That’s why I think the concept of PLs is inadequate / over simplified. It doesn’t really measure training volume per se, just time X intensity in one zone (and maybe an adjacent one too).
Mathematically someone on 6 hr/week can have the same PLs as someone on 20 hr/week. But we know who’s closer to their genetic potential assuming all other factors the same.
TR doesn’t give any indication of training volume in the concept of PLs. Sure RLGL considers whether to scale up or down your training intensity but it still is doing that relative to your current volume. There’s not really progression nor tracking of volume. Adaptive training doesn’t really consider at volume/CTL as far as I understand, other than to keep it steady. It’s basically looking at things on a session by session basis.
If instead it looked at volume and quality of training as separate concepts, then I think it would be a better approach for combining interval training and real world training/racing. A race or group ride is volume. And it may be as effective as or much less effective than an interval session of the same duration. Interval session or not, it might train a few very different areas of fitness (long threshold with attacks sprinkled in there).
If it were me, I’d create a system that broke the ride into loosely defined intervals and rate how close they were to good intervals, assigning them an effectiveness. That part is very tricky and is the perfect application of big data and machine learning. See how it really affected the athlete’s improvement when compared like to like against their other training and peers. You can even do that for TR indoor workouts to judge the effectiveness both of the workout itself and of the benefit to the individual athlete.
Do likewise for volume as well. Then you can show benefit for any workout and indicate to the athlete qualitatively whether they should focus more on training more effectively (intervals) or more volume, and what specific zones are more important to target right now.
Now, this isn’t going to be really precise as far as the expected impact of training. But we don’t have precision today anyways. It’s just “trust the process” and “this is science based” and “excellence tells us that…”. That’s OK but it’s also limited to keeping things fairly simple in order to be easily quantified (intervals, TIZ, TSS, CTL). Any big picture stuff is completely left up to human interpretation or requires fairly strict adherence to a plan.
With my approach I believe that it’s possible to quantify, albeit with larger error bars, how various races, group rides, commutes and just riding around affects training in the big picture way. It also becomes easier to see how an individual responds to different training, including what works for improving compliance to key sessions.
With this approach it becomes easier for adaptive training to adapt not only to moving around workouts but better counting and using non-interval sessions as part of the training program instead of just treating them only as bad (TSS without advancing your training progress, PLs).
This wouldn’t be trivial and it’s got to be done right. How to convince us highly analytical and skeptical toes without giving away the secret sauce is another complexity to this. Plus it could take a lot of resources to figure this out. But it would make a fantastic product.
If I were younger I’d be trying to make this product myself. Hopefully the long delayed WLV2 is something akin to this. That’s the only way it would be justified in taking so, so long to come out IMO.
