Race predictions way off

In my experience, garmins race predictor algorithm is pretty simplistic and only takes vo2max into account.

And vo2max calculation os also very simplistic. If you do a lot of high intensity intervals, your vo2max rises and if you do a lot of steady endurance runs your vo2max drops.

I know, but that still does not answer my question. Why would it be different from, and worse than, an older model?

If I compare my old forerunner 935 with my new 965 I see the same race predictions for the same vo2max. I don’t see any differences in the algorithm.

Has your VO2 Max changed? I found that my old 935 was far more generous with my VO2 max prediction compared to my 935.

Yes, the 965 shows a lower VO2max than the 245. But that just shifts the problem. The question is still why the new 965 shows worse predictions (or V02max estimates, so you will) than the old 245.

That really depends on how you train, like I said, in my experience intervals, were you get your heartrate up and have rests in between to let your heartrate drop again, increases your vo2max. Steady runs, with hardly any fluctuating of heartrate, decreases vo2max.

Do a coopertest or go to a sport medical center to have a better way to determine your vo2max and kinda ignore garmin algorithms. They are more marketing than science.

In my view - and recollection - the calculations were changed on the FW a while back (More than a year ago at least on the 955 if I recall - can't see it on the graphs now as they only go out a year) and the method of reflecting VO2 also changed. All my times predicted got slower but over time have improved to very close to my actuals now. The VO2 max also moves much slower now (which is likely more logical than on the older system) and also the fitness age is limited to current age -9.5 years , before it had a much wider range , mine was about 30 years less than my actual age as an example

peterdb said:

Why would it be different from, and worse than, an older model?

I think the algorithm definitely changed since the older models. I can attest that it changed between 935 and 945 LTE/955.

In the past (e.g. 935), there was a 1:1 correlation with VO2 Max and race predictions, in the sense that if two people had the same VO2 Max, they were both guaranteed to have the same race predictions for all distances. People actually reverse-engineered lookup tables going from Garmin VO2 Max to Garmin race predictions and posted them in the forums. (Not too different from publicly available VO2 Max / VDOT lookup tables for race results.)

With newer models (e.g. 945 LTE/955), it seems that mileage is taken into account, even for race distances as short as 5k (although the relative weighting of mileage vs VO2 Max seems to be higher for longer distances, such as marathons.)

For example, if you don't have enough mileage, then your marathon prediction may not "line up" with other predictions. On a personal note, I stopped running for two weeks, and although my VO2 Max didn't decline (since the watch had no new running data during that time), my race predictions all got slower.

https://www.firstbeatanalytics.com/en/features/race-time-predictor/

How Does It Work?

Your cardiorespiratory fitness level (VO2max) offers good insight into your ability to maintain a certain level of performance over a period of time. VO2max provides a strong basis for predicting how you will perform in in an endurance event, but your VO2max isn’t the only thing that matters. The work you do to prepare should be considered too.

To improve the reliability of your predicted finish times, Firstbeat Analytics takes a number of additional considerations into account. These include long- and short-term trends in your weekly running mileage totals. Similarly, Training Load trends are factored into the analysis, along with specific details like how longer runs contributed to your overall load.

Elbo said:

Do a coopertest or go to a sport medical center to have a better way to determine your vo2max and kinda ignore garmin algorithms. They are more marketing than science.

I typically disregard the absolute numbers, but I think the trends are useful.