Garmin Race Predictor - Useless?

The algorithm that it's based on is flawed (supported by research but the research itself is looking at it in the wrong way). As you say it's heavily weighted off Vo2max which is a cardiovascular health metric rather than a performance metric. Think of it like trying to measure weight lifting strength from your BMI.

Garmin & Runalyze are within a second of each other for me. The prediction looks reasonable enough as well but of course YMMV.

I have a Coros watch, my VO2 max prediction on the watch is 60. I recently did performance testing and my actual VO2 max was 50.

Runalyze is pretty accurate for me, especially after you log a couple of races and use it to adjust the correction factor then it becomes pretty bang on. Although it still does demand non-stop long runs every weekend or you get penalised on the "Marathon Shape" parameter.

I think my marathon predicted time was within a minute of where I actually ended up.

e: Garmin predictions are pretty shite in the opposite direction, where the algorithm doesn't weight long runs as much as interval training, and unlike Runalyze you can't manually adjust it to reflect the real stats you accrue from racing efforts.

Ultimately though, it's all speculation based on "optimal racing conditions" which rarely reflect reality. Use it for setting training goals and paces and tracking progress, but don't bet your house on it

My Garmin is way off but my runalyse is even worse!

I ran a max 5k last week. Garmin predicted something like 24:00. I ran 21:25 which is close to what I expected.

Runalyse gives me a range from 24 to 34 mins!!

I would say useless. It had me almost 10 minutes slower than my actual race times. Garmin doesnt seem to have much of a read on what type of training methodology you are on. I would know that certain types of aggressive interval session are a good indicator of race pace (e.g. 3 x 2 mile) - I'll use that as a rough indicator to know where I'm at for a 10k for example. Then scale it out for a HM by roughly another ten seconds per km

I think the newer watches seem to be getting better, maybe more accurate HR data? I upgraded from the Forerunner 55 to the 165 recently and the race prediction times are fairly close to actual times I've run in recent races.

The 55 never got closer than predicting 1 full minute slower for 5k, and around 10 minutes for a marathon. The 165 is now about 1 second off my most recent 5k, and the marathon time is still about 5 minutes slower but I haven't raced anything long enough to give it much information yet. The 10k and HM times are probably within 30 seconds or so of what I would run currently. Again I'd expect those will probably tighten up a bit if I raced those distances soon.

Unless you’re wearing a heart rate monitor chest or arm strap then it’s pure pie in the sky stuff but you’re into getting proper numbers you’d be wearing them and not using the watch only.

Sure is a good athlete, trains with DTC as well from time to time. He's definitely one of the good guys when it comes to sports science.

I'm not sure if you heard his other podcast with Ross Tucker. Some great information in there.

https://podcasts.apple.com/za/podcast/can-we-trust-sport-science-research/id1461719225?i=1000649056583

Mine predicts me at around 2 mins faster than I really am for every race distance.

I use it purely for humour at this stage.