Andy,

Is there any way to limit the range of durations use to compute mFTP?

Occasionally, my powermeter will spit out a erroneous reading that completely skews the mFTP for all dates within 90 days of the error. For instance, on 10/21/2015, the data from my ride showed a one second reading of about 3,300 watts. The one second readings adjacent to this were both under 200 watts. However, mFTP for all dates within 90 days of this date jumped up about 30%. As a TT-type rider, I would like to eliminate erroneous readings from the computation of mFTP without having to review the data for every single ride for anomalies. This could be done, I hope, by being able to limit the range of durations to say, 15 seconds to 1.5 hours.

Mike

Unfortunately, that would not work, as every point on the mean maximal power curve has some influence on all of the model parameters. Excluding everything under 15 s would therefore alter your mFTP, even if only slightly.

Your best bet would be to clean up your data as described here:

http://help.trainingpeaks.com/...Fix-Bad-Data-in-WKO4

then try to stay on top of things going forward. (You might also think about getting a new powermeter. <g>)

I let will Andy respond to the science and why 90 days is the default and why you should use it but here is a help article and video on ways to find and fix the bad data (which is the better answer if just a couple)...

http://help.trainingpeaks.com/hc/en-us/articles/204895530-How-to-Fix-Bad-Data-in-WKO4







You can also review the mFTP history charts and see the expression formula.


So if you use:

ftp(meanmax(power))
That will give you your Power Duration curve based off any selected time range, output in watts (/weight if want w/kg)

If you use:
ftp(meanmax(power),90)
That will give you your daily historical mFTP over the time range selected BASED off rolling 90 days

Therefor (example):
FTP(meanmax(power),42)
That would give you your daily historical mFTP over the time range selected BASED off rolling 42 days (the 42 is an example, use what number you want).

A cool thing to do in the Power Duration Curve Pack in the child charts (ones in bottom window) is top open the PD METRICS HISTORY and look try the following:
1. Open Config
2. Click DATA SERIES - adding a new series
3. Name mFTP 42 (or what ever time you plan to give it)
4. Select square line and pick unique color
5. Enter the expression: FTP(meanmax(power),42)
6. Close and review

Now you can compare your mFTP based off rolling 90 days and rolling 42 days. Which compares closer to your performance history?

To go deeper, you can add that performance history to the same chart,since tracking FTP lets use 95% of peak 20 minute power to see if correlation:
1. Open Config
2. Click Data series - adding a new series
3. Name it Peak 10 ftp estimates
4. Set to circle, pick color
5. Expression: to get 20 min peaks for each day, we use: meanmax(power,1200) which means meanmax power for 20 minutes (1200 seconds)
6. This give a bunch of dots but cluttered, you might want to just see the greatest 10 performances for the time range so we need to add more to the formula
7. To try this expression (think like algebra, build like equations from inside out): greatest(meanmax(power,1200),10) this will show the greatest 10 - 20 minute performances
8. Now that you have that, that might be good enough but what if you are mentally doing the 95% math now? try...
9. Times the expression to 95%, so: greatest(meanmax(power,1200),10)*.95 That does the math for you.

Sorry, long answer to short question....

PS: save the chart :-)

OK ANDY BEAT ME TO POST

---

Tim Cusick
TrainingPeaks WKO4 Product Development Leader

As I'm sure you know, there are multiple ways to determine "threshold" based on changes in blood lactate. Conceptually, the one that is closest to FTP (in fact, what FTP is a surrogate measure for) would be maximal lactate steady state (MLSS). The latter can only really be estimated from data collected during a ramp test, but based on the literature the method that I would hypothesize would yield the value closest to FTP would be the Dmax approach:

http://www.trainingandracingwithapowermeter.com/...ional-threshold.html

http://www.trainingandracingwithapowermeter.com/...-threshold-pace.html

Thank you both. :-)

I am probably the only dinosaur who is still using Polar ProTrainer5-I find the calendar the easiest to see and find things on and yes it is probably because I have been using it for close to 15 yrs . In wko+3 there was nice little feature to convert my wko file to a .hrm file and then import it into Polar from wko+3. I can't see such a feature in wko4-is there anyway to still do this-other than keep running wko+3?????????????

Hi Andy,

For the last 4 weeks, mFTP has been creeping up as I am doing 4min efforts slightly over threshold (105%). I am doing them a bit harder, with little less recovery so best 30min is creeping up.
FRC is going down and currently at 7KJ.
Elsewhere I think you mentioned FRC is normally higher than W' which I am pretty sure mine is much higher than 7KJ.
I know in the CP model an underestimated W' would probably overstate your CP and currently mFTP is probably a bit high.

All my workouts are on the computrainer so very structured

Are things skewed and what type of work should I do to bring it back in order ? I am pretty sure my data is clean now, Pmax seems right, power duration curves make sense.

Maybe the answer is I need to have more maximal data for the short durations ? I have very little


On the topic of stamina/endurance metric. Is it along the lines of your fatigue profiling in your book ?

