Ive test it 1month ago. It estimate me at 60watt average when i did 160... But was on a gravel bike on gravel road.. they are probably better with road bike on asphalt road
I live in Amsterdam where you have flat windy roads all the time. Due to the wind the readings make no sense. Like 180 watt with a headwind while it's more like 280 watt in reality.
hornetluca that's because strava doesn't plot a gradual steady state of estimating power. It flicks along second to second spiking up in excess of 1000watts and dropping down to 20watts for example. It'll then guess your wattage for a given segment.
@@HeathyRoidz not hard..but it doesnt mean you can average 800-900 watts a minute. Anyone that squats more than 180 lbs without trouble can do 2000 watts for a second.
sarcasm aside that would means it is time to go back to the mothership. your slope may need recalibrating or the unit may need to be replaced if there is a issue with the strain gages.
To improve the comparison, you should take the garmin file and strip out the power channel, and upload that to strava as well. That way, you will eliminate any difference in GPS data between the two recording devices.
Power meters don't rely on GPS data, but strava's estimated power does. (Well, it might rely on elevation data and not GPS, but that could differ between the phone and garmin as well). Given something like www.fitfiletools.com/ can strip out the power channel easily, it would have been just as easy (for post-analysis) to just record on the garmin, and upload the file with and without power to compare.
Oh, I see what you mean. Basically, it would be easier to sync the two graphs for someone who wants to download the raw data and do their own comparisons.
That, plus if the elevation for the phone (which is from some satellite-image measurements, not a barometric altimeter, for my phone at least) is wrong, that will throw off the calculations. And in my experience, elevation data on strava is not very reliable.
Good video. Here is another one I'd like to see. Using the same power (like try to keep it at 100W for the entire test) try different tire pressures and then compare either your speed through the segment or how far you went in a given time. Do it on bumpy, smooth, cobblestones, etc.
Strava estimates power based on many factors some more accurate than others. I'll give a few examples of where small errors creep in and compound into large errors. 1. GPS speed. Speed is a major variable in power estimation (power is force over time). GPS is intentionally inaccurate for civilian use (4 meters RMS) so one GPS reading might be a couple meters behind you and the next a couple meters in front of you. Such situation makes it appear that you traveled further than reality thus causing weird and inaccurate fluctuations in speed. Averaging these errors over a long ride and they'll generally cancel out but for short efforts or in poor GPS it's bad news. 2. Mapping data. The other big variable is grade of the road and the subtle curves of the road. No mapping data is perfect and Strava's maps have poor elevation readings, especially in areas with lots of trees cover. There are a few segments near me that exceed 35% according to Strava but theyre more like 15% max. Again over longer rides these errors average out but greatly affect short efforts. 3. Accurate rider and bike weights. You'd have to be extremely diligent weighing yourself in full kit and bike will full bottles and such to avoid these errors. And even your scale would need to be calibrated well. These errors are unavoidable and won't average out, but they are probably smaller than the other factors. 3. All the other stuff! As others have already notes there are lots of little watts being fudged. Specifically wind resistance and friction. Rider position, aero bikes, helmet and kit selection, even tire width or a gentle breeze affect resistance. Adding to that a worn chain or bearings, rolling resistance of the tires, etc and you can see why just measuring power is so much easier. Finally as someone who recently started using a power meter and previous enjoyed looking at Strava power, I think more important that accuracy (or even consistency) is being able to see power real time. Without being able to look down at your current power you'll never get the benefits of riding with power. And that's not something you can do with Strava power.
Good summary comment. I like to add a few issues as well. 1. GPS: Strava app works with 3 sec. recording. Means you lose accuracy just from the frequency of the amount of data. Garmin has also different settings for the recording intervals. It's highly recommended to NOT use smart-recording. You will end up having greater files, but more data means more accuracy. Furthermore, you should have your phone also not in your back pocket, handlebar mount will have for sure a better GPS signal than in your pockets. 3. Good point. Bike weight and the type of bike setting in Strava is used for the calculations as well. So it isn't enough to properly weight urself. Strava does not take into account wind at all. At least as far as I know. I have no power meter yet and I probably won't get some before next year. So I am using Golden Cheetah for the power calculations, they take the wind and it's direction into account, but the data is still off with massive errors. Especially with a massive headwind and when the route gets these crazy wind as a crosswind, the numbers come out of the calculations much too high. The wind had not such great impact of my overall speed, also the density of forest or houses along the road influence wind speed massively. Also, changes of the wind are very difficult to monitor if you ride 3 hours and the wind turns or massively changes in strength. I rent a power meter for 2 weeks to check my real numbers. So I tune the results regarding my feelings more or less. In the end, calculated Power will never be able to monitor minor or even big improvements or gains at all. It's important to be aware of that. It's just a guessed value. The longer the ride the closer the numbers get. But you can't trust these numbers or do structured training based on these. It is simply not possible. Generally, the Aerobic Power with calculated power is quite higher than measured. For the short Anaerobic Power, it is the quite the opposite: measured power is higher, sometimes even much higher, then calculated.
EagleEye_onBike I have my garmin set to one second recording even when not using my power meter and still the strava estimate was way off. Using a power meter automatically sets gamins to one second recording. People trying to avoid buying one are kidding themselves.
While multiple reviewers have demonstrated the Powertap P1 pedals can be consistent, one common finding is that they usually underestimate power. If that's the case, there would be an even more significant variation between the Strava estimate and the "true" power.
I was referring to the average power, not the sprints, which I believe Strava underreports. If the P1 pedals do under-measure power, that would mean his average power would have in reality been higher than the P1 Pedals report. In that instance, the gap between Strava and the true power generated would have been greater.
Dude, much respect. There is NOBODY doing this ish on the UA-cam. For my own opinion, Strava doesn’t factor in wind direction, rolling resist from extreme temps, or your position on the bike(if you sit upright, it thinks you are pushing less). It’s pretty good for avg on an out and back, but it isn’t trainable interval data.
Cool and interesting comparison. Ive always wondered how similar the results would be. Obviously quite a few variables like wind etc. but its still pretty cool that its reasonably close. Ive been using the StravaStix Chrome plugin that adds a ton of extra info into Strava for you to mull over. Interestingly it generates a "Weighted Power average" It would be interesting to see how that average compares to your PowerTap "weighted average" Vids like this are awesome. Keep them coming :)
Hi, just saw your video - congrats, very interesting and informative. What struck me was the amount of 'spikes' in the Strava calculated data. I frequently process data on the job, and the first step is always checking and cleaning the data. That includes checking the spikes, the possible reasons for the spikes and the elimination of the spikes when they are clearly out of realistic range. This may bring the Strava and power meter data closer to one another. Anyhow, congrats on the interesting and well-made movie & keep up the good work!
