Race Predictor - broken?

Chris76 said:

Any one have any advise, other than to ignore it?

I have had (much) better predictions when I input a specific race event, with a course. Your mileage will vary.

Just understand that the race prediction takes your physiological metrics and your training history into account. From that perspective, for the algorithm, the last race is just one data point among many others.

Stry provides also race predictions. Unlike Garmin, they tell you the range of error. In my case, about 30s for a 5k (!) and a couple of minutes for a HM.

While my race specific predictions are a bit optimistic compared with Stryd predictions, the standard race predictions are way too optimistic. I know this is a sign of some data being off. I found out recently that the calibration of my Stryd was 5 percentage points off (I always use Stryd for pace and distance). That could explain this. It could be that the HR Max I have on my watch is a bit off. Still working on this as well.

You may find some race predictors online that work better for you (hint: runalize.com). For me they are worse (maybe not good enough of an athlete to fit nicely the population feeding these systems :-)

My predictions are off by quite a bit, too. I’ve been training consistently for months, have good HR readings (always use a strap), have my max HR set to the right value… and my half marathon prediction is 9 mins slower than what I just ran in a race. In fact, my 5k prediction is slower than what I’ve run in practice many times over… sometimes covering 25km at a quicker pace than it predicts I could race for a 5km. I can’t figure it out. Their algorithm used to be fantastic, but when they changed it a few years ago to “learn” me, it has never been close since. 

I have similar issues. I ran 5km in 22:43 yesterday, but my racetime prediction is 23:31. 

I have similar issues. I ran 5km in 22:43 yesterday evening, but my racetime prediction today is 23:31, the same as before yesterday.

Race time prediction is based on the VO2 max estimate, and I believe there is something off with that estimate recently. For some reason, my heart rate during training has been significantly higher in the last few weeks than before. I don't know why, it may be that outdoor temperatures have been dropping..? As a consequence, for the same training activity I get much higher active load values, and watch warnings that I should rest more.

My max heart rate during running yesterday was supposedly 191, but the updated max after the run said 186. A few weeks ago my max was 176.

I hope that Garmin developers also look at these forums, and pick up on this. 

Update:

Since i posted this, there has been some changes to the predicted race times, and they are falling.  This seems to be a response on the watch to a fast 5K and fast 10k i ran on the weekend.

The 5 and 10 predictions are now faster than i ran, although not by much, so it is actually giving me good stretch targets for those distances.

The 21.1 has come down, although is still below the time i ran by 3 minutes, rather than 6.

Seems that just following the training plan on the watch, which is heavily biased on slow (slowish) runs, has a negative impact on the race prediction algorithm. 

7348796 said:

My max heart rate during running yesterday was supposedly 191, but the updated max after the run said 186. A few weeks ago my max was 176.

I recommend you disable the HR Max auto detection. Garmin has improved it, but it is still subject to outlier HR data.

Assuming you were wearing a chest strap during your 5k, your ran it as fast as you could and your finished strong, your HR at the end of the 5k is an excellent base to estimate your HR Max. Add 5bpm to that number and that should be your HR Max. Input that value as your HR Max and your VO2 Max model will adjust over a couple of weeks.

7348796 said:

Race time prediction is based on the VO2 max estimate,

The race time predictor does use the Vo2 Max estimate, that is correct, but in combination with your training data over several week

https://www.garmin.com/en-US/garmin-technology/running-science/physiological-measurements/race-time-prediction/

For the Vo2 Max model and the race predictor to be more accurate, it is important to do keep a balanced workout peppered with a range of maximal efforts (ie maximal steady pace for different interval durations).

Chris76 said:

Seems that just following the training plan on the watch, which is heavily biased on slow (slowish) runs, has a negative impact on the race prediction algorithm

The workout suggestions tend to follow a 80/20 polarized approach from my experience. To get good metrics on the watch, from time to time, you should replace the supra-threshold workouts suggested pace by maximal steady pace for the duration of the interval. This is the best data you can provide to the watch about your capabilities.

Chris76 said:

Seems that just following the training plan on the watch, which is heavily biased on slow (slowish) runs, has a negative impact on the race prediction algorithm

The workout suggestions tend to follow a 80/20 polarized approach from my experience. To get good metrics on the watch, from time to time, you should replace the supra-threshold workouts suggested pace by maximal steady pace for the duration of the interval. This is the best data you can provide to the watch about your capabilities.