Race Predictor is not updating after a race

A couple of things:

- your race is just one of many data points used by the race prediction model which uses your VO2 Max and your recent training history to do its thing. A new PB will not alone force a match.

- all predictions come with a standard error, and the standard error is just an average of errors. For a specific user, the error might be much higher, or lower. For example on average, the Vo2 Max estimate has a 3 to 5% error depending on the accuracy of the Max HR. But some users might see a higher error. In the case of race predictions, Garmin doesn't publish the standard error. If you look at Stryd, they have a standard error of about 3mn for the half-marathon. If you use Stryd's standard error, your estimate would be pretty good.

Sorry, but for me, this doesn’t make much sense. While I understand that race prediction is based on a complex, multi-factor algorithm, nothing is more accurate than actual race performance.It is the REAL time i made on this distance. This should take priority over other data. If you have any counterarguments, I’d be happy to hear them.

Even with a standard error of 3 minutes, I’m seeing nearly a 1-minute gap per split, so it's far from “pretty good.” Additionally, a 3-minute standard error in a half marathon means a 6-minute range between the low and high estimates, which equates to a 17s/km difference—quite significant.

9193893 said:

Sorry, but for me, this doesn’t make much sense.

I agree it is not intuitive, no question here.

9193893 said:

While I understand that race prediction is based on a complex, multi-factor algorithm, nothing is more accurate than actual race performance.It is the REAL time i made on this distance

But then again, why would you need a prediction for a distance where you have already done your best... For that distance, you would move on to getting shorter times, no prediction needed.

Vice versa, if you are trying to build your confidence in Garmin's predictions for other distance, you now have one data point about how optimistic/pessimistic Garmin's predictions can be in your case. It is up to you to decide whether you want to use this single data point to adjust in your mind the predictions for the other distances.

9193893 said:

This should take priority over other data

There is a separate UI to track your personal records by distance.

9193893 said:

Additionally, a 3-minute standard error in a half marathon means a 6-minute range between the low and high estimates, which equates to a 17s/km difference—quite significant.

I agree! Again, Garmin doesn't publish its standard error, Maybe it is higher (or lower) than Stryd's... I am just reminding you that any model comes with a standard error. And yes, a few minute on a HM is a big deal!

In my case, my race predictions became optimistic with the latest tweaks that Garmin did to their algorithm. They were just fine before. That said, I always use the training to test my target pace for the race and various other tools (magic mile from jeff galloway, Stryd) to have multiple estimates.