The estimated power can be good in certain cases and certain use cases but if power could really be estimated all the time, power meters manufacturers would be put of business. Last I checked, there are more on the market than ever! Disclaimer: I work for Strava but these words are my own.
For a scruffy looking dude who might be mistaken for homeless, data presentation is world class. No ambiguous boring drawn out snoozefest charts... this data preso is crisp clear spicy front and center. Thumbs up all day.
Thanks for the study. It's very useful. If you want, you could also see the difference between Strava android app power, against imported rides from Garmin devices, such as Edge 1000 and real power. Everytime I use android app, the power shown is about 50% more than the one calculated with Garmin sincronization for same ride.
To base your training on power isn’t only about measuring absolute levels of power. Rather, it’s also about measuring/identifying trends in power (relative levels). Hence, whatever gives me the most consistent(!) readings is the best!
Thanks for the vid and analysis, V. Another way Strava and all power estimators fail is when pack riding. I've also done some comparisons, and any time you are in a draft, estimators will read high, often by 25-50%.
i did the same thing but with a PowerTap hub and a PowerCal HR strap. the results for 30sec average and overall average were always reasonably close. so much so i still use the PowerCal as a backup to my Quarq. this stuff is getting better and better year.
The thing is that training with power only make senses when you have the info live while riding. To have an estimation of your power after the ride is just a gimmic
Good video dude.... I do not have the money to buy a powermeter, I guessed that Strava would give me a good overall view on my performance change over a longer time and your video confirmed it. I added a heart rate monitor to add a measurable variable to the Strava algorithm. Today I did a 123 km ride and tx to yur video I trust what Strava gives me as average power. I still need to calculate my FTP, with Strava and your wisdom, I learned that I will have to take the avg from different sessions. Cheers
Don't trust strava power on flat. Get to know your body (with help if HR monitor) to learn the intensity that you can hold for one hour. 20 mins intervals can teach you a lot about that. Then pick an even uphill (5-8% is perfect) section and ride that at one-hour pace (monitored by HR) on a calm day. That'll be a very good estimate of ftp.
Just a small commentary on the power meter, As i'd assume they'd be using and analogue to digital converter, so the the resolution will purely depend on the sampling rate. That means if the meters doesnt sample fast enough your peak will not get detected. However we are dealing with slow speeds, this means an Arduino's ADC is more than adequate to convert and not miss out.
I compared Strava power to my Power Tap back in 2013 on several occasions. Strava consistently read 15 to 25 watts below my Power Tap average. Since the only Strava variable I can control is weight. I found I could get fairly close by adding the weight of my bike, water and gear ( helmet, spare tubes and such, kit and shoes) into the mix. I eventually dumped the Power Tap hub and just went on Strava power. I wouldn't use it to train by but I does give a rough idea of energy/effort expenditure.
Such an awesome video bro, been wondering about this for a loooong time. Cycle_dr1 had already talked about this and in his case Strava really underestimated his avg power which I think is what happens to me (obviously ahah). The max power is really weird indeed, I have 1000+ values when going downhill without pedalling... One really weird thing is that on really steady rides my Strava power is still all over the place even if the gradient, speed and cadence are constant. I guess I'll have to save up for a power meter.. Great video non the less, keep it up!
Strava's power calculations are based on GPS data that is not 100% true, if you have a super good GPS that updates your position more frequently and faster then the spikes would go away. I am not sure if they have an expected aerodynamic calculation but a strong headwind/tailwind will defiantly yield unfavorable results too. I use Strava power with a grain of salt; I ride the same routes (a lot!), I am aware that a lot of factors are not in Strava's calculations, and I am only using the numbers to gauge my own progress -not bragging rights.
It would be interesting to see what the comparison is when you do (as you suggested) a standard ride without all the spikes and dips. Great video, very informative.
To sum it up, one could use Strava average power or W/kg over a long period of time to track progress. Spikes in max power comes from inaccurate readings on the gps, where speed suddenly jumps or false altitude gains are registered. But over a long period of time, cycling around roughly the same routes, it should give a good estimate of progress.
To measure power without a strain meter, you need five bits of data: speed, acceleration, weight, topography and wind. Strava has access to the first four, but can't get real-time wind data. As a result, it overestimates your power with a tailwind, and underestimates when you ride into a headwind. A relatively new device called Powerpod uses the five factor calculation method; it has a tiny wind gauge built in. Otherwise, I'm pretty sure the only way to measure accurate power is with a strain gauge in your crank, hub or pedals.
I made my own power calculator in Microsoft exel just because I was quite interested to know my usual power when cycling. This was before I even realised you can even get such a thing as a power meter on a bike. It basically involves running the Strava during a cycle and then at certain points during the ride you just stop peddling for a number of seconds. Later once at home you put the info from the Strava into the calculator and it works out the power at that particular moment. The only thing is it's probably an underestimate since it's being measured when you are not peddling, therefore doesn't account for chain/ peddles bearings friction. It would be interesting to know how my calculator method compares to an actual power meter (unfortunately I don't have one on my bike).
Depends a bit on the road surfaces too. Where I live, roads are pretty rough, hence a mellow ride on flat roads at ~160 watts registers on Strava as barely over 100 watts. However, on a smooth surface where I saw 247 watts up a gradient, Strava had 237 for the same segment. For training, accuracy isn't as important as consistency. As you point out, Strava's estimates simply aren't very consistent, hence not a good reference for training.
This has been covered before. Any segment with a tailwind will overestimate Strava wattage by a lot. Only works (barely) for steady climbs with next to no wind and you better not lose that GPS signal since speed is a variable used in the calculation.
I think the most important thing in training with power is the principle of "apples to apples." My Kinetic trainer at home measures power at the wheels and reads 4.5% below a meter reading power at the crank and 2.5% below the wattbike at the gym. As long as you know your zones for the readout you're looking at, you can use it effectively.
Wind REALLY impacts Strava Power. Living in a windy area, it's very common to have someone with a segment time very close to my own but have an estimated power a couple of hundred watts above what my power meter said I did.
