Accurate Power Data For Strava Segments

Is there any better way of getting accurate power data on Strava segments?

I find Strava power estimates to be flaky to say the least, TrainingPeaks seems to be much more accurate but they don't seem to play nice together.

I'm a Strava subscriber but I only have a TP Free account if that makes a difference.

This is purely for my own entertainment about how little I'm able to out out rather than taking all the local KOM's.
«1

Comments

  • dannbodge
    dannbodge Posts: 1,152
    The only way to get accurate power data is to get a power meter and go from there.

    No algorithm will ever be able to produce an accurate reading over an actual power meter
  • It's not going to be perfect, but make sure you account for the weight of things including...
    The kit you're wearing
    Repair kit, food, mobile etc. in your pockets
    Water bottles
    Easily removable accessories on the bike like muguards, lights etc.

    If you use the same bike for multiple uses, for instance commuting and a recreational ride where there might be substantial differences in the weight of things in that list, it might be worth setting up a number of different bike profiles that account for those weight differences.

    Also, if using a GPS computer with a barometer, for consistancy I'd use the "correct elevation" feature on Strava.

    One thing Strava still doesn't account for as far as I'm aware is wind direction and speed, I definitely did not do part of my commute in this morning with an unusual easterly tailwind at z6! :D
    ================
    2020 Voodoo Marasa
    2017 Cube Attain GTC Pro Disc 2016
    2016 Voodoo Wazoo
  • mully79
    mully79 Posts: 904
    I assume Strava sucks at power output.
    Me being 19 stone pushing a 42/23 gear ratio up a 10.5% grade Strava thinks I’m outputting 199W.
    Now I’ve swapped to an easier 42/32 I’m suddenly at 230W.
  • ibr17xvii
    ibr17xvii Posts: 1,065
    dannbodge said:

    The only way to get accurate power data is to get a power meter and go from there.

    No algorithm will ever be able to produce an accurate reading over an actual power meter


    I've got a power meter that's my point.

    Strava gives an estimate which is wildly out & I'm not sure how I can narrow down TP data which seems more accurate to a specific Strava segment or even a specific section of a ride.

    I'm not sure that's even possible hence the question.
  • kingstongraham
    kingstongraham Posts: 26,132
    ibr17xvii said:

    dannbodge said:

    The only way to get accurate power data is to get a power meter and go from there.

    No algorithm will ever be able to produce an accurate reading over an actual power meter


    I've got a power meter that's my point.

    Strava gives an estimate which is wildly out & I'm not sure how I can narrow down TP data which seems more accurate to a specific Strava segment or even a specific section of a ride.

    I'm not sure that's even possible hence the question.
    If you have a power meter, why can you not look at the power for the segment on Strava?
  • gethinceri
    gethinceri Posts: 1,510
    Subscribe to Trainingpeaks and stop subscribing to Strava
  • ibr17xvii
    ibr17xvii Posts: 1,065
    edited August 2021

    ibr17xvii said:

    dannbodge said:

    The only way to get accurate power data is to get a power meter and go from there.

    No algorithm will ever be able to produce an accurate reading over an actual power meter


    I've got a power meter that's my point.

    Strava gives an estimate which is wildly out & I'm not sure how I can narrow down TP data which seems more accurate to a specific Strava segment or even a specific section of a ride.

    I'm not sure that's even possible hence the question.
    If you have a power meter, why can you not look at the power for the segment on Strava?

    Because it’s wildly inaccurate IME.

    My average on a ride today was 154w with Strava & 207w on TP.

    Weighted on Strava v NP on TP was nearer but still different.

    I just assumed the power data on Strava segments is estimated & therefore pretty inaccurate unless anyone knows any different.

    I couldn’t get a KOM if I tried, it’s more for my own curiosity really.
  • ibr17xvii
    ibr17xvii Posts: 1,065

    Subscribe to Trainingpeaks and stop subscribing to Strava


    What would that achieve?
  • kingstongraham
    kingstongraham Posts: 26,132
    ibr17xvii said:

    ibr17xvii said:

    dannbodge said:

    The only way to get accurate power data is to get a power meter and go from there.

    No algorithm will ever be able to produce an accurate reading over an actual power meter


    I've got a power meter that's my point.

    Strava gives an estimate which is wildly out & I'm not sure how I can narrow down TP data which seems more accurate to a specific Strava segment or even a specific section of a ride.

    I'm not sure that's even possible hence the question.
    If you have a power meter, why can you not look at the power for the segment on Strava?

    Because it’s wildly inaccurate IME.

    My average on a ride today was 154w with Strava & 207w on TP.

    Weighted on Strava v NP on TP was nearer but still different.

    I just assumed the power data on Strava segments is estimated & therefore pretty inaccurate unless anyone knows any different.

    I couldn’t get a KOM if I tried, it’s more for my own curiosity really.
    If you upload power data to Strava, that's the power data Strava displays.
  • gethinceri
    gethinceri Posts: 1,510
    ibr17xvii said:

    Subscribe to Trainingpeaks and stop subscribing to Strava


    What would that achieve?
    IME TP has better analytics

  • ibr17xvii
    ibr17xvii Posts: 1,065

    ibr17xvii said:

    ibr17xvii said:

    dannbodge said:

    The only way to get accurate power data is to get a power meter and go from there.

    No algorithm will ever be able to produce an accurate reading over an actual power meter


    I've got a power meter that's my point.

    Strava gives an estimate which is wildly out & I'm not sure how I can narrow down TP data which seems more accurate to a specific Strava segment or even a specific section of a ride.

    I'm not sure that's even possible hence the question.
    If you have a power meter, why can you not look at the power for the segment on Strava?

    Because it’s wildly inaccurate IME.

    My average on a ride today was 154w with Strava & 207w on TP.

    Weighted on Strava v NP on TP was nearer but still different.

    I just assumed the power data on Strava segments is estimated & therefore pretty inaccurate unless anyone knows any different.

    I couldn’t get a KOM if I tried, it’s more for my own curiosity really.
    If you upload power data to Strava, that's the power data Strava displays.

    Why such a big difference between the average on Strava & the average on TP then?

    Of the 2 figures I posted above the TP one seems the most plausible knowing what kind of ride it was.
  • ibr17xvii
    ibr17xvii Posts: 1,065

    ibr17xvii said:

    Subscribe to Trainingpeaks and stop subscribing to Strava


    What would that achieve?
    IME TP has better analytics


    I would agree.

    I got 2 weeks free of TP Premium when I 1st signed up but I’d only just got the power meter then & I was just getting to grips with it so didn’t really have much of a clue.

    Now I’m a bit more confident with it I’d like to analyse my numbers a bit more, particularly on Strava segments but it’s quite expensive to upgrade & not sure I’d get the most out of it.

  • kingstongraham
    kingstongraham Posts: 26,132
    ibr17xvii said:

    ibr17xvii said:

    ibr17xvii said:

    dannbodge said:

    The only way to get accurate power data is to get a power meter and go from there.

    No algorithm will ever be able to produce an accurate reading over an actual power meter


    I've got a power meter that's my point.

    Strava gives an estimate which is wildly out & I'm not sure how I can narrow down TP data which seems more accurate to a specific Strava segment or even a specific section of a ride.

    I'm not sure that's even possible hence the question.
    If you have a power meter, why can you not look at the power for the segment on Strava?

    Because it’s wildly inaccurate IME.

    My average on a ride today was 154w with Strava & 207w on TP.

    Weighted on Strava v NP on TP was nearer but still different.

    I just assumed the power data on Strava segments is estimated & therefore pretty inaccurate unless anyone knows any different.

    I couldn’t get a KOM if I tried, it’s more for my own curiosity really.
    If you upload power data to Strava, that's the power data Strava displays.

    Why such a big difference between the average on Strava & the average on TP then?

    Of the 2 figures I posted above the TP one seems the most plausible knowing what kind of ride it was.
    Are you definitely loading the power data into strava?

