Activity stress does not match workout calculated stress

Website / Web-App
#1

I have noticed that recorded activity stress level saved in TD is always higher than the workout editor in TD (e.g. 60 stress for a workout calculated at 45; 90 for a workout calculated at 60 stress, even though I quit the workout 10 minutes early). Even stranger, the TD activity intervals all show average watts 15-20 points below the workout interval target.

When the activity is sent to Garmin Connect from my bike computer, GC TSS matches pretty closely the stress calculated by the workout in TD.

I don’t have an HRM connected to my TD device, so it can’t really take that into account.

Is there something different about the way activities calculate stress compared to the workout editor?

#2

Hi Jim, that sounds strange. Can you provide screenshots? I just did a comparison in a couple workouts and it looks fine to me. See here

image
image2186×506 144 KB

#3

Hi Alex. Here are my 2 recent ones:

Activity shows 65 and workout shows 50. The 30 second intervals are set to 105%, 315w - but the averages below the activity show 294 to 300 - I thought they felt a little too easy :slight_smile:

end-activity
end-activity1455×1197 96.1 KB
end-workout
end-workout1462×630 38.9 KB

This one is where I skipped the last 10 minutes or so and went into cooldown - activity shows 88 and workout as planned shows 61:

tempo-activity
tempo-activity1445×1200 94.1 KB
tempo-workout
tempo-workout1453×650 37.9 KB

#4

Hmm - I just noticed the intensity scores. For the first activity it’s .1 higher than the workout. And the second one is .25 higher. Could those wattage spikes be causing issues with activity stress calculation?

#5

Yes for sure it is the wattage spikes . I am using Wahoo trainer that has built in power smoothing. It would be interesting to load your workouts into TrainingPeaks to see what they report for total TSS. They might have built in smoothing to their algorythm also.

#6

That could be why GC shows expected stress - if they use an extra smoothing algorithm. I had chart smoothing set to 2 seconds for these 2 rides. I’ll try bumping it up to 7 or so for today’s and let you know.

#7

Our chart smoothing does not affect the stress. We send raw original data every where so that won’t help. I will try to research and see if other platforms use smoothing. I know TP says they have some “magic” error correction.

#8

OK so today’s ride had a better result. I went into settings and found that power match had somehow been turned back on - I originally set it off. Not sure if that makes a difference either, but it also closely matches the Garmin results. Here’s the charts - stress and IF are both acceptably close and the power spikes don’t appear as drastic:

tempo-activity2
tempo-activity21447×1023 80.8 KB
tempo-workout2
tempo-workout21458×552 28.2 KB

The 30 second intervals still average 15-20 points below the target.

#9

Oh yes that looks better. Yeah power match is something like TrainerRoad did, we need a 2.0 version some day. It’s very complicated to get it perfect. The simple solution is obviously just increase those higher intervals a bit… :slight_smile: