par mechgt

Téléchargements: 12,931 (46) • Reviews: 18

 Version: 2.0.7 Mise à jour: Aug 26, 2015

## Time constants for ATL and CTL

### Time constants for ATL and CTL

I was wondering what you think are reasonable time constants for ATL and CTL. So far I have been using the default values (11 and 45 respectively) but thought that especially ATL did not describe reality.

Using 11 for ATL means that it will take ~31 days to get to 5% of whatever level your at (e.g 100 -> 5). 31 days for full recovery seems a very long time to me. Setting it to 5 leads to ~13 days to get to 5%, which to me sounds more reasonable. What do you think?

Using 45 for CTL means that whatever shape your in is down to 5% after ~133 days. Is this reasonable?
RiverWinz

Messages: 2
Inscrit le: Lun Août 17, 2009 1:23 pm

I think that you are caught by some border effect of the math ...

CTL on a span of 45 days means just that the influence of the training older than 45 days is equal to 0.
ATl on a span of 11 days means that you need at 11 days to recover fully from fatigue.
I feel that these values are fairly consistant with my personnal feelings.

When looking at the graph, I'm more interested by the trends and the shape than from any particular values.

Remeber that training should decrease both CTL and ATL on the long run. the same amount of kilometers at the same speed should be easier.

So my advice is to avoid giving to much weight to the figures.
Running with :
acics kayano 18 - Sports tracker pro (android) - polar wearlink bluetooth
my dog : Ithaq (border collee) (Shimly - a happy mixing - retired in 2012)
Donated!

Messages: 124
Inscrit le: Mar Sep 30, 2008 1:56 am
Localisation: Belgium - Liege - 50°39N - 5°35'E

ildibad a écrit:Remeber that training should decrease both CTL and ATL on the long run. the same amount of kilometers at the same speed should be easier.

So my advice is to avoid giving to much weight to the figures.

Agreed, however your CTL and ATL will not decrease with equal training unless you re-test and establish a new higher FTP, I believe. Assuming you raise your FTP then you will have to increase your duration, intensity or both to achieve the same training effect.

That said, I don't even track ATL. My CTL is very slowly climbing and if I get to 100 I'll be surprised. Of course at that point, I'll need to do a field test which should bring my CTL back in line.
Fernando
racerfern
Donated!

Messages: 898
Inscrit le: Mar Juin 06, 2006 6:49 pm
Localisation: East Atlantic Beach, NY

ildibad a écrit:CTL on a span of 45 days means just that the influence of the training older than 45 days is equal to 0.
ATl on a span of 11 days means that you need at 11 days to recover fully from fatigue.

These statements are not entirely accurate. OP has the right idea on the math portion of it. These (CTL & ATL) are decaying values, and mathematically all past trainings will impact your CTL/ATL. The effect of really old activities is in fact very negligible, but it's not quite 0, just to be accurate.

Anyway, 5% sounds like a very arbitrary number to calibrate against. Any reason for this choice? (Not trying to pick, I'm always open to hearing new ideas )

The default numbers (11 & 45 days) were chosen based on Eric Bannister's (and others) TRIMP research, and specifically were found in one of the technical research papers listed here ("Modeling human performance in running"
I believe):

Note that technically everyone's constants are different, but in reality this (Training Load) is a mathematical model, not the law. You might tune it, but as Ildibad said, I wouldn't get too wrapped up in trying to make it "exactly right" (as I don't believe that exists). It uses a mathematical algorithm to model or guess, what we experience, it's not a precise measurement.
Enhance SportTracks with Training Load, Fit Plan and more plugins at mechgt.com. Garmin FR310XT & iBike iPro

mechgt
Donated!

Messages: 1370
Inscrit le: Mer Sep 26, 2007 2:13 pm
Localisation: Atlanta, GA, USA

mechgt a écrit:
Note that technically everyone's constants are different, but in reality this (Training Load) is a mathematical model, not the law. You might tune it, but as Ildibad said, I wouldn't get too wrapped up in trying to make it "exactly right" (as I don't believe that exists). It uses a mathematical algorithm to model or guess, what we experience, it's not a precise measurement.

And btw, what are we supposed to put as "Initial" ATL and CTL, and/or how to estimate those initial values?
Ruskie

Messages: 116
Inscrit le: Dim Mars 08, 2009 11:24 am

I use 42 for CTL and 7 for ATL. Some people use 11 for ATL.
Fernando
racerfern
Donated!

Messages: 898
Inscrit le: Mar Juin 06, 2006 6:49 pm
Localisation: East Atlantic Beach, NY

Yep, those are the values for the constants (I use 11 too, for running). But there are two other values, by default at 0, which are ATL and CTL "initial values". I wonder what they are for and how to estimate them...

Thanks
Marco
Ruskie

Messages: 116
Inscrit le: Dim Mars 08, 2009 11:24 am

Ruskie a écrit:But there are two other values, by default at 0, which are ATL and CTL "initial values". I wonder what they are for and how to estimate them...

