Fitness, form, and function on intervals.icu read 80, 79, and 1%, however in Athletica, they read 44, 62, and -17, which seems way off. Strava essentially matches intervals which causes me to believe those numbers are correct. I can also say that looking through the calendar does seem to indicate all of my data has been imported .
Any idea what is going on here?
Hi everyone,
I’d like to follow up on this thread, as I’ve been wondering about the same thing. I currently have to manually export completed workouts from TrainingPeaks to Athletica, so I end up reviewing data in both platforms. While doing this, I noticed some significant differences in how the metrics are calculated, even though my thresholds are set correctly and are identical on both platforms.
For example:
I’ve been trying to understand the calculation models used by each platform, but I haven’t found a clear explanation yet. If anyone could shed some light on why these metrics differ so much, I’d really appreciate it!
Looking forward to learning more about these data differences. 
Thanks @butiki2 and @holkatriatlonka for the interesting questions!
One of the reasons fitness, fatigue, and form values might differ between platforms like TP, Intervals, Strava, and Athletica is that Athletica calculates training load separately for each modality (run, bike, swim, combined). You can select the different modalities in the Your Performance Potential chart to assess specific values. Importantly, the values shown in Athletica reflect the primary race you’re preparing for: if it’s a marathon, Athletica displays “run” fitness; if it’s a triathlon, it shows “combined” fitness. Does this help clarify?
When it comes to calculating training load, Athletica uses a variety of methods depending on the session modality and the data available. For instance:
For cycling, power is used if available.
If power isn’t available, heart rate (HR) is used.
If neither power nor HR is available, session RPE (sRPE) is used.
For running, speed is used if available.
If speed isn’t available, heart rate (HR) is used.
If neither speed nor HR is available, session RPE (sRPE) is used.
Additionally, training load calculations are influenced by your threshold values. If platforms like Strava, Intervals, or TP are not using the same thresholds as Athletica, it could result in discrepancies.
I hope this explanation helps clarify the differences you’re seeing between platforms! Let us know if you have further questions.
I do appreciate the reply! I would add that the numbers quoted for athletica come from the combined setting on the performance chart. I would also add that probably 95% of my data, (certainly any with intensity), comes from cycling, primarily indoor. while i certainly wont expect them to be identical however when they are off by ~25% it doesnt provide me with a great deal of confidence.
Dear @butiki2 from this sentence I would understand that Athletica fitness and form are 44 and 62, which seems odd. I think you refer to fitness 44, fatigue 62 and form -17? I’m not familiar with “function” in % or other calculations done on intervals. If you want I can have a look at your profile if you DM me with your email or account name. It can well happen that calculations are done in a different way, also because TSS is TP proprietary. Making comparison across platform hard.
Fatigue and fitness are in arbitrary units, which makes it even more difficult to interpret/compare/assess. What is more important is to keep the calculations consistent within the same platform.
Dear @butiki and @holkatriatlonka,
Just replying back with my humble 2 cents here. I have used WKO/Training Peaks since 2008, with just entering in session data by hand to experiencing the transformation to Garmin Express to Garmin Connect. This year, I cancelled my Training Peaks Premium account and I don’t have the connection from Garmin Connect to Strava active. I just don’t feel it’s useful to have my data out there for the world to see. That said, it’s fun to see how you compare on a certain segment.
For many years, I number chased. I strived to have that ever increasing blue line at all costs. I tried to maximize my numbers (TSS, NP, IF, NGP) ETC. I thought that having a certain level of accumulated training load defined me as an athlete. It lead to over-emphasis on the wrong things in a performance and race management situation. Now, I think the best approach is optimization of the data in a way to add confidence to reach a peak race. It’s important to realize it’s not 1 session but a massive amount of sessions over a long period of time that will enhance performance.
These data processing platforms are, all just math to make you contextualize how you feel from a given training load. Cross-platform comparisons are really hard because most do not disclose the calculation, use a static “threshold” and there are many different types of duration based threshold’s the user can input. Also, there is inherent error in the sensors we use. There is day-to-day variation and environmental variation. Just be cognizant of that when driving down in the data details. I think a precise sensor is the better one to own than a very accurate but imprecise sensor.
