TSS / BikeStress inflated when not using Garmins auto pause feature?

[DavidU]

Hello all,

recently I disabled the auto pause feature on my Garmin 810 so that it still records data points even if I stop riding. This seems to inflate the TSS/BikeStress metric in comparison to using auto pause on the same ride.

An example:

Let's assume the following ride:

• 60min @ 200W
• 30min @ 0W
• 60min @ 200W
  
• 2min @ 0W

Let's also assume auto pause would stop the 0W parts after 3 seconds because we're not moving anymore and the rider has a FTP of 300W.

This results in the following metrics:

Auto pause enabled:

• IsoPower 200W
• BikeIntensity 0.667
• BikeStress 89 (time: 120min 6sec)

Auto pause disabled:

• IsoPower 188W
• BikeIntensity 0.627
• BikeStress 100 (time: 152min 0sec)

So, which version is the "truth"?

When describing the TSS formular Dr. Coggan just vaguely mentions s as duration in seconds (see Training and Racing with a Power Meter, p.124). It is not clear whether this refers to the time moving or the total elapsed time. Intuitively that would refer to the time moving, not the total elapsed time because the latter includes non-active times that provide no training stress whatsoever. However, GoldenCheetah, TrainingPeaks and Garmin all seem to consistently use the total elapsed time as they calculate the same TSS when auto pause is disabled.

Imho, for computing these metrics GoldenCheetah should prune data points where the athlete was neither moving nor exerting power. This would lead to consistent results no matter if auto pause was enabled/disabled.

Cheers,

David

[Mark Liversedge]

If you are using garmin smart recording, enable it in preferences.

We recommend to avoid it altogether.

Mark

[Chris Cleeland]

On Thu, Aug 16, 2018 at 8:47 AM DavidU <david...@gmail.com> wrote:

Imho, for computing these metrics GoldenCheetah should prune data points where the athlete was neither moving nor exerting power. This would lead to consistent results no matter if auto pause was enabled/disabled.

If you did that, then you would not be calculating it as intended.  Now, if you stop for a burrito and continue recording, that pause is likely long enough for physiological stress to "reset".  But stopping for a stoplight (which auto-pause will do) and ignoring coasting (which is very common in mass-start races) does not reset physiological stress.  If you calculate them that way, then you're not calculating TSS.  One can decide to argue whether those pauses are significant or not, but the calculation for TSS includes time on the bike whether exerting power or not.  If you change that, then you're no longer calculating a TSS value, and shouldn't call it as such.

[DavidU]

@Mark Nope, no smart recording. One second recording intervals, just with auto pausing disabled. Thus, the Garmin still records data when you come to a stop. This extra data inflates TSS in the way GoldenCheetah currently processes it. I get GoldenCheetahs policy of not fiddling with user data, but this leads to different TSS values for the very same ride depending on whether auto pause was enabled or not.

@Chris That's not the point. Without auto pause you currently gain TSS while stopping for said burrito. I think we can agree that this cannot be intended. Also, that was just a simplified example to demonstrate the problem. You can have a 6-8 hour ride with many small breaks and will notice the same effect.

[Mark Liversedge]

Use Tools > Fix Gaps in Recording, we don't support gaps in recording.

Mark

[DavidU]

Mark, there is no gaps in the recording. For clarification I attached a screenshot of some of the data in question. Please see the time, power and speed columns. I was standing still when power and speed both show zeros, it was not a device or sensor malfunction of any sort.

[Mark Liversedge]

Then how is the data different with and without auto-pause?

Mark

[DavidU]

With auto-pause:

* the garmin would not have recorded all these records that you see in line 6163-6195.

* More generally speaking it does not add records when you come to a halt.

Without auto-pause:

* the garmin still adds records when you stop riding, see line 6163-6195.

* this adds 0 Watt data records while you're standing still.

Currently GoldenCheetah processes all data records as they are. In other words it does not distinguish between riding-data and stopping-data. (see: https://github.com/GoldenCheetah/GoldenCheetah/blob/master/src/Metrics/Coggan.cpp#L53-L103)

Therefore, it uses all 0 Watt records that would not be in the data if auto-pause would have been enabled. It also counts those records as riding time. This leads to a lowered NP, consequently a lowered IF and finally a higher TSS (lower IF but longer riding time).

For further illustration you can watch the effect in action on my Garmin. They obviously seem to compute it in the same way: https://photos.app.goo.gl/8HeyjFaVhe3yjhr39 (please see the TSS, Time [Zeit], Distance [Distanz] fields).

[Mark Liversedge]

On Friday, 17 August 2018 12:29:04 UTC+1, DavidU wrote:

With auto-pause:

* the garmin would not have recorded all these records that you see in line 6163-6195.

* More generally speaking it does not add records when you come to a halt.

This is not supported in GC, use fix gaps in recording to add missing data, which will be zeros.

Without auto-pause:

* the garmin still adds records when you stop riding, see line 6163-6195.

* this adds 0 Watt data records while you're standing still.

Currently GoldenCheetah processes all data records as they are. In other words it does not distinguish between riding-data and stopping-data. (see: https://github.com/GoldenCheetah/GoldenCheetah/blob/master/src/Metrics/Coggan.cpp#L53-L103)

Therefore, it uses all 0 Watt records that would not be in the data if auto-pause would have been enabled. It also counts those records as riding time. This leads to a lowered NP, consequently a lowered IF and finally a higher TSS (lower IF but longer riding time).

GC assumes all data is contiguous with no gaps.

The fact that TSS is accumulated when you are not pedalling or stopped, and is not additive is something to pick up elsewhere.

It has been debated before in wattage for example.

Mark

[DavidU]

Am Freitag, 17. August 2018 15:04:55 UTC+2 schrieb Mark Liversedge:

GC assumes all data is contiguous with no gaps.

The fact that TSS is accumulated when you are not pedalling or stopped, and is not additive is something to pick up elsewhere.

It has been debated before in wattage for example.

Thanks, I'll try to look up the arguments for either side. Just seems counterintuitive to me that you'd gain TSS by waiting at an intersection, taking a leak or having a coffee break.

[Ale Martinez]

Search Wattage forum for RATSS (recovery-augmented TSS), lot of discussions about this.

[DavidU]

Thanks Ale and Mark, I see now what rabbit hole I started digging. You've been there and there is no generally accepted solution. Not what I was hoping for but at least comprehensible.