April 27, 2014

TSS vs. Suffer Score (or are they are complementary?)

Training Peaks software provides the TSS and Strava provides the Suffer Score as methods of determining how hard your ride was. The TSS uses intensity and duration of your power output (from a time series of the watts you generated during your ride). In addition to the Training Peaks software, Garmin power meters also calculate TSS for your ride. TSS and the resulting statistics in Training Peaks provide a lot of info on whether you need a break or you are slacking off in your training. But you need TSS to plug into the software. In the past I was simply guesstimating.

The Suffer Score uses the intensity and duration of your heart output (from a time series of your heart rate during your ride). You need the premium membership to get the Strava Suffer Score. But you get it for every ride where you wear a heart rate monitor. Downside - your followers get to see how high your heart rate was. Get over it.

I have a powermeter for my road bike, but I don't have one for my mountain bike. I could spend a bunch of money to buy one, but I thought I would determine if there is a relationship between TSS and Suffer Score and use that instead. So I fired up Excel, typed in both scores for a bunch of recent road rides and voila! A pretty tight relationship.


The relationship does spread a bit at higher Suffer Scores, but that isn't unexpected. The scores for those rides are a combination of long lower intensity rides and shorter higher intensity rides. Or a combination of both in the same ride. There are recovery rides, sufferfest trainer intervals, outdoor intervals, everything really. The biggest outlier is for storming up Mount Lemmon for 2 hours and then coasting down, stopping for photos, etc. on the way down. There is also a 100 miler in the mix which really had no hard efforts longer than a few minutes. And there are rides from 40 minutes to 6 hours in duration. All in all, given the diversity of rides, I think that is a pretty good relationship (the Suffer Score explains 89% of the variation in TSS). For my mountain bike rides I can now simply obtain the Suffer Score from Strava and use the formula shown on the graph to calculate TSS for my mountain bike. Yay and a real money saver. Y is TSS and X is the Suffer Score by the way.

This relationship will vary from person to person as your heart rate versus power output will also vary. So sad to say, you or a spreadsheet-savvy friend will have to determine your personalized relationship (between TSS and Suffer Score) for best results.

But wait, there's more! Even for an individual, as you get in better and better (or worse and worse) shape, the relationship will change. Some might see that change in the relationship as a problem, but it is really an opportunity to assess how your training (or lack thereof) is going. I plotted the same graph using different series for different time periods and voila!



More colorful for sure, but that's not the point. Sure you can masochistically do fitness tests periodically and see if your fitness is improving. But that's really hard and it assumes that each test is done when the old body is equally rested and capable. Why bother when you can simply plot your everyday rides and find out the same thing? Well maybe to reset your training zones, but really do they change that much? Mine don't change enough to change my targets any. I try to get well into a target zone and not just barely beat the zone boundaries. Don't forget its all one continuum of effort, not compartments. With this method you can use any time periods for comparison, whether you did an FTP test or not. And the comparison allows you to look at the complete range of rides and efforts you do at once - recovery to intervals to endurance. Nice. Perhaps even a training metrics revolution of sorts?

Check out that relationship for April. Suffer Score explains over 96% of the variation in TSS. Almost all other previous rides in 2014 fall below the regression line for April. In April I am getting higher power intensity and duration (TSS score) on average for the same heart rate intensity and duration (Suffer Score). February was the worst and March was intermediate (ignore the Mount Lemmon outlier). So pretty happy about that. This jives with my perception of my fitness sitting in the saddle. It would be interesting to compare years as well, but I'm not curious enough to do that yet. Sometime maybe.

Hopefully this gives you one more (free!) reason to go for a ride. And use the bikes that don't have powermeters too as you now have an accurate handle on the TSS for your training log.

Note to Strava - Suffer score proves to be every bit as useful as TSS. It would be easy for Strava to also mimic the other concepts in Training Peaks and show people how their fitness and freshness are doing....

May 15, 2014 Update - After using this system for a while, I have found that the relationship changes at very high intensities - races and race pace workouts. Like the Mount Lemmon example, the Suffer Score is higher than the TSS relative to normal training rides. The farther to the right of the regression you get, the harder the workout. Note that these have to be really hard efforts at the limit for relatively long periods of time. The kinds of rides you normally cannot achieve in training because you feel so sorry for yourself. So far it is the longer races (2-5 hours) that show this characteristic. But I have yet to see what a 20 min time trial does for example. Soon.