    Or are you recording using the strava app separately from the recording with power?
  • Why such a big difference between the average on Strava & the average on TP then?


    This is an issue with Strava lots of cyclists have picked up on. There are a few Strava support threads which deal with it.

    If you are using a pm then the data will be from your power files, it is not a Strava estimate. The problem appears to be that Strava uses it's own algorithm which can, in some cases, present a big discrepancy in power data. Strava support tend not to agree with this assertion.

    I have the same issue, Strava weighted power is generally 10% lower than my power files (wahoo and quarq) suggest.
  • ugo.santalucia
    ugo.santalucia Posts: 28,172
    Uphill segments can be quite accurate, when it comes to flat, too many variables, like wind speed and direction, CR of the rider, what bike you ride etc...
    left the forum March 2023
  • ibr17xvii
    ibr17xvii Posts: 1,065

    ibr17xvii said:

    ibr17xvii said:

    ibr17xvii said:

    dannbodge said:

    The only way to get accurate power data is to get a power meter and go from there.

    No algorithm will ever be able to produce an accurate reading over an actual power meter


    I've got a power meter that's my point.

    Strava gives an estimate which is wildly out & I'm not sure how I can narrow down TP data which seems more accurate to a specific Strava segment or even a specific section of a ride.

    I'm not sure that's even possible hence the question.
    If you have a power meter, why can you not look at the power for the segment on Strava?

    Because it’s wildly inaccurate IME.

    My average on a ride today was 154w with Strava & 207w on TP.

    Weighted on Strava v NP on TP was nearer but still different.

    I just assumed the power data on Strava segments is estimated & therefore pretty inaccurate unless anyone knows any different.

    I couldn’t get a KOM if I tried, it’s more for my own curiosity really.
    If you upload power data to Strava, that's the power data Strava displays.

    Why such a big difference between the average on Strava & the average on TP then?

    Of the 2 figures I posted above the TP one seems the most plausible knowing what kind of ride it was.
    Are you definitely loading the power data into strava?

    Or are you recording using the strava app separately from the recording with power?

    I'm recording on a Wahoo Bolt v2 & then it gets uploaded direct to Strava & TP.

  • ibr17xvii
    ibr17xvii Posts: 1,065

    Why such a big difference between the average on Strava & the average on TP then?


    This is an issue with Strava lots of cyclists have picked up on. There are a few Strava support threads which deal with it.

    If you are using a pm then the data will be from your power files, it is not a Strava estimate. The problem appears to be that Strava uses it's own algorithm which can, in some cases, present a big discrepancy in power data. Strava support tend not to agree with this assertion.

    I have the same issue, Strava weighted power is generally 10% lower than my power files (wahoo and quarq) suggest.

    This sounds about right to me.

    Having a good look at my most recent ride the "total" value is way off but the segments themselves look more realistic.

    Hard to tell though obviously.
  • ibr17xvii
    ibr17xvii Posts: 1,065

    Uphill segments can be quite accurate, when it comes to flat, too many variables, like wind speed and direction, CR of the rider, what bike you ride etc...


    i think you're right Ugo.
  • kingstongraham
    kingstongraham Posts: 26,132
    Can you post a link?
  • wavefront
    wavefront Posts: 397
    Strava seems to penalise you for stopping and not pedalling. It may be my style of riding but my numbers are pretty close to my real numbers from my PM (I use GC to analyse which turns out identical numbers to TP). On chain tight rides or TT’s - constant power rides where my AP and NP are very very similar or identical. Strava shows only a watt or two difference from GC. On rides which involve more variation, the numbers are different by up to about 8% on strava vs CG. (Saying that, did a 3hr hilly ride no stops apart from junctions at the weekend with a fair bit of downhill coasting, and my AP between Strava / GC were only 1 watt diff, and NP 10w difference)

    Weather, wind direction etc makes no difference to the numbers, only that they may affect your riding style you may notice some differences , but you’re experiencing a big difference so I understand why you’re puzzled.

    Go out and do a short (over 30mins to get NP to adjust properly) chain tight ride at a constant power with little or no stops. Have a look at TP and strava and see what they show - should be within a watt or two.