I hope this helps and doesn’t give you the wrong impression that I am judging you for diving into the data. I applaud you for questioning, but I am just trying to give you a better path versus the path that I followed for too long. Try to enjoy the journey and do the sport/session that makes you happy.
@amn2099 Thanks for the thoughtful feedback, however i feel like my approach to the data is a bit different than what you describe as your previous view. Historically i have viewed it as an interesting distraction, worthy of consideration, but certainly not something which defined or solely directed my fitness/training.
That being said given that Athletica uses this data to quantify my current fitness and create a personalized training plan, the quality and precision of the data takes on a higher value. I certainly understand that there will be differences between platforms, but given they are based on underlying algorithm, i would expect a higher level of correlation.
@andrea not sure how to dm you.
you are correct it is fitness 44, fatigue 62, and form -17 in athletica, and fitness 80, fatigue 79, and form 1% in intervals.icu. In intervals.icu you have the option of presenting form either as a percent or a whole number, with the default being percent. I did toggle the setting to whole number which change 1% to 1. I understand there will be some variation but given that they are all based on the work done by Dr. Bannister, (and since strava and intervals.icu match pretty closely), i have some concern that there is an issue with my data that is throwing off the result in athletica.
Great to have a conversation going on this, as I believe understanding these numbers are part of our journeys as athletes. I’ll throw my two cents here as well.
Bannister’s Training Impulse Model was groundbreaking way to somehow quantify training to allow for planning and monitoring of training. For specially interested, you can find more info in our education section…
Guys , remember that when you have a dynamic threshold detection, compared to a static threshold detection, you are going to get differences in the total load calculations. Strava and Intervals.icu and TP all have static threshold, so the load don’t change much.
Let’s say you do a threshold session (bike or run) close to your CP/ Threshold/ VT2 (FTI 95%) in November, and the load calculation is 95ish. Then you repeat the same session in April, and because you’ve had a great solid base season, you can now crank out more power/pace at CP/Threshold/VT2, but you are using a static threshold from months ago - you are going to get a FTI 105 (because you are now cranking out much more watts/pace) and load 105+. But, in Athletica, automatic detection has been detecting ever so slight increases in your VT2 / CP throughout your base season, and is still giving you a roughly same number of 95. Does this mean you’ve become less fit?
No, of course not.
You are now actually able to produce more external power. “But the Fitness numbers are then not keeping up with me, I should be getting more load points?”
And I think this is where you have to be careful when comparing TSS, Fitness etc numbers across platforms. Yes, using a static calculation relative to an “outdated” threshold, can produce higher load - and artificially higher “Fitness”. What if you are now chasing higher fitness numbers and forgotten to listen to your ever more fatigued body? Are you truly absorbing all that training, or digging yourself to a deeper and darker overtraining hole?
Do the numbers match fitness and our true capacities? I dare to claim NO!
Does your body “know” your fitness number? NO!
Can you “outrun/outbike” yourself at Fitness #80 vs. yourself at Fitness #120? YES! Explanation: you can have insanely high Fitness number but you are stressed AF, overtrained with exhausted HPA (read Prof’s article about unhealthy athletes) and unable to run a 5K with HR through the roof and banger of a headache.
Sorry my long rant, but as you can see I’m just so passionate about this topic. Chasing numbers took me to a severe overtraining, an experience I wish on no-one. Luckily, I was able to climb back out to light with guidance from @Prof. Long story short, but I turned my Fitness/fatigue/form numbers off and reached new heights because I’m not chasing numbers.
Yes, data is fun, useful and interesting, but anytime we talk about humans, there’s not a single math calculation that is close to the truth. We are much more complex than that. Like Prof always tells me: “Feeling trumps everything”.
MJ
Very insightful post! The static vs dynamic thresholds is a very important point, that makes a huge difference.
I probably adjust my threshold setting in intervals.icu, (for a variety of reasons), several times a month. Wouldn’t this negate the significance of the static vs. threshold setting.
It interesting that some much talk seems focused on other applications calculations fueling over training when in my case its a bit of the opposite. Fitness and fatigue reading higher in intervals.icu indicates to me that i am closer to my limits, which given the mix of riding i have done seems more accurate.