Is it possible to restrict a calendar-like report to show only bike activities? If I use 'if(sport="bike",duration)' I still see the non-bike activities in the list

Just use DURATION and pick the BIKE (chain) thing in the sport selection top right.


Try it this way. At the athlete level, build a new chart, add data series names duration and use expression; DURATION and close config box. Click through the sport types and it will change (assuming you have other sports).

---

Tim Cusick
TrainingPeaks WKO4 Product Development Leader

So there's no way to make the calendar reports in the cycling pack automatically filter on cycling only? Am I understanding that correctly? You can only do that by filtering on the chain in the top right?

I re-wrote the expressions in the cycling pack so that PMC is bike only. It looks like that only applied to the athlete I was on at the time. Is there a way to apply that change to all athletes?

The exact relationship between FRC and W' is variable, and obviously depends heavily upon the duration of the efforts used to calculate the latter.

That said, if you haven't done any maximal efforts at shorter durations, it is possible that your FRC is being underestimated. In turn, mFTP might be slightly elevated. This is because, although the parameters statistically well-separated (much more so than in the critical power model, where the dependency is very high), there is still a small "teeter-totter" effect.

In any case, there is one way to find out: choose a duration or durations at which your mean maximal power is lower than predicted by the model, and give it a go. Either you won't be able to significantly improve on your performance (meaning that your estimated FRC is about right), or you'll bump up the curve significantly, resulting in a more accurate estimate of your FRC.

As for stamina, yes, you can consider it somewhat of a corollary to Hunter's fatigue profiling approach, at least as applied to lower intensities (i.e., sub-FTP) and hence longer durations. As I mentioned in the webinar the other night, though, it's more variable than the other model parameters, so don't read too much into modest changes.

You can make it bike only if you want, here is the correct expression: tl(if(sport="bike",tss),ctlconstant)

CTL is an alias, in this case you need use the full formula for CTL which is tl(tss,ctlconstant), tl = training load. Now since you want training load for bike, the if statement needs to go after tl, so you get tl(if(sport="bike",tss),ctlconstant).

A simple way to see this in play for bike, run, swim is to open you chart library and search for "multi" and you will find a multi sport PMC which has all three sport types as if statements so you have 4 CTL lines and some other cool information. In this chart, open config and look at expression, copy and paste...fastest way to build an expression :-)

---

Tim Cusick
TrainingPeaks WKO4 Product Development Leader

Yeah, I did that for the cycling PMC. Can I apply that to all athletes or do I need to go in and rewrite the expression for each athlete?

Also, how do I make the monthly cycling summary report (child report) automatically filter on cycling? Can I write an expression in there to do that? I don't want to have to click the chain for the cycling pack and then switch it to the shoe for the running pack and then both for the multisports pack. I want to set those packs up so they auto-filter and then I want to apply that to all athletes I coach. Can I do that? Is that possible?

To move to other athletes is simple. Just name it as you see fit, for example PMC Super Squid and close the Config box. Notice there is a small DOT to left of chart name. This means this chart has been saved to that athlete and since it is open on your NAV BAR means it will be there each time you open that athlete. Now, click the small down arrow on the right of the chart name to get the same pull down that you select CONFIGURE from but look lower and you will see SAVE CHART TO LIBRARY. So save the chart the library. Once saved the DOT disappears indicating that it is in the library. Now navigate to the new athlete, click the CHART LIBRARY button and in the search box type PMC Super Squid, you will find it in library, double click to open for new athlete now it is remembered on that athlete nav bar. Repeat for those athletes you want that chart to open.

Also, in the NAV BAR to can drag and drop order and that is remembered. For example, at athlete level I drag my PMC chart to the first report so my athletes always open on that, I put the athlete details last (far right) so I can find that easy if needed.

I assume you are looking at the PMC Multi Sport and see that it already gives you the breakdown of BIKE/RUN/SWIM duration by weeks and want to build months? The easiest way to build expressions is to copy. So in that report you can copy the weekly summary for bike expression (for example, can use any sport) it is:

sum(if(sport="bike",duration),"week") The expression sums your duration for bike workouts per week

To make that a month, just edit the word "week" to say "month" (note the " " are needed). This means:

sum(if(sport="bike",duration),"month") expression sums your duration for bike workouts per month

Now a shortcut can be to simply open you library and search that chart (not the full pack) - WEEKLY MULTISPORT SUMMARY REPORT and open it stand alone. Make a copy of it and name copy MONTHLY MULTISPORT SUMMARY REPORT or whatever you want. In this one just edit the existing expressions by changing week to month then link that new report into your PMC SUPER SQUID save it to your library and use your new totally custom PMC report on all (or select) athletes.

You are now starting to build workflow. This is what WKO4 is about. At first the learning curve takes some time but once you get it , you realize it is not just about building charts and millions of possible analytics, it is about building the workflow, the perfect review for you and your athletes that will quickly give you what you want.