Great Content. I wanted to say however, you must also consider they are calculating two different things. Strava is calculating the power transferred to the ground by the bike (at the wheels), while your power meter is calculating power output at the pedals. While i'm not claiming that Strava's calculation is accurate, it will take into account the impact of factors that are not considered on your power meter, such as mechanical advantage/velocity ratio due to the drivetrain. Power readings will always vary to some extent, even between different power meters. The key thing with power measurement is to use the same method of measurement and only use it as a benchmark for measuring progress, not actual power.
Really great video, another interesting one i'd like to see is that there is this google chrome extension called StravistiX, and it pulls precise weather condition data (such as wind, temperature, humidty etc;) and factors that into the equation too. It gives quite a different estimate than the strava one, but i'd be interested to see if it was more accurate if your riding was more variable. As for me, i'm looking to get a power meter in the future, but for the time being, I just use strava power for comparing similar segments (or the same segment), though given that i've been loosing weight, my power estimates on the flats are going down, despite me both feeling more powerful and being practically faster now. I tend to find it seems quite consistent for hills however.
I'm guessing that variability of geospatial data is going to be a key component in Strava's calculations. If you are going up or down small slope changes that strava does't know about, it can't accurately calculate your power. Likewise if Strava's surface data has a bunch of noise (fake micro-hills) it will also produce errors. The good news about random error, it usually averages out over a long enough period.
Weight seems to make a big difference on Strava, I had put my weight and bike weight in but forgot to add in weight for clothing, shoes, helmet, computer and drink/food which amounts to more than you think!
From my observation, strava can be off by so many factors. 1. ride in group 2. wind 3. up and downhill 4. Indoor 5. so many other issues which I don't even know. I train with power so I use Zwift & Trainerroad for indoor workous, and outdoor I my memorize my workouts, analyze in Trainingpeak post-workout. For me Strava is a social app. I use it to observe other cyclists' activities, power output on certain rides.
An issue I have with Strava power is, the numbers are not in real time. When the ride is over, that's when it calculates the power numbers. With a meter you know in real time if you need to put in more effort or less effort.
Strava uses the data metrics in an equation, power meters measure strain. Adding the bikes weight to your weight should make uphill power more accurate
Good job bruh! Next step would be to install the StravastiX for Strava browser plug-in and comparing that estimated power to the power meter. I think it uses another algorithm so it might be more/less accurate.
if you do a uphill segment and record it with your garmin + with your phone, usually the elevation is off by 10% that screws up the whole strava algorythm.
I work for a company that makes high accuracy GPS receivers and have used one to feed into my Strava feed on my phone. The course is much more precise. Most people do not realise that gps reception on a phone is pretty poor, and VC where you ride, under a tree canopy in mountainous country, is really asking a lot of a phone. The power spikes seem to come from interrupted reception, I have found some pretty silly power spikes due to imaginary bursts of speed and my terrain is less demanding than yours. I plan to continue beta testing the high accuracy device on all my rides and compare previous sectors done without. I don't know if Strava uses the phone's motion sensors to smooth the speed but if they do it does not seem very effective.
My mountain bike in the road vs my road bike in the road gives me an average power of about 100-120w more. An hour long ride on my mountain bike usually gives me about 280-300w on strava whereas the same route on my roadbike at the same RPE would give me about 160-180w
Probably, using Strava power meter it is important that results would be the same for the same rides with low deviations. If so, you can then trust it and see some reasonable dynamics comparing to your previous rides.
If you're interested to one-up this video, would you examine the difference between 1-sided and dual-sided power meters? How inaccurate is that other leg making things?
Surely the valid test would be to ride with your power meter, then take the file from your Garmin and strip the power data from it and upload both the original & stripped to Strava - then you're using exactly the same input data streams (ie. No variability due to the phone & Garmin sampling differently)
Nice Vid. Would be interesting to check, how big the difference is, if you'd give cadence to strava to chew on and not only speed. Additionally it would be worth while to check on the difference when using the Stravistix Add-On.
Long ride (3-6 hrs) delta (average) appears to 10-15% high on Strava based on my own anecdotal experience. Hilly ride, fair amount of chipseal, and fairly windy conditions. Interval power isn't even close, especially sprint power. If you have a tailwind on a short segment, Strava will over estimate power by a considerable amount.
Hey- so i know this content was from 5 year ago but hey-some of us are still catching up... Curious to know how the Peloton bike compares to road power output -is that something you'd consider looking at? Also using a trainer with a road bike Vs peloton bike-how do they compare in general.... Anyway if it ever happens great- and thanks for all the amazing video content- it just gets better and better.
Hi @The Vegan Cyclist , nice comparison, but, have you paired cadence sensor with the phone? I think cadence is a very important variable that Satrava takes into account for more precise power estimates... no cadence sensor means that Strava power estimate will be poor... maybe a new comparison? I would do it myself, but i don't have a powermeter to compare :\
Forgot which on my strava-employee friends told me that once you use a powermeter strava uses that data for the power estimate, would have been curious to see if it was as similar if you had a blank account. Geeky video
So I learned that Strava discards spikes, if theres a spike that is not sustained (IE you hit a sprint and then hit 55 for a second and get off the gas right away strava will discard that value, which might be the difference in some of those calculations). Strava seems to lose accuracy as your ride gets longer (headwinds, tailwinds etc)....
just one last piece of advice. Power numbers or wattage is just one metric, you must also learn to read your heart rate even though they say that it lags behind or gives variable readings for some reason or another. What I am saying is that keep an eye on the heart rate when doing short hard efforts, and if you notice that the heart rate is not going down in between hard efforts then it is time to call it a day. Otherwise, you will fall into a cycle of sloppy riding form, and your training effort will be compromised and you might even invite injury, overtraining, and god forbid illness.
It would be kind of interesting to hear from Strava themselves if they think it's their calculation or the GPS signal that is the biggest contributor to these variations. Because if you look att your individual points of location from a ride, calculation power from that must be hard to say the least.
Have you done a video on the bikes you ride and when? It be good to break down drive train, number of teeth, ratios, tire volume, etc. for different rides. I’m currently looking to buy a bike so that might be useful to anyone similar. I’m thinking about getting one bike two wheel sets. Also, trying to figure out how to book you as a guide for Shaver Lake, FYI. Keep at it! Digging the content. Yuh!
