This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

Race time Prediction - NEW algorithm

Hi,

the new race time predictions semes to be hugely more accurate then the previous ones in previous devices. What factors the current agorithm take in account for the calculation ?

thanks

Danilo

  • The reviews I read stated that the new factor is previous milages, so it will scale down the long distance predictions if you did not put in the work. Previous versions only did a look-up based on the vo2max value.

  • Agreed, the times is shows compared to my 230 are much more in tune with reality, esp the longer distance runs (up to marathon)

  • Hi

    This is probably what you are looking for: https://www.firstbeat.com/en/consumer-feature/race-time-predictor/

    The short answer to your question is that 3 key elements were added. 

    1. Long- and short-term trends in running mileage. 

    2. Long- and short-term trends in your Training Load

    3. Performing "long runs" that would prepare you to actually run those longer race distances where endurance matters. 

    VO2max still plays a significant role in the prediction, but more as a baseline to begin the prediction rather than the sole generator as it had been previously. I think everyone can expect a vast improvement in the meaningfulness of their predicted finish times, and I'm more than happy to hear everyone's personal experiences - and how it measures up to real-world performances. 

  • I got the new 945 (been using it for slightly over one month) and my race time prediction is the worst of all of my Garmin device I have ever owned. The predictions are way to slow (for example 21min for 5k, did a trainingsession includong 5k sub 20min) and it gets worse the longer the distance gets(About 15min off at halfmarathon). Is my software broken or am i doing something wrong? (Filled in my one Heartrate zones, so these are accurate)

  • I have mine for a week and a half. Race prediction times started off way too slow. But after every run I do, they get adjusted and start getting closer to my actual times. 

    Offcourse its still settling, since it doesn't have enough data yet. Wonder how well it will turn out after a few weeks. 

    Vo2max readings at least seem to be better. They were allways low in the summer because of the heat. The 945 started off with the base value it got from my 3hr. And it's starting to rise after every run. It's almost at the same level I normally get in the winter already.

  • +1. Far more accurate. Only had mine a week. Did a 10k on Sunday and got within 10 seconds of the predicted time. Which is great considering it’s still learning about me. 

    My 10K prediction has come down by about 40 seconds and now is about 50 seconds faster than my PR(which was set on Sunday)

    interestingly my prediction for HM is a min slower than my PR!

    VO2 score is accurate too. 

  • Mixed results for me (VO2 max @ 61) > Predicted / Actual

    5K > 17:35 / 17:37 > this one is spot on !

    10K > 37:05 / 35:58 > 1'07' slower

    HM > 1:22:15 / 1:18:56 > 3:19 slower

    Heart Rate is the best I've seen in any watch until now. Smooth graph, very similar to my Tickr-X which is amazing for an optical sensor !

  • Mine are almost right on spot from 5k to marathon. I use Stryd, HRM-Run and 945 for months.

  • Interesting, mine are way off. 

    has my 5k @ 22:02 my fastest 5k was 24:30

    has my half at 1:45:52 my fastest half was 1:54 

    And it has my marathon at 3:58  my longest run was 20 miles @ 3:10 again not even close  

    that’s quite the spread. I wonder how the heck Garmin comes up with these numbers. 

    I had the watch for 5 months now. 

  • Garmin uses research papers that are shown to give good results on average. But nobody is average. We all deviate from it. Also, the race predictor represents your theoretical potential. It means what you could achieve, peaked and rested, when all the conditions are right, including the right pacing strategy so that you can squeeze out as much performance as possible.

    So it's 1) theoretical 2) a best case scenario