Using TrImp to calculate CTL, ATL and TSB
- Part 2
Back in this post I was wondering how to use TrImp for more than just setting a target goal score for the week. Well today I'm gonna tell you how I did that.
If this is the first time you've heard of TrImp (Training Impulse), then I suggest you read some past postings:
First post on TrImp
Second post
Using TrImp to calculate CTL, ATL and TSB - Part 1
There's more that I've posted about TrImp and to get to it I suggest using the search box on the right. Go ahead, type in "trimp", hit the search button and you should get a list of results that will occupy hours of time wasting.
Now that you're all cought up, here's a chart of the year so far as daily TrImp values, ATL, CTL and TSB goes. (click it to see it bigger)

Sometimes viewing the whole year to date chart makes it hard to see what's really going on, so here is a chart showing some of the most recent weeks. (again clicky to see biggie)

With these you can see that this past weekend (Memorial day weekend), I really didn't get out and ride at all. It's the flat TrImp line before the slight spike up for yesterday. Really what you can see is how missing a day or so here and there doesn't affect the CTL (chronic training load) as much as it affects ATL (acute training load) and TSB (training stress balance). You want CTL to continue to rise as the season progresses. ATL will jump up and down more, but should show an upward trend. TSB will also jump up and down, but you want it to rise prior to target events, and expect it to go down after them or during times of intense training.
Now here's how to use TrImp scores and Excel (or whatever spreadsheet program you choose) to get ATL, CTL and TSB. Well at least this is how I did it. I use SportTracks with the TrImp plugin to give me the TrImp scores for the day. I take that score and enter it into a spreadsheet. (again- click for big. you may want to open it in a new window so you can refer back to it.)

You can set up your spreadsheet with whatever other info you want, but you need a column for TrImp scores. Then next to it make a column for ATL, CTL and TSB. Now it helps to understand what that ATL is a 7 day rolling average, CTL is a 42 day rolling average, and TSB is the difference between the two.
In my spreadsheet the Trimp score column is "G", ATL is "H", CTL is "I" and TSB is "J." It's easier to start with the most recent entry. In mine it's row "150." So in cell "H150" you want to enter the formula "=AVERAGE(G144:G150)" without the quotes. This makes an average of the current entry through the past 7 days. Then select the cell and extend the formula all the way up the column.
Now you can just stop at the seventh day of your training log. Or you can just adjust the formula for those begining days so that there is some sort of average to chart. This is how I handled it. My Trimp score date begins in cell "G3", so in cell "H9" I enter the formula "=AVERAGE($G$3:G9)" and drag it up the rest of the way up the ATL column.
Now you have to do the same thing for the CTL column, except you need to go back 42 days for the rolling average. In cell "I150" I used the forumla "=AVERAGE(G108:G150)". I extended it all the way up to "I46". Then in cell "I45" I used formula "=AVERAGE($G$3:G45)" and extended it up the rest of the way.
Finally, since TSB is just the difference between the two, in "J150" I used the formula "=I150-H150" and extended it up to the top.
That's it. Now you can use the chart wizard to create pretty charts, and pretend you know what they mean like I do.
And I just wasted my lunch break writing this post instead of going out for a ride. Now I'm grouchy and mad at you. Oh well...
Click here for Part 1.
Click here for Part 3.
ATLATLAcute+training+loadAcute+training+loadCTLCTLChronic+training+loadChronic+training+loadTSBTSBTraining+Stress+BalanceTraining+Stress+BalanceTraining+ImpulseTraining+ImpulseChartChartExcelExcelSpreadsheetSpreadsheetcyclingcyclingbikebikebicyclingbicyclingGunnarGunnarCoachingCoachingetcetcSoroosSoroosBeerBeerScoreScore
Wednesday 05/28/2008 | 01:38 pm
Matt Rossing on Thursday 06/05/2008 | 12:00 pm
There is one small but significant difference between what you are doing and what wko+ does. unfortunately it makes the formula in excel much much more complex as far as i can tell, although i have very little excel experience. your ATL formula gives the same weight to your ride 7 days ago as it does for today's ride. the same for CTL--the ride 42 days ago has the same impact as 1 or 2 days ago. the model used by wko+, and by the guy who first designed the TSB model uses a weighted rolling average--giving more weight to more recent workouts. the link with the formula is here:
http://www.cyclingpeakssoftware.com/power411/performancemanagerscience.asp
unfortunately i have no idea, except by typing out 42 days worth of exponential functions and adding them up, how to do this on excel. I am sure there is a way, but i haven't done this kind of algebra in years and i only spent about twenty minutes trying to figure it out. I spent a few emails asking the trainingpeaks people about this and eventually got a response from hunter allen, their ceo, that they've tried to incorporate it into their software but never had adequate results when they've tested it.
If i do figure out the formula, i will try to let you know.
Matt Rossing
Gunnar on Thursday 06/05/2008 | 12:48 pm
Wow, thanks Matt. Now that you explain it, I sorta understand what they mean by a weighted average. I think I've tried to read the page you linked to, but it just goes way over my head. Please do let me know if you figure out the formula, as while what I'm using seems to work for my purpose, it sure would be nice to be more accurate to the real model.
Matt Rossing on Thursday 06/05/2008 | 06:12 pm
yeah, i had to read it a few times to understand it too. basically the longer it has been since a particular effort, the less effect it has on today's performance. i like to think of CTL as describing "fitness" and ATL as describing "fatigue," so yesterday's ride will have a bigger impact on your "fatigue" than the ride a week ago, and for fitness it goes much further back in time.












Write a comment