Yeah, the wind would probably have a big effect on power. Strava probably gives you the same power for a segment completed at the same time every time. Which of course would make a big difference on power if you're going with a headwind or a tailwind 👍
Ride loops, especially on windy days. Out and back routes are good too, especially if the route coincides with the wind directions. With quartering or swirling winds, larger loops work better. Both negate the influence of headwinds and tailwinds. Strava power estimates are reasonably accurate then. Any 5-10 mile or larger loop or out and back route is good. Doesn't matter whether it's flat or hilly, it evens out. Otherwise, nope. Especially shorter, unidirectional segments. Strava will exaggerate power with tailwind, and ignore real effort into headwinds. Or if wind conditions change significantly during a ride Strava estimates may be off. But include both head/tail wind effects on a loop or out and back route and it evens out.
I use mywindsock.com average power which links with Strava because it calculates your power based on wind speeds as well I believe. Its good to see your power after rides for kilo/watts and to make sure your progressing I think. Far better than going off of speed.
Another important thing to me would be how repeatable are the Strava figures? If you did the same test tomorrow, with the same power from yourself, would the Strava figures be similar to today?
Vegan Cyclist, Alter your weight in settings until you get close to matching values in a mixed ride. The kind of rides that one ordinarily does. Rides with all the variables as in hills, acceleration and steady speed. I did it years ago with my Power Tap and I had to add my weight, bike weight and a little extra to get them to come close. : )
I'm not on Strava but it would be interesting to do this comparison. My max power on a typical ride is usually around 625 watts, Strava would probably turn me into Peter Sagan on a sprint lol.
Strava is a nice tool and we all agree about this. But the point is: if Strava is so inaccurate in (actual) elevation measurement - which is always far more than the real one - how can it be reliable in estimating a power output ?
It kinda work only for the solo efforts and those without drafts. Strava will calculate a peloton effort just as it was a solo one which will be way too off. And lots of other cases as well like the bike and kit aero profile etc....
A few variables like wind and gearing Strava doesn't take into account. Those 2 things could be a pretty big determining factor. My experience is Strava power tends to be way off in most cases.
You touched on something I was researching.. Downhill. Trying to find out: if you don't pedal downhill, how does this affect overall ride avg power (Strava), compared with spinning over at high cadence (at similar speed)? E.g if you have 0 cadence going downhill, does that affect your overall ride avg power output? Is it better for overall avg power output to spin at high cadence downhill or coast? Thanks!
The power is good read from your garmin, as soon as strava uploads my power is way off,I am not talking about estimated power,I asked why its because the segments apparently, VC YER.
+The Vegan Cyclist - Thanks for doing such comparison. How is the Powertap P1 holding off so far? I am in the market to get a P1s due to budget constraints. If possible could you do a comparison between a P1 and a P1s? Thanks!
Did you ensure that the phone recorded the same speeds as your head unit did? The gps reporting may have lags and therefore will report exorbitantly fast time between subsequent waypoints, hence, 1000+ watt peaks. Try 2 speed sensors: 1 for no power, the other with power.
Hey bro really appreciate this experiment I was looking for this for a while (don't have a power meter yet, but I'm already looking for one) I wonder how much wind did you face? We're i live there are freaking crosswinds that Puch you out of the road and always believe that this windy situation ruin the accuracy of Strava power
algorithm depends on EXACT weight of your bike in riding condition (w water, tools, spare, etc) and the EXACT WEIGHT of your body with shoes, clothes, helmet, etc. if those #s are accurate, the avg power on Strava should be very accurate (and could also be calculated on the fly for like 10 sec or 30 sec avg)
there s gps error to take in . Usualy for a phone, the gps data is preety buggy. Try and use your garmin without power meter and compare that data to your phone, you will see the diff. Also the max power is higher because of the lag of the data comming in. Gps devices record data in bucket data structs.Depending on those bucket size your gpx file is more or less accurate. If the bucket is big you will get a huge lag of your surge which translates in a huge power. There s nothing wrong with the Strava algorithm, it s your recording device thats prone to errors.
What about riding into a strong headwind? I'd have to say that it's probably grossly wrong. I've had some rides where I was going into the red and strava said I was doing 150 watts, I'd have to say I was doing 250-300. I don't currently have a power meter hooked up...suspect I need one.
I would love to see an updated test to see if Strava power has gotten any better, especially with all the updates Strava has made recently.
Yes please
+1 on this
THIS! Makes sense, power estimates should be better, but I guess those power bursts/peaks are the same, at least on my rides there are always some
Ive test it 1month ago. It estimate me at 60watt average when i did 160... But was on a gravel bike on gravel road.. they are probably better with road bike on asphalt road
@@cricrimc thanks for the info
A windy day would also affect the accuracy of a steady state ride.
Henrywildeberry Yes I get high wattages while riding with a tail wind in Strava and vice versa for head wind.
It would be slick if Strava could use the weather as a variable in determining power.
Its way off for me . Strava has no clue. Buy a power meter.
Yes the weather looks pretty calm there. I get big differences in tail and headwinds even though my heart rate is the same.
I live in Amsterdam where you have flat windy roads all the time. Due to the wind the readings make no sense. Like 180 watt with a headwind while it's more like 280 watt in reality.
Those 1200 watts are always in any of my Strava rides.
hornetluca that's because strava doesn't plot a gradual steady state of estimating power. It flicks along second to second spiking up in excess of 1000watts and dropping down to 20watts for example. It'll then guess your wattage for a given segment.
My power meter says 1200 watts cos I'm a beast.
1200 is not a big number. Sprinters can do 2000 watts.
@@felixg4785 can you do 2000W?
@@HeathyRoidz not hard..but it doesnt mean you can average 800-900 watts a minute. Anyone that squats more than 180 lbs without trouble can do 2000 watts for a second.
I threw mine away. It was always reading low. Strava power for life.
Must of been a Stages one sided
SilentGTboy ya it is. I was just kidding. Like blaming my low power on the power meter. It's hard to type sarcasm.
Yer, my powertap consistently reads about 40% under, I just add it on in my head ; )
sarcasm aside that would means it is time to go back to the mothership. your slope may need recalibrating or the unit may need to be replaced if there is a issue with the strain gages.
Its way off for me . Strava has no clue. Buy a power meter.
You make the coolest videos about cycling, especially for beginners who are trying to figure all of this out. Keep em coming!!!