    FWIW, I’m sure I have a free TP account and whilst all the analytics aren’t there, it’ll show you NP/AP for your ride to allow you to compare.

    Out of interest what PM are you using?
  • ibr17xvii
    ibr17xvii Posts: 1,065
    wavefront said:

    Strava seems to penalise you for stopping and not pedalling. It may be my style of riding but my numbers are pretty close to my real numbers from my PM (I use GC to analyse which turns out identical numbers to TP). On chain tight rides or TT’s - constant power rides where my AP and NP are very very similar or identical. Strava shows only a watt or two difference from GC. On rides which involve more variation, the numbers are different by up to about 8% on strava vs CG. (Saying that, did a 3hr hilly ride no stops apart from junctions at the weekend with a fair bit of downhill coasting, and my AP between Strava / GC were only 1 watt diff, and NP 10w difference)

    Weather, wind direction etc makes no difference to the numbers, only that they may affect your riding style you may notice some differences , but you’re experiencing a big difference so I understand why you’re puzzled.

    Go out and do a short (over 30mins to get NP to adjust properly) chain tight ride at a constant power with little or no stops. Have a look at TP and strava and see what they show - should be within a watt or two.

    FWIW, I’m sure I have a free TP account and whilst all the analytics aren’t there, it’ll show you NP/AP for your ride to allow you to compare.

    Out of interest what PM are you using?


    Interesting. I have a 4iiii single sided PM.

    That does make sense actually cos on my ride yesterday I mentioned above it was a hilly ride with close to 5000ft elevation that did have a fair bit of downhill & I also stopped twice so I guess that’s why Strava said it was a 154w average but TP much higher.

    FWIW I’ve always thought the Strava estimate was ridiculously low on the averages even before I got the PM.

    It’s not a dealbreaker & it’s certainly not ruining my life but I like to see my data on specific segments to try & track improvements over time. I always go off the average & NP on TP for the total ride anyway.

    If the segments are pretty accurate then that’s good enough. I think you can narrow it down to a specific section / segment on TP but only on the premium version
  • dannbodge
    dannbodge Posts: 1,152
    ibr17xvii said:

    ibr17xvii said:

    dannbodge said:

    The only way to get accurate power data is to get a power meter and go from there.

    No algorithm will ever be able to produce an accurate reading over an actual power meter


    I've got a power meter that's my point.

    Strava gives an estimate which is wildly out & I'm not sure how I can narrow down TP data which seems more accurate to a specific Strava segment or even a specific section of a ride.

    I'm not sure that's even possible hence the question.
    If you have a power meter, why can you not look at the power for the segment on Strava?

    Because it’s wildly inaccurate IME.

    My average on a ride today was 154w with Strava & 207w on TP.

    Weighted on Strava v NP on TP was nearer but still different.

    I just assumed the power data on Strava segments is estimated & therefore pretty inaccurate unless anyone knows any different.

    I couldn’t get a KOM if I tried, it’s more for my own curiosity really.
    It will be down to the formula or algorithm that is used to calculate avg power.

    On weighted power/NP etc, Training peaks uses the Dr Koggan formula and Strava doesn't.
  • kingstongraham
    kingstongraham Posts: 26,132
    Average power shouldn't need an algorithm though, should it? To be that different, one must be ignoring zeroes or stopped time, I'd say.
  • ibr17xvii
    ibr17xvii Posts: 1,065
    I think @wavefront is right.

    Just had a look back at a few rides where I’ve not stopped for any length of time & the numbers are reasonably similar. Strava’s average weighted is still lower by between 15-20 watts but it’s closer.

    Also agree with @kingstongraham , an average should be an average, don’t see why an algorithm needs to get involved.
  • pblakeney
    pblakeney Posts: 25,596
    ibr17xvii said:



    Also agree with @kingstongraham , an average should be an average, don’t see why an algorithm needs to get involved.

