For those of you who use TrainingPeaks WKO+ software, it's a great tool to analyze data and train efficiently (and if you don't use it, you should!). I'm not a long time user or fan of TSS (training stress scores), but have recently adopted using them after I got a power meter in November. I'm finding that for the bike, it has been quite useful and I'm growing in support of the idea. As I started to really embrace it for the bike, I found that I longingly wished for it on the swim and run. While I waited for the latest version of WKO+ to include a running TSS (rTSS), I made my own spreadsheet that calculated it in the interim. It worked well. But now, WKO+ does have rTSS (still no swim yet), which is supposed to make my life easier since I don't have to calculate it on my own, right? Well, I can't get over a hump that I see as a fundamental flaw in how it is currently designed to calculate in WKO+.
I don't really have any particular issues with the bike TSS. In short, it uses the quotient of your normalized power (NP) output from your functional threshold power (FTP) to calculate an intensity factor (IF), where IF = FTP/NP. FTP is defined around a 40K TT, so they use a general time of an hour. TSS is then calculated for each workout using the formula (workout duration * IF^2 * 100). The "100" standardizes the score to 1 hour. For the bike, I'm okay with this. Really fast guys may do a 40K TT in 55 min, and slower guys at 1:05, but that's still pretty close to an hour. Probably not enough to drastically change TSS one way or another, or worse yet, change training zones (based on FTP) to less efficient ones.
But I do have a problem with the run TSS. Instead of using FTP to calculate IF, it uses Normalized Graded Pace and Threshold Pace to calculate IF. I'm okay with this part...it's the same concept, just a different metric. But the formula is identical to the bike TSS from what I can tell: duration * IF^2 * 100. In my opinion, it shouldn't be standardized to an hour! Here's why: in the bike scenario, fast/slow people don't differ much from an hour for 40K TT. In a running scnerio, using a 10K as the surrogate for the 40K TT, there is a huge difference between fast/slow people, with faster runners being in the low 30-minute times and slower being near an hour. So if you are a faster runner and use your 10K time as Threshold Pace, the current formula of rTSS in WKO+ severely underestimates your stress for the workout. If you wanted to solve for this, all you have to do is change your Threshold Pace to be your 1-hr run pace. However, there are problems with that, too. Since running/riding near threshold is *supposed* in target LT the most, running at a slower pace for intervals doesn't seem to be as effective at improving LT than it would be to run near 10K pace for the intervals. Entering a Threshold Pace based on a 1-hour run completely changes the training zones to ineffective ranges. Entering your 10K pace underestimates your rTSS, inhibiting your potential to maximize training/racing.
It seems that rTSS is a good idea in principle, but it was designed by people that think like cyclists, not like a runner. In my opinion, the way to solve this is to change the "100" in the formula to calculate rTSS. Instead of a fixed 100 value (= 1 hour), it needs to adjust to a specific 10K time. This would be very easy to do. Just replace the "100" in the above formula for rTSS with "(60 / (threshold pace * 6.2)) * 100" if you use min./mile or "(60 / (threshold pace * 10)) * 100" if you use min./km. The resulting formula for min/mile would be "rTSS = duration * IF^2 * ((60 / (threshold pace * 6.2)) * 100)." Sure, I could do this in a spreadsheet and then override the existing values in WKO+, but then what's the point of having the software calculate rTSS in the first place?