To improve the comparison, you should take the garmin file and strip out the power channel, and upload that to strava as well. That way, you will eliminate any difference in GPS data between the two recording devices.
Walter Somerville fuck that
What's the point? Power meters don't rely on GPS data at all.
Power meters don't rely on GPS data, but strava's estimated power does. (Well, it might rely on elevation data and not GPS, but that could differ between the phone and garmin as well). Given something like www.fitfiletools.com/ can strip out the power channel easily, it would have been just as easy (for post-analysis) to just record on the garmin, and upload the file with and without power to compare.
Oh, I see what you mean. Basically, it would be easier to sync the two graphs for someone who wants to download the raw data and do their own comparisons.
That, plus if the elevation for the phone (which is from some satellite-image measurements, not a barometric altimeter, for my phone at least) is wrong, that will throw off the calculations. And in my experience, elevation data on strava is not very reliable.
Good video. Here is another one I'd like to see. Using the same power (like try to keep it at 100W for the entire test) try different tire pressures and then compare either your speed through the segment or how far you went in a given time. Do it on bumpy, smooth, cobblestones, etc.
Drafting in a fast peleton, Strava will give you a really wrong reading!
Strava estimates power based on many factors some more accurate than others. I'll give a few examples of where small errors creep in and compound into large errors.
1. GPS speed. Speed is a major variable in power estimation (power is force over time). GPS is intentionally inaccurate for civilian use (4 meters RMS) so one GPS reading might be a couple meters behind you and the next a couple meters in front of you. Such situation makes it appear that you traveled further than reality thus causing weird and inaccurate fluctuations in speed. Averaging these errors over a long ride and they'll generally cancel out but for short efforts or in poor GPS it's bad news.
2. Mapping data. The other big variable is grade of the road and the subtle curves of the road. No mapping data is perfect and Strava's maps have poor elevation readings, especially in areas with lots of trees cover. There are a few segments near me that exceed 35% according to Strava but theyre more like 15% max. Again over longer rides these errors average out but greatly affect short efforts.
3. Accurate rider and bike weights. You'd have to be extremely diligent weighing yourself in full kit and bike will full bottles and such to avoid these errors. And even your scale would need to be calibrated well. These errors are unavoidable and won't average out, but they are probably smaller than the other factors.
3. All the other stuff! As others have already notes there are lots of little watts being fudged. Specifically wind resistance and friction. Rider position, aero bikes, helmet and kit selection, even tire width or a gentle breeze affect resistance. Adding to that a worn chain or bearings, rolling resistance of the tires, etc and you can see why just measuring power is so much easier.
Finally as someone who recently started using a power meter and previous enjoyed looking at Strava power, I think more important that accuracy (or even consistency) is being able to see power real time. Without being able to look down at your current power you'll never get the benefits of riding with power. And that's not something you can do with Strava power.
Joshua Piccari
power = force x velocity = energy/time
Good summary comment. I like to add a few issues as well.
1. GPS:
Strava app works with 3 sec. recording. Means you lose accuracy just from the frequency of the amount of data. Garmin has also different settings for the recording intervals. It's highly recommended to NOT use smart-recording. You will end up having greater files, but more data means more accuracy.
Furthermore, you should have your phone also not in your back pocket, handlebar mount will have for sure a better GPS signal than in your pockets.
3. Good point. Bike weight and the type of bike setting in Strava is used for the calculations as well. So it isn't enough to properly weight urself.
Strava does not take into account wind at all. At least as far as I know.
I have no power meter yet and I probably won't get some before next year. So I am using Golden Cheetah for the power calculations, they take the wind and it's direction into account, but the data is still off with massive errors. Especially with a massive headwind and when the route gets these crazy wind as a crosswind, the numbers come out of the calculations much too high. The wind had not such great impact of my overall speed, also the density of forest or houses along the road influence wind speed massively. Also, changes of the wind are very difficult to monitor if you ride 3 hours and the wind turns or massively changes in strength.
I rent a power meter for 2 weeks to check my real numbers. So I tune the results regarding my feelings more or less.
In the end, calculated Power will never be able to monitor minor or even big improvements or gains at all. It's important to be aware of that. It's just a guessed value. The longer the ride the closer the numbers get. But you can't trust these numbers or do structured training based on these. It is simply not possible.
Generally, the Aerobic Power with calculated power is quite higher than measured. For the short Anaerobic Power, it is the quite the opposite: measured power is higher, sometimes even much higher, then calculated.
EagleEye_onBike I have my garmin set to one second recording even when not using my power meter and still the strava estimate was way off. Using a power meter automatically sets gamins to one second recording. People trying to avoid buying one are kidding themselves.
experiment54 People who are trying to avoid buying power meters can't afford them.
Harrison Devine Everyone can save up. I know its a foreign concept in todays age of gimme gimme now. 4ii and stages are cheap.
While multiple reviewers have demonstrated the Powertap P1 pedals can be consistent, one common finding is that they usually underestimate power. If that's the case, there would be an even more significant variation between the Strava estimate and the "true" power.
I was referring to the average power, not the sprints, which I believe Strava underreports. If the P1 pedals do under-measure power, that would mean his average power would have in reality been higher than the P1 Pedals report. In that instance, the gap between Strava and the true power generated would have been greater.
Dude, much respect. There is NOBODY doing this ish on the UA-cam. For my own opinion, Strava doesn’t factor in wind direction, rolling resist from extreme temps, or your position on the bike(if you sit upright, it thinks you are pushing less). It’s pretty good for avg on an out and back, but it isn’t trainable interval data.
Cool and interesting comparison. Ive always wondered how similar the results would be. Obviously quite a few variables like wind etc. but its still pretty cool that its reasonably close. Ive been using the StravaStix Chrome plugin that adds a ton of extra info into Strava for you to mull over. Interestingly it generates a "Weighted Power average" It would be interesting to see how that average compares to your PowerTap "weighted average"
Vids like this are awesome. Keep them coming :)
Hi, just saw your video - congrats, very interesting and informative. What struck me was the amount of 'spikes' in the Strava calculated data. I frequently process data on the job, and the first step is always checking and cleaning the data. That includes checking the spikes, the possible reasons for the spikes and the elimination of the spikes when they are clearly out of realistic range. This may bring the Strava and power meter data closer to one another. Anyhow, congrats on the interesting and well-made movie & keep up the good work!
make sure you put in bike weight for strava, that's important too ;)
The estimated power can be good in certain cases and certain use cases but if power could really be estimated all the time, power meters manufacturers would be put of business. Last I checked, there are more on the market than ever!
