Skip to main content

Good news! We’re updating our algorithms to make leaderboards more credible, and taking steps to proactively prevent suspicious efforts from appearing on leaderboards – so you can trust that the results you see are accurate. We're withholding from leaderboards any activities that appear incorrectly labeled (e.g. a bike ride getting uploaded as a run) or have faulty GPS data.

With these and other ongoing changes, we can better ensure your efforts will get the ranking they deserve and you can trust that the CR, KOM or QOM (plus all the other times on the leaderboard) are the real deal. These changes will be for all activities moving forward – we can’t yet capture past activities. And of course, you can still report an activity if you think there’s one the algorithm missed. Learn more about that here.

This is a big undertaking – we saw around 26,000 1-mile running splits that would have been faster than the world record last week! – and the work will be ongoing. Here’s how we’re committing to cleaning up leaderboards right now:

The Details 📝

✔️ There’s a new threshold for flagging: Activities with too much erroneous data will be automatically flagged – which means all segment efforts from those activities will be withheld from leaderboards.
✔️ We’re doubling down on catching bike rides (or downhill ski runs, car rides, etc.) marked as runs with new run-specific parameters that will flag activities based on their distance and pace data.

Another example I’ve come across: what looks to be trail running categorised as a ‘hike’. One example I saw was 13km in 1hr 17 mins, with HOKA CarbonX2-1 shoes which are running shoes. They even cheekily called it a ‘fast hike’. Everyone in the top 10 posted similar times. Now, I’m not sure how an algorithm would be able to stop this as depending on the elevation etc fell running might be at a walking pace in certain sections.

For reference: I did the same section in 2 hours at a decent hiking pace.


I often find that when using some Garmin devices (I test many of them) with the energy-saving mode activated, the GPS detects fewer points.
I have noticed that some KOMs appear on my profile but unfortunately (!) they are incorrect.
It's easy to identify them because they have a higher average speed than the maximum speed, which is actually impossible.
This occurs on very short segments.
I suggest implementing a simple yet effective check: a segment where the average speed is higher than the maximum is mathematically incorrect and impossible, it should be not reported in the KOM lists.
What do you think about this?


segment matching seems to use a different algorithm from ride segment analysis, so it should be readily apparent. it's quite common in built-up areas and short distance efforts, the graph displays a very different set of metrics than the segment matcher.



We don't have any updates to share right now, but your input and examples have been passed on to the Team.  


I'm still seeing world record breaking runs (rides uploaded as runs) and rides (forgot to stop their activity recording and drove home) at the very top of leaderboards. Athletes who have had their activities flagged and who have been reported directly to Strava Support continue to upload activities that go against your terms of use. Could you update us on what has been done since January 2024 to "Improv(e) Leaderboard Accuracy"?


Last month this guy took a cab to the airport and then TOOK A PLANE to another city (Avg speed 286.2 km/h - Max speed 872.3 km/h). Did Strava automatically remove this activity from the leaderboard? No!!! How come?!

https://www.strava.com/activities/11030771091/overview


Have no fear, dear paying customers! Here at Strava "we are continually evaluating and improving the algorithm"


I'm good with doing my bit to help keep leaderboards clean, and I can clean a segment up over time myself with the flag option. The only problem is I feel like I'm the only one doing so, and after a few flag submissions I am 'capped' by disabling my ability to do so for a while. I get that leaderboard integrity is community-driven, and that the onus is on us to flag these times, but even then I get limited in doing so.


@Bernhardt - I agree.  I am like you.  I will go through regularly and keep the local segments cleaned up, but it can be hard when they only give you 10 flags in a 24 hour period.  I have another post here (Modify Activity Flagging Process - Strava Community Hub) where I made suggestions to improve the flagging process such as resetting at a set time rather than just 24 hours after the last flag and also approving local ambassadors who would have additional flagging capabilities.  Feel free to check that thread out and add your support (kudos) if you agree with the ideas.


Meanwhile in 5k Brooks Challenge...


Look at these awesome activities that were not flagged automatically:

https://www.strava.com/activities/11162527264#3216134935065909580
https://www.strava.com/activities/11214927165/overview

 

400 km/h average speed!


Impossible!

(Sarcasm)
We’re updating our algorithms to make leaderboards more credible, and taking steps to proactively prevent suspicious efforts from appearing on leaderboards – so you can trust that the results you see are accurate.

(/sarcasm)


@DannyDindia wrote:

Look at these awesome activities that were not flagged automatically:

https://www.strava.com/activities/11162527264#3216134935065909580
https://www.strava.com/activities/11214927165/overview

 

400 km/h average speed!


- Strava


I think it's a great idea. But more importantly Strava does not seem to think so.


it's not always athlete-driven errors though, and individual segments cannot be flagged for having invalid algorithm matching results.

the top two on this segment have glitched times; https://www.strava.com/segments/11425394

if the 'analysis' tab is selected for each effort, it shows max and avg speeds considerably lower than the segment match results.


Well, If you don't process past activities righ after having a decent algorithm, we are doomed because they are piling and with 10 flags a day it's impossible to keep local segments clean.


Strava made this announcement over a year ago, and I'm still getting multiple notifications per day from people that have taken my KOMs and are very obviously driving cars. When is this actually going to be implemented?


@mal @Jane11 @Scout - What is so hard about creating some basic controls on activities that would filter out OBVIOUSLY bad activities?  How can someone load a "ride" that has an average of 45mph with an average heart rate of 67bpm over 56 miles and your system does not question it?  This is F-ing ridiculous!  See the activity below for just one example.  This happens ALL THE TIME, yet somehow your programmers can't figure out how to screen these out?  

Afternoon Ride | Ride | Strava (https://www.strava.com/activities/12044135676)


 438 mile per hour KOM. How does this get past auto flagging? Ridiculous.


Obvious troll or spam bot. Leaning towards spam bot. That reply is 100% AI generated.


https://www.strava.com/activities/12486956951#3273408790740594804
Date: August 30, 2024
Max Speed: 104.8 km/h
Elapsed Time: 575h

How come this wasn't automatically flagged?


@ARoger  I think this ride was edited after you have viewed it (maybe with "correct distance") as the high values aren't there.


**bleep**, I posted the wrong one. I've edited my post.


@ARoger - It looks like the new link you corrected it to goes to an activity that appears to have been edited as well.  Yes, the elapsed time still shows the 575 hours, but the actual activity is only showing 100 miles and just under 9 hours of total time, a very realistic set of stats for biking.  There appears to possibly be some anomalies in the data, showing a max speed of just over 61mph, but that could just be residual from what was cropped out just like the elapsed time.   I've had this happen when I have recorded an entire triathlon as one activity, then went back to "split" it into separate activities after.  One of the files (usually the first one) will retain the total elapsed time of the original file even after being split or cropped down to just the relevant part.


👋 Hello!
Thanks to everyone who has commented and provided feedback. We are excited to share a segment update with you that will improve leaderboard accuracy! 


Check it out here: We’re Changing Segments – For The Better

We will also be closing this post for comments as we will be reviewing the above post for any feedback and comments. Thank you!


Reply