Garmin watch 'Race Predictor'?!?

I have a 245 and I find the race predictor to be garbage for me. I'm guessing it is because I have a super high resting heart rate (I'm in my mid 40s, yet easy runs are like 180 bpm). My long runs (~13 miles) are sometimes at a faster pace than it thinks I can race a half-marathon. Most other times are way off too.

old west wrote:

I have a 245 and I find the race predictor to be garbage for me. I'm guessing it is because I have a super high resting heart rate (I'm in my mid 40s, yet easy runs are like 180 bpm). My long runs (~13 miles) are sometimes at a faster pace than it thinks I can race a half-marathon. Most other times are way off too.

I find it hard to believe anyone is running 'easy' with at 180 bpm. I suspect there is something wrong with your HR measurment. Are you using a strap, because some people (not all, including me) have reported that wrist-based measurement is too inaccurate. Also, a chest strap can also be inaccurate, depending upon the model, typically due to the electrodes getting too dry, and electrode gel can usually solve that problem.

The race predictor on my Garmin is hilariously wrong.

At present:

VO2 max = 59

5k = 18:02

10k = 38:00

Half = 1:24:26

Marathon = 3:02:02

The most ridiculous part about this is my watch will tell me I am capable of an 18’ 5k minutes after uploading an activity that Garmin where I covered “5.05k” in 16:14. I understand it is just making a conversion from VO2 max but you would think Garmin would factor some common sense real world data into their calculation. Similarly I better that HM prediction frequently during long run workouts with an average HR in the 150s.

The strange thing is other applications like Runalyze take my Garmin generated data and interpret it pretty correctly (albeit a little optimistically) so I am confident it is the Garmin algorithm and not just wonky watch data.

I am pretty amazed yours is so accurate honestly.

I’ve had my watch for quite a while, and use a chest strap when running, It’s always ranged from mid 18s to low 19s until recently when I strung together a fewgood workouts.

I’ve been using a 935 for the last few years but had a 945 for a couple of weeks (tried it for the 36-hour battery life which totally didn’t live up to expectations). The 935 estimator is generally way off but the 945 was much closer. Both rely on VO2 max estimates which rely on HR data...and we know the issues we can have with that. But at least the 945 (and the 245 because the updated Firstbeat algorithm is used in both) can adjust predictions even without a full point change in the VO2 max estimate.

For example, if you have a 56 VO2 max estimate on the 935 it will always give you the same predicted 5k time. Get to 57 and it says that pre-determined estimate for that number. The 5k number will be the one that correlates to a marathon and vice versa. If you’re training for a 5k then your marathon time will be ridiculously inaccurate - sure, maybe you are estimated to run a 18:26 min 5k but it will say you can run a 2:56:32 marathon.

The 245/945 do a much better job of looking at data based on how you run. If all you do are runs 6 miles and under, for example, your 5k estimate might still be 18:26 but your marathon estimate may be like 3:30 or something. And even with a VO2 max estimate of 56 it will micro-adjust. By this I mean it might estimate an 18:26 one day, then 18:18, or perhaps 18:34 even though your VO2 max remains 56. I found it to be much more accurate once I ran with it for a couple of weeks.

Importantly, your 245 adds weather into the equation. If it’s a hot day, it’ll consider that data point whereas the previous generation didn’t.

Relying on wrist-based HR data is always wonky though. Sometimes my 935 gives me insane marathon estimates of like 2:45...yeah, not happening haha. But if you run hilly routes in the heat, it’ll give you woefully slow estimates. Don’t even think about going on trail runs...ugh.

I think it’s mostly a gimmick but the 245/945 do have more accurate estimates than prior versions.

Not sure it really matters if my HR on Garmin is accurate or not, but it's what likely leads to my race predictors being way off. It thinks I'm struggling to run 7 min pace when I'm not. As for my high HR, I tried a chest strap long ago (high school) and was easily >210 bpm.

Check out

https://www.firstbeat.com/en/

Garmin licenses the algorithms from them traing status /VO2 Max

As other posters have pointed out it only knows your hear rate and the heart rate variability since the Garmin Heart Rate sensor is good its not as good as a chest strap for a number of reasons. Therefore the algorithm has to work with some less then perfect data. It doesn't know if your heart rate is high or variable because its hot or because you stayed up late or because you had to may beers last night

VO2 max over an extended time period is probable;y pretty accurate but race predictor not so much especially at longer distances

I did a mile TT in 5:21 this wknd. Not the best, or worst, weather conditions. As a result I’d have to say 17:02 (just under 5:30 pace) is pretty unrealistic. The Garmin updated the estimate to 17:10. I’d say in ideal conditions with a fast course and people to run with I’d do 17:35 +/- 15s. Just no way at all I can touch 17:10 or lower. Unfortunately with COVID going on and summer heat, will be a while before I can really test it.

Thanks for your informative post.