Disclaimer: I work for Strava but these words are my own.
If you work for Strava can you get me some KOMs please.
The only thing Strava power is good for is convincing cyclists "F**k this shit, I need a real power meter."
For a scruffy looking dude who might be mistaken for homeless, data presentation is world class. No ambiguous boring drawn out snoozefest charts... this data preso is crisp clear spicy front and center. Thumbs up all day.
Thanks for the study. It's very useful. If you want, you could also see the difference between Strava android app power, against imported rides from Garmin devices, such as Edge 1000 and real power. Everytime I use android app, the power shown is about 50% more than the one calculated with Garmin sincronization for same ride.
To base your training on power isn’t only about measuring absolute levels of power. Rather, it’s also about measuring/identifying trends in power (relative levels). Hence, whatever gives me the most consistent(!) readings is the best!
Thanks for the vid and analysis, V. Another way Strava and all power estimators fail is when pack riding. I've also done some comparisons, and any time you are in a draft, estimators will read high, often by 25-50%.
i did the same thing but with a PowerTap hub and a PowerCal HR strap. the results for 30sec average and overall average were always reasonably close. so much so i still use the PowerCal as a backup to my Quarq. this stuff is getting better and better year.
The thing is that training with power only make senses when you have
the info live while riding. To have an estimation of your power after the ride is just a gimmic
+roland H totally, but like I said if someone wanted to check watt to kilo, it's not nearly as far off add I thought it would be.
The Vegan Cyclist True, can give someone just starting to train an Idea where he is powerwise 👍
You could use it as a very rough guide for tss. Only real rough.
i think there is a connect app that estimates power for you. probably pretty closer results to starva.
Its way off for me . Strava has no clue. Buy a power meter.
Good video dude.... I do not have the money to buy a powermeter, I guessed that Strava would give me a good overall view on my performance change over a longer time and your video confirmed it. I added a heart rate monitor to add a measurable variable to the Strava algorithm. Today I did a 123 km ride and tx to yur video I trust what Strava gives me as average power. I still need to calculate my FTP, with Strava and your wisdom, I learned that I will have to take the avg from different sessions. Cheers
Don't trust strava power on flat. Get to know your body (with help if HR monitor) to learn the intensity that you can hold for one hour. 20 mins intervals can teach you a lot about that. Then pick an even uphill (5-8% is perfect) section and ride that at one-hour pace (monitored by HR) on a calm day. That'll be a very good estimate of ftp.
TX...
Just a small commentary on the power meter, As i'd assume they'd be using and analogue to digital converter, so the the resolution will purely depend on the sampling rate. That means if the meters doesnt sample fast enough your peak will not get detected. However we are dealing with slow speeds, this means an Arduino's ADC is more than adequate to convert and not miss out.
I compared Strava power to my Power Tap back in 2013 on several occasions. Strava consistently read 15 to 25 watts below my Power Tap average. Since the only Strava variable I can control is weight. I found I could get fairly close by adding the weight of my bike, water and gear ( helmet, spare tubes and such, kit and shoes) into the mix. I eventually dumped the Power Tap hub and just went on Strava power. I wouldn't use it to train by but I does give a rough idea of energy/effort expenditure.
This video needs to be done again as it’s 3 years on and the Strava algorithm must’ve improved by now.
I doubt - ua-cam.com/video/xfnRz4u4Cgc/v-deo.html
Such an awesome video bro, been wondering about this for a loooong time. Cycle_dr1 had already talked about this and in his case Strava really underestimated his avg power which I think is what happens to me (obviously ahah). The max power is really weird indeed, I have 1000+ values when going downhill without pedalling... One really weird thing is that on really steady rides my Strava power is still all over the place even if the gradient, speed and cadence are constant. I guess I'll have to save up for a power meter.. Great video non the less, keep it up!
Ok I am completely sold on the power now. I will grab them next week.
Strava's power calculations are based on GPS data that is not 100% true, if you have a super good GPS that updates your position more frequently and faster then the spikes would go away. I am not sure if they have an expected aerodynamic calculation but a strong headwind/tailwind will defiantly yield unfavorable results too. I use Strava power with a grain of salt; I ride the same routes (a lot!), I am aware that a lot of factors are not in Strava's calculations, and I am only using the numbers to gauge my own progress -not bragging rights.
It would be interesting to see what the comparison is when you do (as you suggested) a standard ride without all the spikes and dips.
Great video, very informative.
To sum it up, one could use Strava average power or W/kg over a long period of time to track progress.
Spikes in max power comes from inaccurate readings on the gps, where speed suddenly jumps or false altitude gains are registered. But over a long period of time, cycling around roughly the same routes, it should give a good estimate of progress.
To measure power without a strain meter, you need five bits of data: speed, acceleration, weight, topography and wind. Strava has access to the first four, but can't get real-time wind data. As a result, it overestimates your power with a tailwind, and underestimates when you ride into a headwind. A relatively new device called Powerpod uses the five factor calculation method; it has a tiny wind gauge built in. Otherwise, I'm pretty sure the only way to measure accurate power is with a strain gauge in your crank, hub or pedals.
your channel is cool real useful and super chill!
I made my own power calculator in Microsoft exel just because I was quite interested to know my usual power when cycling. This was before I even realised you can even get such a thing as a power meter on a bike. It basically involves running the Strava during a cycle and then at certain points during the ride you just stop peddling for a number of seconds. Later once at home you put the info from the Strava into the calculator and it works out the power at that particular moment.
The only thing is it's probably an underestimate since it's being measured when you are not peddling, therefore doesn't account for chain/ peddles bearings friction.
It would be interesting to know how my calculator method compares to an actual power meter (unfortunately I don't have one on my bike).
Depends a bit on the road surfaces too. Where I live, roads are pretty rough, hence a mellow ride on flat roads at ~160 watts registers on Strava as barely over 100 watts. However, on a smooth surface where I saw 247 watts up a gradient, Strava had 237 for the same segment.
For training, accuracy isn't as important as consistency. As you point out, Strava's estimates simply aren't very consistent, hence not a good reference for training.
