Is race predictor utterly broken?

I think everything above a marathon cannot be really predicted by Garmin. This is also true for races with a lot of elevation, but there are signs that at least the latter soon will change for the better

Done half marathon 4 days ago in 1:26:52, so race predictor is quite good

VO2max went from 57 to nearly 59 today in 5 weeks

VO2max is not stuck, you have to train better

At the very least, they should take your marathon prediction, and use incrementally slower paces for your ultra marathon predictions. Predicting that I do a 42k at 5:36/km while at the same time predicting that I do a 90k at 4:40/km is totally ridiculous.

Also, daily suggested workouts seems to be tied in with the race widget and it's prediction, hence the suggested paces are too fast. Obviously most people will not use daily suggested workouts to train for an ultra, but some might, since it's part of the daily morning report, and they will totally be doing the wrong training.

I probably bring nothing new to the table, but here are my observations.

My vo2max is moving, race predictions are a little bit optimistic, but not too far off. I guess the predictions are more correct if I run on track, no wind, perfect temp and using a pacer

That said, my vo2max is also flatlining in periods when doing indoor track and treadmill - which makes sense.

The modes Track Run and Run do affect my vo2max just after saving the activity. Just a thought, but other modes like Trail run may also be excluded as vo2max activity, have you tried different modes?

I have also noticed the race predictions under the vo2max menu is different when comparing to a future race in my Garmin calendar. My vo2max 10k prediction is 38:21, but my September 10k race is predicted to be 38:48.

The reason could be that they use the same algo but adjust for race elevation data (+/-61m) to make a more correct assumption.

Just a minor bump and to put things into perspective...

I ran a marathon today, breaking my PB and getting a sub 2:45...  (just ;) )

Garmin predicted (and still predicts) a 2:47 so that's pretty damn good. The result today really was about the max in it...

I was also wearing a stryd, that was predicting something in the range of 3:02 and now still IS predicting a 2:59.

Just noting that Garmin isn't the only one that can be off, and completely disregards actual results, Stryd which for many is THE standard wrt accuracy is even worse in this case...

Stryd is precise that's the fact, but only for pace/distance. It's race predictions are based on so called critical power. In order to keep the CP correct you need to periodically run all kind of different lengths nearly max effort. That's thing that I don't want to do - keep feeding data to the gadget instead of training. So for me this part of Stryd is unusable and race predictions are of course off.

Marek said:

you need to periodically run all kind of different lengths nearly max effort.

That’s kinda stupid. If you do some distance at near max effort you can use a online race time calculator to get a time for a desired distance.

All these companies are full of marketing. They want you to believe their approach is onso scientific, but in reality it is quite simplistic. But they charge a premium price for it.

I just wanna add that the race predictor seems to be u0dated with the new release. The prediction for my upcoming 24 km trail run with 1000 m elevation was way 1:50 before the update which changed to approx 2h after the update. It also suggests now to do more hilly runs as preparation

Congratulations on your PB  insanely good performance hope to run a marathon myself one day. Also good to see another good prediction from Garmin

My current achievable / believable offsets from predictions are +2 mins for 5k, +3 mins for 10k. I suppose they assume 100% maximum effort.  The HM and M predictions are fairly big extrapolations, haven't run those for years.