Dubious Strava KOM/QOM records

What’s with all the b/s KOM/QOM records for Zwift segments being put up on Strava?

Take, for example, the recent records for the Epic KOM: Most of them posted in February and March of 2021. And all of them in 12-15 minute range. That’s cycling up a 4% grade, for almost 10km, at more than 35 km/hr. Some back-of-the-envelope math suggests a normal-sized adult male would need to put out well over 800 watts to do that. But Strava records them as only doing ± 250 watts. And you’ll find similar nonsense on all the other Zwift climbs.

Are people doing Zwift routes using “smart trainers” that simply don’t simulate grade? Because the numbers just don’t add up on most of the KOM records for Zwift climbs on Strava.

Yeah. Zwift physics aren’t always great (calculation of Draft, Double Draft & Sticky Draft), but in Feb / March were alot of races up those hills so some put a bit more effort in, maybe some cheated with weight & height, …
The gradiant simulation on the trainer doesn’t do too much (alot of videos are out there discussing that matter, e. g. GP Lama).

1 Like

It’s also possible these times were set during group workouts or meetups with “keep everyone together”. I’ve see some odd speeds when the rubber banding does its thing, and there were a couple of other posts on here recently where people late joining meetups were causing group speeds to fly up.

1 Like

Maybe.

But I was just looking at some guy who apparently cruised up the Volcano KOM at almost 30 miles an hour, a hair over 5 minutes, while averaging a strenuous 188 watts.

That’s one heck of a draft. And who were they following?

This isn’t sour grapes or envy at better athletes. I know darned well there are many, many truly remarkable athletes and cyclists out there. And many of those people participate on Zwift. People who can put out truly epic amounts of power and endurance. But their achievements are a) matched by real world performance and b) don’t fundamentally contradict the laws of physics.

The keep together bug has caused it for the most part - there was a bug that inflated the speed so there was a load of people who smashed up the hills whilst producing relatively low wkg… Not there fault, another one of zwifts improvements…

3 Likes

I recently bought a Wahoo Kickr Snap and rode it for about a week before getting the bright idea to turn on my head unit to see how my Power Tap P1s compared. I’d assumed the Snap was perfectly accurate, since it has lots of reviews that say so. I was shocked to find that despite all of the calibrations being performed exactly per the instructions, the Snap was reading 17% high! I fixed it with a bit of fumbling around doing factory spindowns over the next several rides and now it behaves within 0-3% of the Power Tap (so no ill will towards Wahoo).

This isn’t necessarily the answer to your question, but I’ve wondered how many people out there are running a trainer with a huge error and have no idea. Without a 2nd meter reference I’d have been legitimately clueless that I’m not actually a 4w/kg rider.

One thing I don’t understand is how some of the top Strava Segment result elevation graphs are NOT Zwift elevation. How is the Segment matched, if the elevation is not the same as Zwift?

I’ve seen that too, even in my own rides. I’m not a strong rider, and yet with a 5% grade I can if I choose to pass a lot of others by increasing my speed and cadence. Watts does not increase comparatively with the grade, nor does it decrease when going down hill. I have kept the same cadence both uphill and downhill with the same watt output. Personally, I don’t keep track of watts for that reason and others, but I know others do. Also, one thing, riding a bike in real life, there is usually some road resistance, and more importantly air resistance -(wind speed). With the showing of the grades in % is good, the length of the grade in metres, or yards if you so choose, would be helpful too.

The zwift KOM’s on strava are a bad joke. Look at all the guys over 55 beating the world tour guys up the alp last year. The pros did around 36-38 minutes and they are on page 2 of the leaders I think (kiwatoski, Rohan Denis). I clicked on a few old guy (me) age group leader board strava pics they guys were big and over weight and one guy even typed “don’t read too much into the time or numbers” or something like that. Meaning he did not do it legit, but still left his name up there on top? Ego I guess.

Just do what you can and realize it’s cartoon video game riding and some are doing more gaming manipulation than true racing.

1 Like

There is obviously a breakdown in the way Zwift data is used by Strava. Which is a bit of a shame. I like both platforms (Zwift and Strava) but the KOMs records are worse than useless.