This has been covered before. Any segment with a tailwind will overestimate Strava wattage by a lot. Only works (barely) for steady climbs with next to no wind and you better not lose that GPS signal since speed is a variable used in the calculation.
I think the most important thing in training with power is the principle of "apples to apples." My Kinetic trainer at home measures power at the wheels and reads 4.5% below a meter reading power at the crank and 2.5% below the wattbike at the gym. As long as you know your zones for the readout you're looking at, you can use it effectively.
I have a sigma unit which uses a algorithm to calculate so if the number is constant an u just use tht numbers your working with something
Head or tail winds trash the Strava figures even on steady state rides.
Wind REALLY impacts Strava Power. Living in a windy area, it's very common to have someone with a segment time very close to my own but have an estimated power a couple of hundred watts above what my power meter said I did.
Cool video mate. love it.
Great Content. I wanted to say however, you must also consider they are calculating two different things. Strava is calculating the power transferred to the ground by the bike (at the wheels), while your power meter is calculating power output at the pedals. While i'm not claiming that Strava's calculation is accurate, it will take into account the impact of factors that are not considered on your power meter, such as mechanical advantage/velocity ratio due to the drivetrain. Power readings will always vary to some extent, even between different power meters. The key thing with power measurement is to use the same method of measurement and only use it as a benchmark for measuring progress, not actual power.
Really great video, another interesting one i'd like to see is that there is this google chrome extension called StravistiX, and it pulls precise weather condition data (such as wind, temperature, humidty etc;) and factors that into the equation too. It gives quite a different estimate than the strava one, but i'd be interested to see if it was more accurate if your riding was more variable.
As for me, i'm looking to get a power meter in the future, but for the time being, I just use strava power for comparing similar segments (or the same segment), though given that i've been loosing weight, my power estimates on the flats are going down, despite me both feeling more powerful and being practically faster now.
I tend to find it seems quite consistent for hills however.
great video thanks and 'love you " was too cute!
I'm guessing that variability of geospatial data is going to be a key component in Strava's calculations. If you are going up or down small slope changes that strava does't know about, it can't accurately calculate your power. Likewise if Strava's surface data has a bunch of noise (fake micro-hills) it will also produce errors. The good news about random error, it usually averages out over a long enough period.
Weight seems to make a big difference on Strava, I had put my weight and bike weight in but forgot to add in weight for clothing, shoes, helmet, computer and drink/food which amounts to more than you think!
From my observation, strava can be off by so many factors.
1. ride in group
2. wind
3. up and downhill
4. Indoor
5. so many other issues which I don't even know.
I train with power so I use Zwift & Trainerroad for indoor workous, and outdoor I my memorize my workouts, analyze in Trainingpeak post-workout.
For me Strava is a social app. I use it to observe other cyclists' activities, power output on certain rides.
An issue I have with Strava power is, the numbers are not in real time. When the ride is over, that's when it calculates the power numbers. With a meter you know in real time if you need to put in more effort or less effort.
Strava uses the data metrics in an equation, power meters measure strain. Adding the bikes weight to your weight should make uphill power more accurate
Enjoyed the breakdown VC
Very informational and interesting!
Good job bruh! Next step would be to install the StravastiX for Strava browser plug-in and comparing that estimated power to the power meter. I think it uses another algorithm so it might be more/less accurate.
On days when I'm depressed, I just use Strava power 😂🤣
if you do a uphill segment and record it with your garmin + with your phone, usually the elevation is off by 10% that screws up the whole strava algorythm.
Thank you VC, you could also challenge Strava accuracy vs GPS to check gradient and length of climbs.
I work for a company that makes high accuracy GPS receivers and have used one to feed into my Strava feed on my phone. The course is much more precise.
Most people do not realise that gps reception on a phone is pretty poor, and VC where you ride, under a tree canopy in mountainous country, is really asking a lot of a phone. The power spikes seem to come from interrupted reception, I have found some pretty silly power spikes due to imaginary bursts of speed and my terrain is less demanding than yours.
I plan to continue beta testing the high accuracy device on all my rides and compare previous sectors done without.
I don't know if Strava uses the phone's motion sensors to smooth the speed but if they do it does not seem very effective.
Good comparison! Nice.
My mountain bike in the road vs my road bike in the road gives me an average power of about 100-120w more. An hour long ride on my mountain bike usually gives me about 280-300w on strava whereas the same route on my roadbike at the same RPE would give me about 160-180w
Probably, using Strava power meter it is important that results would be the same for the same rides with low deviations. If so, you can then trust it and see some reasonable dynamics comparing to your previous rides.
If you're interested to one-up this video, would you examine the difference between 1-sided and dual-sided power meters? How inaccurate is that other leg making things?
Surely the valid test would be to ride with your power meter, then take the file from your Garmin and strip the power data from it and upload both the original & stripped to Strava - then you're using exactly the same input data streams (ie. No variability due to the phone & Garmin sampling differently)
Nice Vid. Would be interesting to check, how big the difference is, if you'd give cadence to strava to chew on and not only speed. Additionally it would be worth while to check on the difference when using the Stravistix Add-On.
Long ride (3-6 hrs) delta (average) appears to 10-15% high on Strava based on my own anecdotal experience. Hilly ride, fair amount of chipseal, and fairly windy conditions. Interval power isn't even close, especially sprint power. If you have a tailwind on a short segment, Strava will over estimate power by a considerable amount.
Hey- so i know this content was from 5 year ago but hey-some of us are still catching up... Curious to know how the Peloton bike compares to road power output -is that something you'd consider looking at? Also using a trainer with a road bike Vs peloton bike-how do they compare in general.... Anyway if it ever happens great- and thanks for all the amazing video content- it just gets better and better.
Hi @The Vegan Cyclist , nice comparison, but, have you paired cadence sensor with the phone? I think cadence is a very important variable that Satrava takes into account for more precise power estimates... no cadence sensor means that Strava power estimate will be poor... maybe a new comparison? I would do it myself, but i don't have a powermeter to compare :\
really interesting. glad you tested this
Forgot which on my strava-employee friends told me that once you use a powermeter strava uses that data for the power estimate, would have been curious to see if it was as similar if you had a blank account.
