Loading Notifications...

Strava segment anomalies

Please Register as a New User in order to reply to this topic.

I've been using Strava for about 9 years, probably for the last 5 or 6 I have not been paying much attention to segments, but in the last couple of months I have been working hard to get back to fitness after prelonged illness and injury and find the segments set up on certain speed sessions useful.

I did some hill reps as part of a small group last night, I felt I was going really well, much, much better than when I did a similar session on the same hills a couple of months ago.

I uploaded my watch and it shows me slower than both the times I did a couple of months ago, but also slower than people I was consistently beating in the session.

For example is gives my best time last night on one hill as 1:19, compared to 1:15 a few months ago and 1:11 for one of the people I was consistently beating. Looking at the stationary to stationary times on the Strava analysis of the run it was about 1:05.

It has occurred to me that I may be stopping a short distance before the end of the segment, but looking at this in more detail, I am pretty sure this isn't the case. It is not the first time that this has happened. Any ideas?

Report
 Basemetal 29 Jul 2020
In reply to The New NickB:

You're slower than you thought you were?

I had a similar experiece on a hilly circuit the the other day where I thought I was flying round but was 3 minutes over my PB...

Report
In reply to Basemetal:

> You're slower than you thought you were?

> I had a similar experiece on a hilly circuit the the other day where I thought I was flying round but was 3 minutes over my PB...

All the evidence suggests otherwise, particularly other runners who recorded quicker times, despite finishing well behind me.

Report
 Basemetal 29 Jul 2020
In reply to The New NickB:

Joking apart, that is wierd.

Report
 compost 29 Jul 2020
In reply to The New NickB:

10 seconds on a hill rep is, what, 20m? That's pretty close to the GPS accuracy threshold, which can be made less reliable by factors like cloud cover, atmospheric conditions, leaf cover (thicker leaves on the trees now than a few months ago) etc. 

OS map and stopwatch next week?!

Report
 bigbobbyking 29 Jul 2020
In reply to The New NickB:

Strava have some useful info about this kind of thing here: https://support.strava.com/hc/en-us/articles/216918187

"It's important to keep in mind that each GPS device records its own unique set of data even if, for example, you and your friend are recording the same activity on the same day. When we match you to a segment, we choose the GPS points from your file that fall closest to the segment's start and end. Because of the nature of GPS and the differences in device recording intervals, it's unlikely that there will be a GPS point exactly on top of the segment's start/end. This means that you can be timed over portions of your activity that are slightly longer or shorter than the actual segment length. Everyone is susceptible to this but you may get a short match that works in your favor or a long match that adds a few extra seconds."

Were you using the same device to record your previous and current activities? Seems most likely some GPS error against you and in favour of your running mates is to blame.

Report
In reply to compost:

You must run up very steep hills, 10 seconds is about 50m at the sort of pace I'm looking at. It is doing something strange though.

Report
 DancingOnRock 29 Jul 2020
In reply to The New NickB:

A friend of mine has a problem with his watch misreporting speed. His watch will show 6:00/mi while the rest of us are doing 9:00/mi.

Could be an issue with your watch. 

Report
In reply to The New NickB:

> You must run up very steep hills, 10 seconds is about 50m at the sort of pace I'm looking at. It is doing something strange though.

You must be very fit if you're running up hills at 20 seconds per hundred metres pace. That's equal to 3 minutes 20 seconds per km pace. Up hills!!! 

Report
 Nik Jennings 29 Jul 2020
In reply to The New NickB:

What are you using to record your time on? I know of at least a couple of people who were using watches to record and they were only taking a measurement every 4 seconds! This meant that when on a road bike their trace was often cutting corners so missing segments.

Report
 Yanis Nayu 29 Jul 2020
In reply to The New NickB:

Bloke I know got a KoM last night with an average speed for the segment which was about 10mph faster than his max speed for the segment.

Report
 gazhbo 29 Jul 2020
In reply to The New NickB:

I think the shorter the segment the less reliable the times are.  My running club had some local segment challenges at the beginning of lockdown and I definitely benefited from some dubious satellite positioning.  Strava had me running a segment slightly over a mile, 10s quicker than I ran a mile.  I beat a lot of people who are far quicker than me.  You have probably got the raw end of that.

Report
 Wilderbeest 29 Jul 2020
In reply to The New NickB:

Borrow your friends watch next time ?

Report
In reply to Wanderer100:

> You must be very fit if you're running up hills at 20 seconds per hundred metres pace. That's equal to 3 minutes 20 seconds per km pace. Up hills!!! 

For 300-350m. I can do 1km on the flat in 3:20, I can only do 300-350m at that pace on a 5% hill.

Report
In reply to Nik Jennings:

Garmin Forerunner 235

Report
In reply to The New NickB:

> For 300-350m. I can do 1km on the flat in 3:20, I can only do 300-350m at that pace on a 5% hill.