Still, while fitness and fatigue numbers in AIE are ~25% off, (and in my case lower), from the other apps, the difference in form is less significant.
Interesting, what workouts/data do you use to adjust your thresholds that often?
Also, I am wondering does Athletica use heart rate to calculate training load in workouts where heart-rate is assigned/used by Garmin to set the workout targets? Or does it still use power/pace?
Hello,
@Andrea outlines the priority of session data that is used for actual training load calculation. For cycling: power, hr, or session RPE is used in that order. For running, it’s pace, HR, session RPE in that order. They have not implemented running power, yet since there is a more complicated handling involved because certain situations in outdoor environments pace or power might be better to use.
For prescribed training load, the user has the option to select how training load is prescribed. It’s in the settings, profile, plan settings
Also, not to get off topic here but I think it’s worth to talk about prescription of training, what you actually plan to do on any given day, and how you actually feel. The best signal is in the brain, and no one math model can tell us what is best to do in the moment. Essentially, all math models are wrong, some can be useful.
Hey,
I am not understanding what you mean with
What are you basing your decision to change your thresholds so often? Day form? I don’t think physiologically your Power or Pace at your VT2 change much over a month. We had many discussions internally how sensitive our automatic detection should be, and definitely day-to-day variation is too sensitive. We are also biased when manually changing frequently… day form? Too optimistic? Bad sleep? Angry? Too many factors to consider to change your threshold often… and what difference does it really make?
My old Ski Prof used to always ask: “yes, it’s great, but does it work, and if it does, what difference does it make?”. Basically, so what?
In my opinion you may be basing your training decisions too heavily on arbitrary numbers if you go by fitness/fatigue (regardless of platform)…
Would you not “know” when you are closer to your limits by how you are feeling, how motivated you are, intuition? Surely you can tell if you are feeling fatigued or not.
Best, MJ
What i was trying to say in the quote you provided was that based on my activity the data in intervals.icu more accurately represents what i believe to be my current physiological state than the data in Athletica. (Fitness and fatigue are high and balanced due to a zwift stage race i did earlier this month combined with a good amount of recovery).
I typically change my FTP setting based on feedback from either zwift, (when it tells me it has changed) or based intervals.icu’s estimated FTP feature, (i had a big 10w jump back in early november to 259w due to holding 268w for 23 minutes during a race, since then its slowly drifted back down to 249w).
While i enjoy looking at the numbers, they don’t really inform my training (specifically fitness, fatigue, and form. I do use FTP in zwift to set levels in their internal training plans, and influencing my category in races). Beyond being a curious distraction, I also might look to them, along with the data from my garmin, to confirm how i feeling on a given day.
My current training “plan” is as follows: Walk for at least an hour daily. Unless i an participating in a zwift stage race, make every other ride zone 2. If i did a zone two ride yesterday and there is no race today, participate in a zwift challenge If i done my zone 2 ride and ther is no race or challenge i am interested in, stick in a zwift structured workout. Lastly, stick in a run every third or 4th day (no ride on this day).
I am not unhappy with my current performance and have seen slow improvement over the years however as someone who is a techno-optimist and a believer in striving for improvement, i was hopeful an application like Athletica might optimize my effort and make me a stronger athlete.
Cheers,
Bob
Is there a way to have run workout load prescribed by heart rate if you don’t use a Garmin? Thanks!
You can try and build one inside the Athletica platform. The best way I found is to take a pre-existing workout based on heart rate, save it to the library. The icon to do that looks like
It’s a bit of a learning curve to save it correctly, just play around with steps in the workout.
If you edit a previous workout, you can see how the creators have structured it.
Thank you for enlightening me on how you use the fitness/form/fatigue numbers.
I don’t really understand the whole zFTP calculations, but remember we use Critical power method vs. zFTP so slight differences there. You can read up on how we use Critical power here..
Those Zwift races and challenges can sneak up on you, IMHO. Stick to Athletica’s plan and sprinkle a race or two on Zwift carefully and you’ll start to see great improvements 
MJ