Geeky video
+Emile Goguely you know I kinda thought about that. Either way, power needs to be accurate, can't really use the strava stuff for much
So I learned that Strava discards spikes, if theres a spike that is not sustained (IE you hit a sprint and then hit 55 for a second and get off the gas right away strava will discard that value, which might be the difference in some of those calculations). Strava seems to lose accuracy as your ride gets longer (headwinds, tailwinds etc)....
just one last piece of advice. Power numbers or wattage is just one metric, you must also learn to read your heart rate even though they say that it lags behind or gives variable readings for some reason or another. What I am saying is that keep an eye on the heart rate when doing short hard efforts, and if you notice that the heart rate is not going down in between hard efforts then it is time to call it a day. Otherwise, you will fall into a cycle of sloppy riding form, and your training effort will be compromised and you might even invite injury, overtraining, and god forbid illness.
It would be kind of interesting to hear from Strava themselves if they think it's their calculation or the GPS signal that is the biggest contributor to these variations. Because if you look att your individual points of location from a ride, calculation power from that must be hard to say the least.
And once again....I am not even close to this name VEGAN. :)
But i am really enjoying your videos!
Have you done a video on the bikes you ride and when? It be good to break down drive train, number of teeth, ratios, tire volume, etc. for different rides. I’m currently looking to buy a bike so that might be useful to anyone similar. I’m thinking about getting one bike two wheel sets. Also, trying to figure out how to book you as a guide for Shaver Lake, FYI. Keep at it! Digging the content. Yuh!
Great idea for a video. I’ve often wondered.
Yeah, the wind would probably have a big effect on power. Strava probably gives you the same power for a segment completed at the same time every time. Which of course would make a big difference on power if you're going with a headwind or a tailwind 👍
Ride loops, especially on windy days. Out and back routes are good too, especially if the route coincides with the wind directions. With quartering or swirling winds, larger loops work better. Both negate the influence of headwinds and tailwinds. Strava power estimates are reasonably accurate then. Any 5-10 mile or larger loop or out and back route is good. Doesn't matter whether it's flat or hilly, it evens out.
Otherwise, nope. Especially shorter, unidirectional segments. Strava will exaggerate power with tailwind, and ignore real effort into headwinds. Or if wind conditions change significantly during a ride Strava estimates may be off.
But include both head/tail wind effects on a loop or out and back route and it evens out.
I use mywindsock.com average power which links with Strava because it calculates your power based on wind speeds as well I believe. Its good to see your power after rides for kilo/watts and to make sure your progressing I think. Far better than going off of speed.
Another important thing to me would be how repeatable are the Strava figures?
If you did the same test tomorrow, with the same power from yourself, would the Strava figures be similar to today?
Very interesting video! It seems that if Strava smoothed it's estimations, the power would be a lot more accurate - pretty good overall though!
Vegan Cyclist, Alter your weight in settings until you get close to matching values in a mixed ride. The kind of rides that one ordinarily does. Rides with all the variables as in hills, acceleration and steady speed. I did it years ago with my Power Tap and I had to add my weight, bike weight and a little extra to get them to come close. : )
I'm not on Strava but it would be interesting to do this comparison. My max power on a typical ride is usually around 625 watts, Strava would probably turn me into Peter Sagan on a sprint lol.
Strava is a nice tool and we all agree about this. But the point is: if Strava is so inaccurate in (actual) elevation measurement - which is always far more than the real one - how can it be reliable in estimating a power output ?
Thanks for comparing!
It kinda work only for the solo efforts and those without drafts. Strava will calculate a peloton effort just as it was a solo one which will be way too off. And lots of other cases as well like the bike and kit aero profile etc....
A few variables like wind and gearing Strava doesn't take into account. Those 2 things could be a pretty big determining factor. My experience is Strava power tends to be way off in most cases.
You touched on something I was researching.. Downhill. Trying to find out: if you don't pedal downhill, how does this affect overall ride avg power (Strava), compared with spinning over at high cadence (at similar speed)? E.g if you have 0 cadence going downhill, does that affect your overall ride avg power output? Is it better for overall avg power output to spin at high cadence downhill or coast? Thanks!
Thank you for the info, quick question though - Do you weigh in with all the riding gear or regular morning weigh-in?
The power is good read from your garmin, as soon as strava uploads my power is way off,I am not talking about estimated power,I asked why its because the segments apparently, VC YER.
+The Vegan Cyclist - Thanks for doing such comparison. How is the Powertap P1 holding off so far? I am in the market to get a P1s due to budget constraints. If possible could you do a comparison between a P1 and a P1s? Thanks!
Did you ensure that the phone recorded the same speeds as your head unit did? The gps reporting may have lags and therefore will report exorbitantly fast time between subsequent waypoints, hence, 1000+ watt peaks.
Try 2 speed sensors: 1 for no power, the other with power.
Great video very useful, other thing would see if a not smart trainer like my Cyclops fluid 2 ZPower is close to a real power on zwift?
Hey bro really appreciate this experiment I was looking for this for a while (don't have a power meter yet, but I'm already looking for one) I wonder how much wind did you face? We're i live there are freaking crosswinds that Puch you out of the road and always believe that this windy situation ruin the accuracy of Strava power
Thank you so much for the good info
algorithm depends on EXACT weight of your bike in riding condition (w water, tools, spare, etc) and the EXACT WEIGHT of your body with shoes, clothes, helmet, etc. if those #s are accurate, the avg power on Strava should be very accurate (and could also be calculated on the fly for like 10 sec or 30 sec avg)
there s gps error to take in . Usualy for a phone, the gps data is preety buggy. Try and use your garmin without power meter and compare that data to your phone, you will see the diff. Also the max power is higher because of the lag of the data comming in. Gps devices record data in bucket data structs.Depending on those bucket size your gpx file is more or less accurate. If the bucket is big you will get a huge lag of your surge which translates in a huge power. There s nothing wrong with the Strava algorithm, it s your recording device thats prone to errors.
Interesting video. Just out of curiosity what is the bike weight that you put in?
Could you do a follow up video using a cadence sensor with strava.
do this again please to see if Strava has gotten any better
What about riding into a strong headwind? I'd have to say that it's probably grossly wrong. I've had some rides where I was going into the red and strava said I was doing 150 watts, I'd have to say I was doing 250-300. I don't currently have a power meter hooked up...suspect I need one.
I just went for a ride today and checked strava. It was hitting some peak power which is 2x what I could achieve on a smart trainer.
Would appreciate a review of the power edge pro CL by xuxuxu (garmin IQ connect) vs actual pm.