    At a guess, counting or negating times of freewheeling towards the overall average?
    The above may be fact, or fiction, I may be serious, I may be jesting.
    I am not sure. You have no chance.
    Veronese68 wrote:
    PB is the most sensible person on here.
  • kingstongraham
    kingstongraham Posts: 26,132
    Are you sure it's Strava that's inaccurate?
  • ibr17xvii
    ibr17xvii Posts: 1,065
    pblakeney said:

    ibr17xvii said:



    Also agree with @kingstongraham , an average should be an average, don’t see why an algorithm needs to get involved.

    At a guess, counting or negating times of freewheeling towards the overall average?

    I think so.
  • ibr17xvii
    ibr17xvii Posts: 1,065

    Are you sure it's Strava that's inaccurate?

    No in all honesty! I suppose I’m presuming the PM is more accurate (which it should be) & how I’m feeling when riding.

    Ride today: easy spin, no stops apart from traffic lights, no elevation to speak of (240ft) just cruising basically - Strava average 136w, weighted 144w - TP average 163w NP 155.

    Based on the numbers I was seeing on screen & how I was going 136w seems low to me, TP 163w seems about right.

    “It’s a rabbit hole” as Shane Miller would say. Think I’ll just accept the differences & go with TP. Like I said it’s not ruining my life, I was just curious.
  • kingstongraham
    kingstongraham Posts: 26,132
    ibr17xvii said:

    Are you sure it's Strava that's inaccurate?

    No in all honesty! I suppose I’m presuming the PM is more accurate (which it should be) & how I’m feeling when riding.

    Ride today: easy spin, no stops apart from traffic lights, no elevation to speak of (240ft) just cruising basically - Strava average 136w, weighted 144w - TP average 163w NP 155.

    Based on the numbers I was seeing on screen & how I was going 136w seems low to me, TP 163w seems about right.

    “It’s a rabbit hole” as Shane Miller would say. Think I’ll just accept the differences & go with TP. Like I said it’s not ruining my life, I was just curious.
    If the power number is coming from the PM in both cases, it's not PM vs Strava, it's TP vs Strava.

    Normalised power should never be lower than average power though (for a whole ride assuming it's over 20 minutes), so something looks wrong there.
  • ibr17xvii said:

    wavefront said:

    Strava seems to penalise you for stopping and not pedalling. It may be my style of riding but my numbers are pretty close to my real numbers from my PM (I use GC to analyse which turns out identical numbers to TP). On chain tight rides or TT’s - constant power rides where my AP and NP are very very similar or identical. Strava shows only a watt or two difference from GC. On rides which involve more variation, the numbers are different by up to about 8% on strava vs CG. (Saying that, did a 3hr hilly ride no stops apart from junctions at the weekend with a fair bit of downhill coasting, and my AP between Strava / GC were only 1 watt diff, and NP 10w difference)

    Weather, wind direction etc makes no difference to the numbers, only that they may affect your riding style you may notice some differences , but you’re experiencing a big difference so I understand why you’re puzzled.

    Go out and do a short (over 30mins to get NP to adjust properly) chain tight ride at a constant power with little or no stops. Have a look at TP and strava and see what they show - should be within a watt or two.

    FWIW, I’m sure I have a free TP account and whilst all the analytics aren’t there, it’ll show you NP/AP for your ride to allow you to compare.

    Out of interest what PM are you using?


    Interesting. I have a 4iiii single sided PM.

    That does make sense actually cos on my ride yesterday I mentioned above it was a hilly ride with close to 5000ft elevation that did have a fair bit of downhill & I also stopped twice so I guess that’s why Strava said it was a 154w average but TP much higher.

    FWIW I’ve always thought the Strava estimate was ridiculously low on the averages even before I got the PM.

    It’s not a dealbreaker & it’s certainly not ruining my life but I like to see my data on specific segments to try & track improvements over time. I always go off the average & NP on TP for the total ride anyway.

    If the segments are pretty accurate then that’s good enough. I think you can narrow it down to a specific section / segment on TP but only on the premium version
    Is your auto pause on ?
    I know it can mess with average speed ,not sure if it does the same with power