Hi Michael,
I give up! How did you generate the data in your table? And how did you derive the Ave HR and Ave Pwr for the 20 minute segments?
When I tried to download the intervals.icu data for my latest LIT ride I got a spreadsheet with 9000 rows (1 per second). I can’t see how to avoid tedious busy work firstly identifying the segments then averaging the HR and Pwr in each segment. I do concede I’m no Excel wizard.though!
Ray: I cheated! I use TrainerRoad, on which you can define intervals of whatever length, and it provides summaries for you.
I have just done all that work for Kosio. It is a bit tedious, but then I never considered that I was going to be in demand! There would need to be a lot of work to make this process applicable to any ride – for example, different rides have different periods of missing data [headunit paused or stopped recording], the rides are of different lengths and so on.
Then you have to decide what you are going to do about Pwr = 0 or Pwr very low [eg coasting to a stop or gently pedalling while going downhill]. These make a real difference to the results.
Sorry about that.
Kosio_Varbenov: that is a good idea.
And yes, on reflection, the HR settings and the power setting do not affect the trends – they are there for Seiler to be able to compare individuals or for you to compare yourself over time.
To everyone else: if you are going to get results that are meaningful and that you can interpret, then my experience is that all data points with Pwr = 0 should be dropped from the analysis. And, as Kosio suggested, all data points where you are just pedalling gently [eg going downhill] also need to be dropped.
On the first question – HR for endurance rides. This depends on your definition of endurance. For some – including me – an endurance ride is less than about 0.70 of HR max, which in practice I round up to 130. Seiler proposes a percentage of HR capacity [= HR max - HR rest] above HR rest. And for both of these, yes, you have to vary power to keep HR below what you have set.Others propose a power definition, such as 55 - 75% of FTP. Choose your weapon.
On your decoupling question – your bottom graph mixes both possible decoupling and problems with the data record. [See my long post with graph and table above]. At the beginning of you ride, there are evidently many zero points and periods of near coasting. So there is apparent decoupling – this should be ignored. In the middle of your ride you have a period of rest, when power is low and HR gradually falls – another period of apparent decoupling which you should ignore. The rest of the decoupling trace is pretty much constant. What is important is not the level of the decoupling trace, but its trend – if it is going up, you need to improve your endurance at that power. If it is flat you have two options [1] ride for the same duration at slightly higher power; [2] ride for longer durations at the same power. Both will provide a stimulus for your body to adapt, according to Seiler.
No need to apologise. You’ve played a leading role in this discussion.
I suspect you search for your 20 minute segments after completing the ride. I approached it from the front end by generating successive 20 min segments in Workout Creator. I’m still searching for a way to extract the segment data without manual invovement!
BTW I was chuffed when I saw that you virtually climbed the Stelvio today. It was THE highlight of our trip to the Dolomites in 2018.
Only nearly, Ray – I ran out of time. And today I did Tormalet. Both in endurance zone.
On your search for 20-minute segments, see this thread in the forum: Average/min/max information on selected part of workout. That will solve all your problems.
Tx for all the analysis and discussion guys. The decoupling calc uses the w/HR ratio not heart rate reserve:
powerHR[i] = watts60s[i] / hr[i]
decoupling[i] = (firstHalfRatio - powerHR[i]) * 100 / firstHalfRatio
I will have another look at how to better clean this data in a little while. I want to try get some HR/w data on the fitness page. That work is using the much more cleaned up and lag adjusted 1m data points used on the other decoupling charts.
Now that you mention this, I have another idea 
It would be great to see how your efficiency compares to riders similar to you FTP wise… like how is the usual power/HR over different zones/watts/%FTP of people with similar fitness levels. Basically who has the best zone 2, or less cardiac drift at different watts/times.
I know this is a bit ahead, but just want it to throw it out before I forget.
Its generally not good to try compare HR based metrics between people as there is a lot of variation. The are good for yourself over time.
At the end of his presentation, this is what Seiler indicates that he wanted to do: to compare people and to compare performance over time by normalising HR to HR capacity or HR reserve and normalising Power to best 6-minute power [or FTZP].
Daniel_Hernandez: remember that this indicator really only works effectively for steady power sessions. A good 20 minutes to get the HR / power relation stabilised, and then the stable power session over which observations are being made. Effectively this means that the indicator could be used for endurance [Seiler’s idea], tempo or sweet spot continuous efforts.
Hello,
what I understood from the video is that @ 50 % HRR you should be around half of your aerobic capabilities, so quite good to normalize this with 50 % of Max Aerobic Power/Power@VO2max hence the ratio around 1
@david thank you for your work, it is really awesome to witness this site develop the way it is going.
I was using and testing this new graph and today during my training, I did 2x20mnts sweetspot blocks
https://intervals.icu/activities/3384320395
As I was going into the second block, I was watching my HR climbing up slowly but surely, and thus I started to experiment with cadence, I was lowering my cadence and due to this, I was even able to drop my HR to couple points, you can see it between minute 45 and 46.
So, in light of this, I would like to suggest you add to the cardiac drift graph the field “cadence” since it does have an impact and can influence the drift.
Thank you
yup I have noticed that too, but frankly if you want to compare one session to another you try to work in the same cadence to analyze HR or power/HR ratio, usually you work at your natural cadence for this kind of workout, don’t you ?
Yes I do like most people, however, I noticed that cadence is also impacted by fatigue level, usually at the start of the workout I do higher cadence than at the end when in hanging on for my dear life 
Ok, but IMHO Efficiency factor is mostly something useful to monitor endurance and easy training.
So if you are dropping your natural cadence on an endurance ride, you don’t need to have a complex cardiac drift graph or In/Ex S ratio to pinpoint where you are getting really fatigued on a long endurance bout, don’t you ?
Ok, so after a more detailed playing around, I made some interesting observations which are described here.
Interesting to know whether @david, you think this might influence the way the charts look in the future. Also thanks to @Michael_Webber for his input.
Looking forward to hearing your thought
Kosio_Varbenov: your link takes me to Scribd, with a random list of books. Is this a mistake? If not, can you provide another link?
Thanks.
Its says your “doc has been removed” when I try follow the link. Keen to see what you have been up to!
You should send it to Stephen Seiler via Twitter !