Like I said, you must be very fit. My best 1km time is 3.35 and that nearly killed me!! 

Report
In reply to Wanderer100:

It’s all relative, an international 1500m runner would be past 1400m by the time I got to 1km.

Report
 Nic Barber 08:05 Thu
In reply to The New NickB:

Best way to be completely accurate would be to use the lap function on your watch and markers on the ground.

I'd mainly say go off how you feel. The tolerances and errors of the human body are probably narrower than those of myriad GPS devices!

I use strava segments as an indication only, and then only over longer segments really as segments will be less accurate the shorter they are.

Report
In reply to The New NickB:

The other thing you can do is download the .tcx file from Garmin. That will open in notepad or a host of GPX freeware analysis software and you can see the actual distances in metres and velocities in m/s. Connect and Strava are very blunt tools. 

Report
In reply to The New NickB:

Frustrating but I think you just shrug your shoulders and try not to worry too much. I've done shortish races and seen wildly different distances from mate's tracks. It's a fun tool and very useful but definitely has limits. As I've got older and slower my local PBs have disappeared and not all of the leaderboards look legit to me. About the only hill segment I'm still top 3 on *I'm* suspicious of. I should flag myself really but it was an important run for personal reasons so don't want to wipe it

Report
In reply to The New NickB:

Just been looking at Strava and something weird is definitely going on. I did a ride with a mate a couple of days ago, and one climb segment in particular we were doing as part of a local challenge. We rode together (-ish) and he was faster than me by 13 seconds (he was on a gravel bike, I was on a full suss MTB that’s my excuse and I’m sticking to it...). We got back to his, had a coffee and uploaded our respective rides, and he was 13s faster, marvellous. I’ve just looked at our times for the segment again and I am now 2s faster, his time has had 15s added to it.

Now I was recording directly onto my phone, he was using a garmin watch, pure wild speculation but...
I wonder if maybe they are not quite as out of the ransomware woods as they think, could there be a bug in their system that is messing around with time recordings?

Report
In reply to Nik Jennings:

Strava does odd things to imported data. 

Report
In reply to The New NickB:

The end points of the segment can be very critical. It is best to start and stop the effort at points that are out of the segment by a decent margin. I once did a nice hill climb on the bike under wet and icy conditions (several slips, very slow). I came back later in the year on a nice dry day intending to do it faster. I clearly succeeded in this but according to Strava I was a full 15 mins slower. I eventually found the reason was that I had stopped for lunch before starting the segment and the lunch break had been included in the segment time. The segment started at a gate and I had lunch just before the gate. I assumed that GPS errors led Strava to think I had lunch inside the segment. To check I created a slightly shorter segment starting a few m from the gate and was rewarded with a time about 35 mins faster. I suggest creating your own private slightly shortened test segments to see if there is a big jump in time near either the start or finish. If there is it would suggest the time is influenced by errors in the start and/or stop location (so although you started or stopped at the right place it recorded you as being in the segment before you actually started or when you stopped). 

Post edited at 14:22
Report
 timjones 14:37 Thu
In reply to The New NickB:

> For example is gives my best time last night on one hill as 1:19, compared to 1:15 a few months ago and 1:11 for one of the people I was consistently beating. Looking at the stationary to stationary times on the Strava analysis of the run it was about 1:05.

Allowing for the accuracy of the GPS and the sample/recording rate I wouldn't be surprised by those sort of anomalies. Short segments can show up throw really bizarre "results".

Report
 timjones 14:39 Thu
In reply to The New NickB:

> You must run up very steep hills, 10 seconds is about 50m at the sort of pace I'm looking at. It is doing something strange though.

If you're running uphill at a pace of about 3:20/km you are either exceedingly fit or the hill isn't very steep.

Report
In reply to timjones:

Longer efforts on a 4% hill, shorter efforts on a 7% hill. Ours efforts tend to be longer than some recommend (200-350m). Try and sustain something about 10-15 seconds a mile quicker than 5k race pace.

Report
In reply to harold walmsley:

I’ve had a little play around, it does seem my watch has a bit of a GPS black spot at the top of the hill.

Report
In reply to timjones:

My best ever 1km time was 3:43, and that was downhill at a perfect gradient, just at the point where I would have fallen over had it been any steeper or I'd gone any faster. That was was 5 years ago, already in my 50s, but long exposure to others and my own disposition has taught me that I will likely be forever slow.

Report
 jk25002 19:46 Thu
In reply to DancingOnRock:

definitely. I use a suunto watch. For cycling, on Movescount a graph of my speed is smooth, imported to Strava it gets modulated with a ~0.05Hz signal at ~10% of the amplitude... sometimes...

For 1min efforts a lap timer will be more accurate.

Report

Please Register as a New User in order to reply to this topic.