TSS Flaws? (Race data inside)


#1

Ok, knowledgeable people… I’m officially calling out TSS as BS :thinking:

I’m attaching two power graphs below as visual aides. For reference, my FTP is 280.

Graph #1 is from a prep ride I did 8 days ago. It was a ride that included many stoplights, stop signs, was mostly meant to be a sweet spot and relatively intense NP ride, but not meant to gas me. After this ride I felt refreshed-yet-taxed and like I’d put in a good, solid effort. The stats for graph #1 are:

  • NP: 260
  • IF: 0.93
  • Duration: 1:59
  • TSS: 168

Graph #1:

Graph #2 is from a 33 mile gravel race I did yesterday. From the start, it was a hammer fest. My goal was to stay with the ‘A’ selection for the entire race, and 2 miles into the race there was a ~2 min, 13% grade “selection”. Then I was caught behind many riders who made it over the hill, but were blowing up and dropping from the group right afterward – so I was forced to pull around them into the wind and bridge to the main peloton. In between all of this there was hill-after-hill-after-hill. In fact, for the first 10 miles of this race, all that was going through my head was, “I f#cking hate riding bikes. This is dumb. I’m never going to ride a motherf#cking bike again” – I mean, I was completely pinned. Red lined. Call it what you will. I was on my limit. Then I settled in for the middle of the race and found my rhythm in the pack, and then the final 8 miles were full gas yet again. Today I am sore in places I didn’t know existed. I feel like I was in a boxing match.

The stats for graph #2 are:

  • NP: 254
  • IF: 0.91
  • Duration: 1:42
  • TSS: 141

Graph #2:

Now . . . look at those power files. There is NO WAY you can tell me that #1 has 27 more TSS points than #2. No. Way.

Which ride looks easier to you? With all the data and metrics at our disposal, there has got to be a better way to quantify training stress.

Based on RPE, ride #2 was 5x harder than ride #1. I’m wasted today. After ride #1 I was mildly fatigued.

Yes, ride #1 was 17 minutes longer…but there is no way that accounts for the difference. Right?

Anyone else struggle with this?


#2

You’ve struck upon the root cause of the saying ‘not all TSS is created equal’.

The TSS formula is an effort to approximate the impact of various intensities into a single quantifiable number. It is not perfect but it is, as far as I’m aware, the best tool we have access to thus far.

Most riders find that lower percentages of their FTP generate TSS in a less taxing manner than higher percentages of their FTP. That is to say - if you earn 10 TSS at 130% FTP you will feel the impact of that more than if you earn 10 TSS at 50% FTP even though you spent significantly more time on your bike earning the 10 TSS at 50% FTP.


#3

Whilst TSS taken into account the relative difficulties of efforts above and below FTP, it doesn’t take account of your abilities above and below FTP. It makes the same assumption for a good TT racer, who has very little ability to generate power above FTP, as it does for someone who can attack on short steep climb all day. If the TT racer tried to keep up with the punchy rider the effort would soon have them off the back even although the TSS would be exactly the same.

You can see the effect of this if you go out and do a session with 6 x 30 second all out efforts with five minutes rest between. A punchy rider would get much more TSS and a higher IF than the weeker rider if they have the same FTP even though in terms of RPE, both riders will have done an equal effort.

The training platform XERT tries to get around this problem but I’m not sure how they work out their score.

Mike


#4

Post up your time in zones, and I think you’ll open the window to how different the two rides are.

And then I’ll also have to quote “Not all TSS is created equal” Good metric for steady state efforts, but say for my MTB rides, TSS is significantly lower than just about anything I do per unit time, but there is so much more need for recovery.


#5

Would it be too obvious to point out that with a shorter duration and lower average intensity with ride #2, a lower TSS is not surprising?

However, looking carefully at the graphs, ride #1 shows a lot of sustained riding around FTP, whereas ride #2 features a lot more sharp spikes to 200% of FTP, as well as some sustained riding in the middle below FTP. Especially if you haven’t been training your anaerobic system, ride #2 may well feel harder. Knocking the tops off those peaks (keeping them to say, 150% of FTP) won’t make a huge difference to TSS, but will make a difference to how you feel.


#6

The problem is more likely with the NP formula than the TSS formula in this situation, as your RPE for ride 2 would suggest a much higher IF vs ride 1.

However, do also suspect the TSS formula has flaws also.

It would be an interesting study to update these formulas… I’m sure now with power data ubiquitously available, there is a much bigger data set to work with than when the formulas were originally developed.


#7

Interesting point, but I will say that I ramped to this race for the last 6 weeks using “Short Power Build” (low volume supplemented by outside rides like Graph #1) – meaning, I’ve been doing nothing a lot of anaerobic and Vo2 system(s) training. This doesn’t mean I’m good at it, but I have been paying attention to it.

All of these takes are interesting. I do think the issue may be with NP math, vs. TSS math. That’s a worthwhile point.


#8

Totally agree, I’ve found TSS near-useless for MTB rides, the fatigue and needed recovery just never seems to line up for me (not just MTB vs road vs trainer, but among MTB rides).


#9

The difference in duration completely accounts for the difference in TSS. They have a similar NP, and are 84-85 TSS/hr.

The heart rate suggests the latter was quite a bit harder.

Maybe look into time-in-zone or something that will estimate W’bal to see a difference. But I am surprised that the first ride has such a high IF.

Of course, TSS only measures one thing. If it perfectly tracked RPE, human performance would be a simpler system to understand.


#10

Any chance you could post up the two .fit files? I’m interested to have a look.

Mike


#11

I’m not sure how to do that…

I recorded one ride on a Hammerhead Karoo (graph #1) and the second ride on a Wahoo Elemnt Bolt. Both synced to Strava and Strava syncs to TrainerRoad, which is how these were generated.

If there is an easy way to do it, I’d be happy to.


#12

If you can’t post the files - would you mind linking to the strava rides?


#13

Sure. See below:

Graph #1: https://www.strava.com/activities/1919151786

Graph #2: https://www.strava.com/activities/1930364201

Would love to know what you think/find.


#14

In the workout page click the three circles and select download.

Mike


#15

There’s a pretty huge gap in the time in zone charts on those two rides that really explains why the second one was so much more impactful on your body. You spent a ton more time over 375 watts in the second ride - that’s what likely crushed you


#16

I guess that’s my point— shouldn’t the TSS formula account for that? (I know it doesn’t currently)

To accurately reflect true training “stress” I would think #2 would be more stressful by any measure.

Or am I thinking about TSS wrong? Is graph #1 truly just as, or more, stressful as graph #2? And I just don’t feel it as intensely?


#17

Thanks for the instructions. Didn’t know I could do that from TR.

It won’t let me upload the .fit files here on the forum, though.