I understand why Zwift doesn’t want to restrict users to only those that have proper smart trainers. So its possible to sign up with an old-school trainer and a speed sensor. And Zwift put a lot of effort into working out good power curves for those trainers. And if used and adjusted correctly those power curves can give the user a pretty accurate experience. But there are so many things that can wrong. The resistance knob can be too loose. The tire pressure can be wrong.

How to fix this issue? I don’t know all the technical details, but I wish that Strava would include information on the make and model of the trainer being used to record these virtual KOMs. That would, I think, allow people to separate the junk from the real achievements. I think this is more of a Strava problem than it is a Zwift problem.

This is a huge bug/feature in the Meet-up code that by attempting to keep people in the group is accelerating the whole group up hills way faster than anyone in the group is actually pushing hard enough to go. This is not something complicated like hardware calibration - THIS IS A SETTING THAT YOU SELECT IN ZWIFT!! Strava leader-boards for Zwift are completely dominated by these low power, super-speed MEET-UP efforts. It actually makes a complete mockery of Zwift’s relationship with Strava, nevermind it’s relationship with cyclists and reality!!

1 Like

OK. This “Meet Up” bug is something I haven’t previously understood or encountered. But if it is the source of these ridiculous KOMs, then something needs to be done about it, by someone. Either Zwift, or Strava, or the individual participants. Because it is making a mockery of the whole process.

How can you possibly measure your progress as a cyclist against your peers if a certain percentage are posting results that are completely fictitious? Surely it would be possible for people using “Meet Up” to have their times on stages nullified?

It happens in Keep Together meetups, because those are the ones where Zwift regulates the riders’ speeds in relation to others.

I agree that Keep Together meetups shouldn’t count for Strava segments. But I doubt that Strava has a mechanism to recognise this.

Rob, so true, and its for that reason, I just for the most part ride my own pace,(pushing it too sometimes) and knowing I will never beat those ghost chaps who are never seen on the path who end up at the end within minutes. I would guess that they somehow are grossly cheating making the route as if they were doing around 80 km/hr. That is the only way I can see it being done, perhaps they have 2 speed sensors on their bikes, thus doubling the speed they are actually getting. Heck, I once took a second magnet and mounted it on my bike and wow, I was doing 80 - 90 km/hr so easily! Perhaps I should put 4 sensors on!, but, you know, I’ll leave it at 1.

David

Well it makes me question the value of “Keep Together” meet-ups at all.

For as long as there has been cycling, there’s been an issue keeping groups together. And the answer comes down to: Either the faster riders slow down, or the slower riders work a little harder. Or some combination. Or you just don’t ride in the same group. It seems to work just fine in the real world. Why do we need it in Zwift?

They can keep “Keep Together Meet Ups” if they want. But Zwift has got to make it so that segment data doesn’t get exported to Strava. Because it’s messing up Strava segments for everyone else.

And shame on anyone letting Strava record speed and climb data you didn’t really earn.

You’re missing the point of this bug. It doesn’t just make the slower riders go faster, it makes all the riders go unrealistically fast - much faster than the fastest rider would otherwise go.

Even more reason for Zwift to dump it, then.

I can tolerate riding through an active volcano. I can hold my nose and deal with riding up glass skyways a couple hundred feet over Central Park. I can even tolerate a “difficulty” setting that gives your bike an infinitely variable transmission.

But a meet-up setting that lets <2.0 watt/kg poseurs think they’re climbing HC segments faster than Marco Pantani on crystal meth? That’s really taking things too far.

Because Zwift isn’t outdoor riding. Keep Together meetups are a great feature and used extensively by my club.

Instead of splitting into multiple groups (like outdoors), everyone can enjoy being in the same ride, no matter their ability. This makes it very social and inclusive. It also allows everyone to have as hard or as easy a session as they want, while being kept together on the road.

Calling for a feature with lots of benefits to be dropped just because of a bug causing a side effect to a third-party service doesn’t make any sense.

It’s a bug. Zwift should fix the bug.

3 Likes

This is untrue. There is no setting in Zwift that makes everyone go much faster than usual.

There is a bug in Meetups, when the Keep Together option is selected, that sometimes the speed algorithm goes wrong and everyone goes too fast.

This bug has been discussed plenty of times here on the forum in the past, and Zwift are well aware of it.

It used to be quite a rare thing to happen, but a few updates ago it became more frequent.

It’s on Zwift to fix the bug.

1 Like

That makes more sense.
I thought the rubber band sped up the slower people but slowed the front people.