They are used to extend your moving averages back before your history. Assuming you have been in training prior to ST recording, then they would be your estimate of that training. If no change in training then use your current values. Or if the ST usage start was well into the past, leave it a 0.

texmurphy
Donated!

Messages: 2072
Inscrit le: Mer Juil 05, 2006 7:38 pm
Localisation: Maryland, USA

texmurphy a écrit:
Ruskie a écrit:But there are two other values, by default at 0, which are ATL and CTL "initial values". I wonder what they are for and how to estimate them...

They are used to extend your moving averages back before your history. Assuming you have been in training prior to ST recording, then they would be your estimate of that training. If no change in training then use your current values. Or if the ST usage start was well into the past, leave it a 0.

Oh! Then they are like p0 in the trimp formula... well I have data starting early last year, I do not think those constants may make a big difference then...

Thank you!
Ruskie

Messages: 116
Inscrit le: Dim Mars 08, 2009 11:24 am

It seems to me that there are two ways to look at the CTL. As a resource to be run down, or as a fund to be built up. Given that generally we are not in the game of de-training I tend to the latter. And John Hellemans who is an outstanding NZ triathlete coach always worked on the basis that meaningful fitness changes took 8 weeks, so based on that if you are trying to assess the value of your current training load I would be tempted to go out to 56 days.
Keith Sherman
Garmin Edge 800
Specialized Roubaix SL3 '11
Giant Anthem '10

ksherman
Donated!

Messages: 223
Inscrit le: Dim Juil 11, 2010 10:00 pm
Localisation: Australia

### Re: Time constants for ATL and CTL

I think this has especially to do with the math border effect. However, I do not think that the value of CTL and ATL holds true forever as it is subjected to change. And since they are decaying values, equaling zero is for the most part impossible.
hello
vaajin

Messages: 4
Inscrit le: Ven Nov 11, 2011 4:06 am

### Re: Time constants for ATL and CTL

Hi,

I've been collecting experience with Training Load for a few months now. It's very interesting to see how various workouts add up differently.
Right now, I'm trying to be conservative and regard CTL as a better parameter to describe and add up my training efforts than "weekly mileage". I think it primarily does just that.

A few points, though.
I'm 48 years old and wonder whether there are any data or experience how age affects the ATL and CTL time constants. Clearly, I'm not recovering as fast as I did when I was 22...
It may also take me more time to "build up" CTL.
I'm using 12 and 45 right now.

Second, I have the feeling that when I do really hard intervals, like 5 x 1 km way below my current 10k race speed in anaerobic territory and reaching up to 95% of my heart rate reserve, it does give me relatively few TRIMP points - compared to let's say a relatively easy run at constant speed and maybe at 69% of heart rate reserve (that would be around 77% of maximal heart rate).

Do you have the same observations?

Thank you,

Michael
sunbeam

Messages: 2
Inscrit le: Dim Sep 21, 2008 7:49 am

### Re: Time constants for ATL and CTL

sunbeam a écrit:Hi,

I'm 48 years old and wonder whether there are any data or experience how age affects the ATL and CTL time constants. Clearly, I'm not recovering as fast as I did when I was 22...
It may also take me more time to "build up" CTL.
I'm using 12 and 45 right now.

Second, I have the feeling that when I do really hard intervals, like 5 x 1 km way below my current 10k race speed in anaerobic territory and reaching up to 95% of my heart rate reserve, it does give me relatively few TRIMP points - compared to let's say a relatively easy run at constant speed and maybe at 69% of heart rate reserve (that would be around 77% of maximal heart rate).

yes Michael..i have the same age 'problem' and question. I share EXACTLY your 5x 1km issue with TRIMP.

also does it apply to triathletes who do, say, only 2 lots of running 2 lots of swimming and 2 lots of biking each week compared to 6 lots of running for a runner?
also does it equally apply to IM training as say to someone training for one mile racing.

just wondering!

the5krunner

Messages: 365
Inscrit le: Dim Fév 22, 2009 1:41 pm

### Re: Time constants for ATL and CTL

Since TRIMP is calculated based on the score for each heart rate zone, you may edit the default score rating so that the score of higher HR zones (i.e. close to HR max) is non-proportionaly higher. For example:

Settings
settings.png (106.39 Kio) Consulté 4653 fois

If you plot the above score rating you see that the relationhip between score and %HR is non-linear after 160bpm.

Score chart
chart.png (13.23 Kio) Consulté 4653 fois

As a result, activities that are perfromed at a HR close to HRmax will get a much higer TRIMP value.
ghinop

Messages: 1
Inscrit le: Ven Déc 30, 2011 5:46 am

### Re: Time constants for ATL and CTL

If you gave the highest zone a factor of 1000 the TRIMP would be even higher...but it would not mean that it was 'correct'. So how did you arrive at the factors you use please?

Glad you made me check these though as for some reason a few of the factors were blank so I filled the gaps in!

the5krunner

Messages: 365
Inscrit le: Dim Fév 22, 2009 1:41 pm