Race predictions way off

I think HR max is correct. Garmin has ascertained it to be 186. My max recorded recently (3 months ago) was 185, from a chest strap obviously. That was when I almost died on the finish line!

The true difference between my 5k and 10k paces is about 6 secs/km. Despite every race being recorded on Garmin, the watch still insists the difference is 12 secs/km. I guess that's an improvement, albeit a slow one, as it used to give me 15secs/km difference.

Jens said:

Garmin has ascertained it to be 186. My max recorded recently (3 months ago) was 185, from a chest strap obviously

Looks right indeed.

Jens said:

The true difference between my 5k and 10k paces is about 6 secs/km

Interesting. It seems low, but every individual is different... Do you know your 1mi PB by any chance? When you enter your times in online race predictors is the 10k more "challenged" as well?

I have never run 1 mile flat out, that just seems painful, but I would guess it won't be much different to my 5k pace, maybe improved a bit only by a sprint finish.

On online calculators, if I input my 5k time, my 10k prediction is about 1min slow. If I input my 10k time, my 5k prediction would be unachievable. There is one calculator, however, on Marathon Handbook, the age graded one, where it requests gender, I find that almost spot on. Interestingly, if I change my gender to male (everything else stays the same), then the predictions are off. There is a YouTube video by Tim Grose where he looks at the proportion of men who can do both a sub 3 marathon and a sub 40min 10k and found that it was nearly all of them. However for women, he did not find that to be the case. Relatively, far fewer women who can do a sub 3 marathon were capable of doing a sub 40 10k.

Jens said:

I have never run 1 mile flat out, that just seems painful, but I would guess it won't be much different to my 5k pace, maybe improved a bit only by a sprint finish.

I was asking because there are good correlations between 1mi, 5k and 10k, but then the rest of your answer put that idea on the back burner.

Jens said:

Interestingly, if I change my gender to male (everything else stays the same), then the predictions are off.

Very interesting. Thank you for sharing. I don't know whether the Garmin's race prediction algo uses gender, it doesn't look like it does...

When you plug your times in a benchmark tool like this one (which factors gender in), do you find 10k is your strong suit?

runrepeat.com/how-do-you-masure-up-the-runners-percentile-calculator

This is getting crazier. No. According to that calculator, my strongest is 5k, followed by HM then 10k. That is among women and overall. I really need a new 10k PB this year! 

It's not for want of trying. I have had bad weather conditions, and I suffer more than most in the wind. That is why I prefer to look at power rather than speed. That way I can compare runs irrespective of terrain, hills and weather. I don't understand it, but the power metric seems to be consistent. Incidentally, my LT power according to Garmin is 10w lower than my actual HM power and 21w below the power of a recent 10k. I ran a 10k split within a HM equal to the 10k prediction. It's very poor that Garmin ignores actuals. Even if the LT speeds are wrong because of conditions, the power should be close? For a LT test, I may have to wait a while. The weather is pretty grim now and I do most of my runs indoors.

Jens said:

It's very poor that Garmin ignores actuals. Even if the LT speeds are wrong because of conditions, the power should be close?

Yes, but the algorithms are all based on pace. Maybe Garmin will update their research with grade-adjusted pace or even power, but it is note easy to execute all the research and validation required, before even changing the implementation on the devices. Who knows? Maybe it is in the works...

Are you using Stryd? The Stryd power is not "better" than Garmin's, but the Stryd's platform uses power for predictions so you might get better results that way. The existence of a power-duration curve with Stryd is very useful per se, for training planning.

I have Styrd, but I stopped using it a few years ago. It's reassuring to hear that you think Garmin's power is comparable to Stryd's.

I commented on this thread because I wanted to share my experience with Garmin's race predictor. I don't need it because I have a better work around. For every race that I do now, whatever the distance, I calculate an index for the course. It's purely applicable to me and covers all aspects: conditions, weather, terrain etc. The index is speed/power. The higher it is, the faster the course was for me on the day. I am looking at getting the most speed for less power. With that, I can then apply my current power (say I just did a windy and hilly race, put in a lot of effort, but the time was disappointing). Therefore my potential speed is index (for another course) x recent power. With the speed, I can calculate my time. I have it all on a spreadsheet. ;) 

Etupes25 said:

The point was to reiterate how Firstbeat is estimated ventilation (HR, HRV). Under exercise, the ventilation alreay reflects the weight, age, gender influences.

Exactly. Weight also influences speed, and since we know the VO2 Max prediction algorithm primarily measures speed vs HR, I think that using weight as a direct input would be a form of "double-counting".

Again I will point out that if weight is used for VO2 Max prediction, then if I lose weight and stay at the same speed at the same HR (for whatever reason), then, paradoxically, my predicted VO2 Max will increase even though I can't actually run faster.

(I realize that there are other influences on speed, such as running form.)

Etupes25 said:

Most of the VO2 Max estimation formulas i have seen are expressed as relative so that they can be used by any individual. At the same time, I have read papers where the unit chosen was the absolute one to establish relationships between ventilation and blood lactate concentration for example.

Etupes25 said:

Thinking of it, the topic of units for the VO2 Max is a bit of a red herring in our discussion since weight is known data. 

I don't think it's a red herring if the goal is to determine whether weight is used for VO2 Max prediction or not. I want to know because I'm interested in how the algorithm actually works.

However, it seems that short of getting an answer directly from Garmin or FirstBeat, we'll never know, unless someone wants to do the experiment I suggested.

Another experiment which could be done without two watches would be to train normally with your correct weight entered and obtain a VO2 Max estimation. Now double your weight on your Garmin watch and train normally. If weight is used for VO2 Max estimation, then all other things being equal, your VO2 Max (and race predictions) should go through the roof.

There used to be a FirstBeat employee who would answer questions in these forums, but I was never able to get a clear (to me) answer on the weight issue.