Weighed power, a stupid metric?

The one used by Strava...
every time I do a time trial, it seems the weighed average is lower than the actual average by a few Watts, which seems absurd... almost as if I was so regular in my output, that I exceed the steady state...

The normalised power, as used by Garmin Connect and others, seems more realistic, being just slightly over the real average by a handful of Watts

Is it just me?
left the forum March 2023

Comments

  • N0bodyOfTheGoat
    N0bodyOfTheGoat Posts: 5,906
    edited April 2022
    Strava's "weighted power" is their own algorithm, not the one derived by Dr Coggan, which the Strava plugin Elevate (formerly known as Stravistix) uses.

    Coggan's formula tends to give me higher values, last night's ~2 hour ride was 261 vs 253.

    Elevate now has a beta desktop version of the plugin, just make sure you backup your Elevate ride history data, a recent release accidentally wiped loads of old data... Which takes an age to get back, as Strava limits apps to ~600 activities every 4-5 hours and there might well be an additional daily limit, getting back ~4300 when I remembered to continue the sync took several days!
    https://chrome.google.com/webstore/detail/elevate-for-strava/dhiaggccakkgdfcadnklkbljcgicpckn?hl=en
    ================
    2020 Voodoo Marasa
    2017 Cube Attain GTC Pro Disc 2016
    2016 Voodoo Wazoo
  • ugo.santalucia
    ugo.santalucia Posts: 28,301
    There clearly is a bug of some sort... if my output was perfectly linear, then the weighed power should be = to the real power... any deviation should lead to higher weighed power than the real... so I don't understand how it can be lower... it just makes no sense as a number
    left the forum March 2023
  • Yes, Strava's weighted power is notoriously inaccurate. Their is various Strava support threads with people complaining about this. Strava claim there is nothing wrong with it.

    It is generally 20-30 watts lower than the NP reading given by both of my power meters.
  • joe2019
    joe2019 Posts: 1,338
    You have a power meter, why do you care what Strava says it is?
  • ugo.santalucia
    ugo.santalucia Posts: 28,301
    joe2019 said:

    You have a power meter, why do you care what Strava says it is?

    Weighed power is based on power data from the PM... it's a way to estimate the real effort involved... for instance if you go up and down a lot, you will find that the average power is quite low, but you are knackered... normalised power or weighed power gives you a better estimate than the real average. It's useful in TT to get an idea of how hard you have gone over different courses...
    For instance in the hilly TT we did in March my average power was 255, whereas yesterday on a flat TT I averaged 275... however, the normalised values are almost identical
    left the forum March 2023
  • joe2019
    joe2019 Posts: 1,338

    joe2019 said:

    You have a power meter, why do you care what Strava says it is?

    Weighed power is based on power data from the PM... it's a way to estimate the real effort involved... for instance if you go up and down a lot, you will find that the average power is quite low, but you are knackered... normalised power or weighed power gives you a better estimate than the real average. It's useful in TT to get an idea of how hard you have gone over different courses...
    For instance in the hilly TT we did in March my average power was 255, whereas yesterday on a flat TT I averaged 275... however, the normalised values are almost identical

    Yes, of course, but you get a normalised power figure from your power meter, just wondering why you care what Strava says?
  • ugo.santalucia
    ugo.santalucia Posts: 28,301
    edited April 2022
    joe2019 said:

    joe2019 said:

    You have a power meter, why do you care what Strava says it is?

    Weighed power is based on power data from the PM... it's a way to estimate the real effort involved... for instance if you go up and down a lot, you will find that the average power is quite low, but you are knackered... normalised power or weighed power gives you a better estimate than the real average. It's useful in TT to get an idea of how hard you have gone over different courses...
    For instance in the hilly TT we did in March my average power was 255, whereas yesterday on a flat TT I averaged 275... however, the normalised values are almost identical

    Yes, of course, but you get a normalised power figure from your power meter, just wondering why you care what Strava says?
    I was just wondering if it is just me, or this anomaly is in essence a problem in the algorithm…
    Strata is a good archive of previous activities, much more accessible than Garmin Connect

    left the forum March 2023
  • joe2019
    joe2019 Posts: 1,338

    joe2019 said:

    joe2019 said:

    You have a power meter, why do you care what Strava says it is?

    Weighed power is based on power data from the PM... it's a way to estimate the real effort involved... for instance if you go up and down a lot, you will find that the average power is quite low, but you are knackered... normalised power or weighed power gives you a better estimate than the real average. It's useful in TT to get an idea of how hard you have gone over different courses...
    For instance in the hilly TT we did in March my average power was 255, whereas yesterday on a flat TT I averaged 275... however, the normalised values are almost identical

    Yes, of course, but you get a normalised power figure from your power meter, just wondering why you care what Strava says?
    I was just wondering if it is just me, or this anomaly is in essence a problem in the algorithm…
    Strata is a good archive of previous activities, much more accessible than Garmin Connect


    Fair enough.

  • wavefront
    wavefront Posts: 397
    I’ve got a number of rides in Golden Cheetah (same numbers derived from Wahoo head unit) where my NP is less than my average by a watt or too - as you said those rides where the effort is very constant. So it happens outside of Strava too. However I trust the numbers in GC.

    I don’t have a link, but read somewhere that For anything less than around 20 odd minutes NP should be ignored, and av power is more telling. Over 20 mins, NP starts to be quite useful. I’ll see if I can dig it out. I think it’s those occasions, on shorter efforts. where NP can be quite a lot less that AV. So when measuring shorter intervals, for example always measure using AV power.

  • me-109
    me-109 Posts: 1,915
    Blah! Blah! Blah! None of it gives you bragging rights down the pub / teashop / macchiolattecino coffee house. Not like MAX POWERRRRRRR.......

    :D
  • dannbodge
    dannbodge Posts: 1,152

    Strava's "weighted power" is their own algorithm, not the one derived by Dr Coggan, which the Strava plugin Elevate (formerly known as Stravistix) uses.

    Coggan's formula tends to give me higher values, last night's ~2 hour ride was 261 vs 253.

    Elevate now has a beta desktop version of the plugin, just make sure you backup your Elevate ride history data, a recent release accidentally wiped loads of old data... Which takes an age to get back, as Strava limits apps to ~600 activities every 4-5 hours and there might well be an additional daily limit, getting back ~4300 when I remembered to continue the sync took several days!
    https://chrome.google.com/webstore/detail/elevate-for-strava/dhiaggccakkgdfcadnklkbljcgicpckn?hl=en

    What he said ^^
  • davidof
    davidof Posts: 3,058
    edited August 2022
    It is basically just a quadratic mean. It should never be less than the average. @ugo.santalucia explains why quite well why.
    BASI Nordic Ski Instructor
    Instagramme
  • wavefront
    wavefront Posts: 397
    davidof said:

    It is basically just a quadratic mean. It should never be less than the average. @ugo.santalucia explains why quite well why.

    https://help.trainingpeaks.com/hc/en-us/articles/218243287-Why-is-my-normalized-power-lower-than-my-average-power-