My wrist-based HR measurement seems to be pretty good so far, and as you said, the (newer) 945 predictor is much better than the (older) 935, because:

(1) it apparently takes into account total volume (type) of training and thus does not give equivalent distances across-the-board based on VO2 max alone;

(2) it also micro-adjusts the predictions, even if the VO2 max number stays the same.

I think I saw something is the advertisement literature about temperature for the 245. How is the 245 taking into account temperature? Is its chipset monitoring temperature, even though a temperature measurement is NOT provided to the user on that model? Or is the algorithm assessing temperature based upon the GPS location, and cross-correlating with weather data for that locale?

NX64000 wrote:

Check out

https://www.firstbeat.com/en/

Garmin licenses the algorithms from them traing status /VO2 Max

It doesn't know if your heart rate is high or variable because its hot or because you stayed up late or because you had to may beers last night

VO2 max over an extended time period is probable;y pretty accurate but race predictor not so much especially at longer distances

But the algorithm does seem to be taking into account the type/volume of training to reduce predictions for longer distance races. And as Murdock stated, the 245 is somehow taking into account 'weather'.

Since Garmin is licensing Firstbeat algorithms, does that mean all the data collected by the watch is being tracked by Firstbeat, as well?

cool story, bro wrote:

[quote]NX64000 wrote:

And as Murdock stated, the 245 is somehow taking into account 'weather'.

I'd be shocked if it's not taking into account heat. I think that's why it's overstating my ability . . . I do really well in the heat, so it's looking at my workouts and projecting something much better than realistic [for me] for better weather conditions.

Where do you even access the Garmin race predictor?

I have a Garmin and have never seen this.

Which Garmin model do you have? That feature is not available on all models.

The race predictor is waaay off for me (245) I would say around 7 minutes too fast for 10k and maybe 20 minutes too fast for hm. I dont know why it's so messed up, the heartrate from the watch seems pretty accurate when I compare it to a strap or a manual measurement and I've used the watch in 100s if runs. Maybe I just really suck at giving it my all in races.

walkingfastlol wrote:

The race predictor is waaay off for me (245) I would say around 7 minutes too fast for 10k and maybe 20 minutes too fast for hm. I dont know why it's so messed up, the heartrate from the watch seems pretty accurate when I compare it to a strap or a manual measurement and I've used the watch in 100s if runs. Maybe I just really suck at giving it my all in races.

Interesting...since you seem to have an idea on your actual times up thru the HM, how does your VDOT (per Daniels' or McMillan's calculator, NOT the Garmin VO2max number) line-up across the distances? (i.e., are you aerobically and/or structurally deficient with extension of the race distance?)

I have the 245M and the race predictor is pretty off for me. My V02 Max reading is 57 on the watch and predicts my 10k is 39:23, yet I ran a 10k 2 weeks ago in 38:24. Likewise, the week after I ran a 5k in 18:13, yet the race predictor says I should've run that in 18:41.

The corresponding VDOT values for a 38:24 10k is 54.5 and 5k 18:13 is 55.5.

doogiski wrote:

I have the 245M and the race predictor is pretty off for me. My V02 Max reading is 57 on the watch and predicts my 10k is 39:23, yet I ran a 10k 2 weeks ago in 38:24. Likewise, the week after I ran a 5k in 18:13, yet the race predictor says I should've run that in 18:41.

The corresponding VDOT values for a 38:24 10k is 54.5 and 5k 18:13 is 55.5.

Were the course distances certified? Were the routes you train on similar to the courses? Was the weather (temp/humidity) in the races the same as when you train? etc., etc.

For the sake of example, let's assume the answer to those questions is "yes". Your 38:24 is equivalent to a 19:00 5k per Daniels' calculator, or a 18:58 5k per McMillan's calculator, yet you ran 18:13 (whether or not you were fully recovered), indicative of being aerobically and/or structurally weaker at the 10k relative to the 5k, and the Age Grade Tables indicate that, because (assuming you are a 40 yr old male) your 5k was 75.3% while your 10k was only 70.4%.

The watch was:

(1) actually in the 'ball-park' in its absolute time predictions,

(2) telling you your VDOT was not lining up, indicative of giving you a stronger 5k prediction relative to your 10k result. In that sense, its 5k prediction did better than using the calculators alone.

My mistake, I plugged your predicted 10k into the calculators rather than the actual race result.

For the sake of example, let's assume the answer to those questions is "yes". Your 38:24 is equivalent to a 18:31 5k per Daniels' calculator, or an 18:29 5k per McMillan's calculator, yet you ran 18:13 (whether or not you were fully recovered), indicative of being a little aerobically and/or structurally weaker at the 10k relative to the 5k, and the Age Grade Tables indicate that, because (assuming you are a 40 yr old male) your 5k was 75.3% while your 10k was only 72.2%.

Murdock wrote:

I think it’s mostly a gimmick but the 245/945 do have more accurate estimates than prior versions.

Yeah, pretty much, but another data point to use along-side the calculators.