering from illness
  • I need more recovery
  • My training is on track.

So thats why I find EF interesting but as its not displayed within TR I do all the analysis on a spreadsheet.

I expect TR has some big data tools and enough user data to do quite an accurate analysis of how accurately EF predicts ramp test results. That could be an interested study and it may prove me wrong.


1 Like

Reading between the lines I think its a marketing decision. TrainerRoad as a product isn’t looking to be as data heavy as say WKO5 or Golden Cheetah. As a product they are trying to position themselves as an easy to use training product where you do your workouts and can track progress pretty easily.

Thankfully Intervals.icu tracks EF, CTL and other useful metrics and is free. I’d say its a great compliment to TR if you’re more of a data nerd than the average TR user (the average TR user isn’t in the forum for instance)