Help seeding CTL



marcovelo125

New Member
Feb 1, 2012
12
0
0
I could use a bit of help thinking about whether/ how to seed my ctl. Here's the situation:

I've been using a power meter for several years now. I got a late start training this year: began only Jan 1. more or less. A couple of months no training before that.

With no seeding of CTL after 4 weeks training my current CTL is ~40 and my TSB is ~ -60 (which doesn't correspond to how I feel (based on past experience): more like -30 at the worst.). I'm currently doing around 450-500 TSS pts/wk and tolerating it pretty well. Tired but not shattered.

It could be that I underestimated FTP a bit, which would skew the other values, but I doubt I'm off by much more than 10 watts, though. (I'll do a FTP test at the end of this recovery week.)

On the one hand, because I wasn't training before Jan 1, I feel there is no reason to seed a CTL value higher than 0. On the other hand, starting from 0, ATL, CTL, TSB and rate of CTL increase are looking a little unusual and don't seem to correspond to how I feel (e.g. -60 TSB).

Mostly what I'm interested in is rate of CTL increase. Because of my late start I want to push that as much as prudent. Starting from 0 CTL, the rate of increase looks a bit high but not outrageous: ~ 7-8 pts/week. Normally that's a lot, but I thought I could start out a bit hot as long as I felt okay and dial it down as necessary. If I seed the CTL value (according to average hours or average TSS), it looks practically flat, and would seem to suggest I need to train more, but I don't feel as though I can tolerate enough more TSS to get it up to a reasonable rate of increase.

One of my questions is: what are the long term statistical consequences of seeding or not seeding? Will either one skew the data significantly for a significant period of time such that it makes the values unreliable?

Should I stick with the 0 CTL or would it be wiser to pick a CTL value that generates a TSB which corresponds to perceived fatigue? Does any of this matter that much or will it even out in a couple of months?

Are there considerations I've missed?

Thanks for checking my thinking,

-marco
 
Originally Posted by marcovelo125 .

I've been using a power meter for several years now. I got a late start training this year: began only Jan 1. more or less. A couple of months no training before that.
Since you have all historical power data and no need to guesstimate actual prior training stress / training loads, then there is no requirement for seeding the performance manager. Seeding is for when you have no prior data to reasonably represent your current ATL/CTL (i.e. you have been training but only just started using a power meter and power analysis software).

Accept that when starting out after a long break, and having had plenty of solid training experience in seasons prior, that you are capable of sustaining a higher rate CTL increase in the early weeks, but that rate will likely need to drop a bit as CTL rises.

Also accept that it might take a couple of months for the TSB numbers to match sensations (it might not though).
 
I wouldn't seed CTL as it's only an average over the past 42 days. Right? It will catch up with itself quickly.

Am I off here?
 
Originally Posted by awilki01 .

I wouldn't seed CTL as it's only an average over the past 42 days. Right? It will catch up with itself quickly.

Am I off here?

It (as with ATL) is an exponentially weighted moving average that accounts for all training stress dating back to the Big Bang. It's just the contribution of stress scores from rides in the distant past are subsequently weighted significantly less in the averaging such that they effectively provide a zero contribution.

Hence a ride from 43 days ago is weighted slightly less than one done 42 days ago, but it is still included in the CTL calculation.
 
Originally Posted by Alex Simmons .



It (as with ATL) is an exponentially weighted moving average that accounts for all training stress dating back to the Big Bang. It's just the contribution of stress scores from rides in the distant past are subsequently weighted significantly less in the averaging such that they effectively provide a zero contribution.

Hence a ride from 43 days ago is weighted slightly less than one done 42 days ago, but it is still included in the CTL calculation.
I just reread some information (albeit not in great detail). 42 days is a time constant used in the CTL calculation. As you get further and further out, the CTL curve is less affected by prior training. I would say it becomes insignificant. I just created a new chart in WKO+ that goes out for the rest of the year. If I stop training right now with my current CTL of 55, it falls to 0 towards the end of July. No comments on my meager 55 CTL please /img/vbsmilies/smilies/smile.gif.

Now, I realize that WKO+ does not fully implement the entire formula for Banister's Impulse-Response model and has it's own set of limitations, but I think it is probably the best we have now. I don't know the exact formula WKO+ uses, but I remember back in the Navy as an electrician, we calculated inductance and capacitive charge/discharge rates over 5 time constants.

Here is a good read, if interested:
http://home.trainingpeaks.com/articles/cycling/the-science-of-the-performance-manager.aspx

One thing the article points out are some limitations. One that caught my eye is the following quote:

While the impulse-response model can be used to accurately describe changes in performance over time, it has not been possible to link the structure of the model to specific, training-induced physiological events relevant to fatigue and adaptation, e.g., glycogen resynthesis, mitochondrial biogenesis.
To me, this says that for a given CTL value, my performance will be better this year than last year. I've seen this at a great degree since I'm new to the sport (1.5 years). I'm already exceeding the performance I had last year even though my CTL is lower now than it was then. That is because of what I believe to be the beneficial physical adaptations that have taken place in my body as a result of training stress. As the years pass, however, that increase will start to slow dramatically. I think I'm witness to this now because I'm still very new to the sport and have quite a bit of headroom for more physical adaptations. But, one day, I will hit a genetic limit and will need to find other ways to increase performance through more research and better peaking/tapering methods.

I know I've totally thrown us off into the techno-weeds here.....