I still remember day one when we started on WKO4 we spent a lot of time talking about what reports and data to put in. We spent hours and hours talking about the reports people wanted to see, we talked about what Tri coaches wanted vs. Cycling coaches vs. runners..... The reality is the list was massive as everyone wants "their reports". So that is what we built. It takes some learning but does the job amazingly well.

---

Tim Cusick
TrainingPeaks WKO4 Product Development Leader

Do you (or Hunter or Kevin) still have those sheets of paper?

If so, I think it would be cool to have them framed.

Comparing my TSS per workout with Trainingpeaks.com WKO is off by a 1-3 points on nearly every workout so my CTL is a few points higher in WKO4 than online. This is on bike and run. Any ideas how that could be?

It would be great if you guys could add Friel Run HR Zones to the training levels on the Athlete details page.

You mean Trainingpeaks.com is off slightly, right? ;)

WKO4 handles gaps in recording a bit more elegantly, resulting in slightly different values for TSS, and hence the PMC metrics.

Great....my biggest pet peeve with WKO3 was that the workout data didn't always match TP, usually run pace. I was hoping with WKO4 and syncing with TP rather than just uploading or downloading that maybe WKO and TP would finally agree so I'm looking at the same numbers whether I'm in WKO, TP.com or the app.

Most of these are indoor workouts without gaps and WKO is still different than TP. The difference is bigger on workouts with gaps, but it still doesn't agree on workouts that have no gaps.

Can you explain a little how mFTP is calculated? I haven't been doing a lot of short, hard efforts lately or any 20' FTP tests but I did do an hour at 310 watts (AP, indoors) and it's showing my mFTP at 302 right now. Looking back at 2014, I was doing more short, hard efforts back then and had done some 20' tests (so I have all out 5' and 20' efforts but don't have that in my 2016 data since I haven't done that test in a while) and it's showing that my mFTP was 321 but I did an indoor TT during that time where I held 308 watts for 55 minutes. I'm trying to understand how the mFTP could be so different for those periods when both time periods have a ride that's more or less an hour all out and the AP/NP for both rides were very close.

What gets exchanged is the raw data file, along with ancillary things like notes. For TP and WKO4 to be completely in lock-step would require significant changes (and investment) on the part of TP.


mFTP is calculated by fitting a multi-parameter mathematical model of the power-duration relationship to your mean maximal power data, using a custom-built Levenberg-Marquardt solver. The underlying assumption is obviously that your mean maximal power data are reasonably representative of the best you could do at each duration - if not, the parameter estimates will clearly be "off." As it turns out, if you scoop up 90+ d of data from people who are actively training for competition, that assumption generally holds.

In your particular case, you have two mFTP estimates (i.e., 302 and 321 W) that are each w/in 4% of your maximal power for 1 h at around the same time (i.e., 310 and 308 W). That's w/in the limit-of-agreement that is to be expected, i.e., I see nothing wrong with your data.

(Hijack alert)


Here is a cool chart for this group. Check this out.

Dean Golich from CTS was tracking the effects of positioning/Cda/Aerodynamics on power in similar steady state efforts then creating a "correction" to determine what power it would take to maintain speed in reduced aerodynamic positions. Kevin Williams and Andy helped him build it in WKO4 to review track and repeat steady state TT efforts. Check it out, very cool for anyone wanting to understand aerodynamic effect on power to maintain a speed.

https://www.dropbox.com/s/76gcvjenflddlbm/Dean%20Golich%20Power%20at%20Speeds.wko4chart?dl=0


Just download the file and drag and drop if over WKO4 workout, it will load.

---

Tim Cusick
TrainingPeaks WKO4 Product Development Leader

Hi Tim

I am trying to do a simple Time in Zone chart

X axis is date, organized by weeks
Y axis is hours and a stacked bar chart with Z5 on top of Z4 on top of Z3...all different colors

What type of expression would I use for this ?

thanks

marc

Marc,

Sorry slow reply, glad to help.

So start what the data channel you want to build: Power

power

Now lets figure out what we want to do with power. In a time in zone chart, you need to group the summed time together into a "bin" so we get:

bin ( power )

That begins to tell the system to bin the power into groups but we need to tell it what groups. I think you are using coggan classic zones. Which would be: cogganclassic BUT time and name statements need quotes so it is "cogganclassic"

so you tell the system to use the "cogganclassic" to bin power by:

bin(power,"cogganclassic")

This will give you each bin. Time range can be controlled in the RHE so need to limit it to a week. In the RHE you can also use command or control (hold control down while selecting multiple time ranges) to compare the charts. This will overlay the data but you can also configure the chart to Time Series Separate if you want to see if distribution chart separately.

Hope that helps!

---

Tim Cusick
TrainingPeaks WKO4 Product Development Leader

Tim, the reason I wanted to split the files is to eliminate the corrupting effects of some very long pauses on the MMP curve, as has now been amply discussed on the Wattage list. I didn't succeed using the protocol you sketched out, so I am resorting to using WKO3 to split them, where it is very easy. Would be nice if it could be similarly straight forward in WKO4.

thanks Tim

I had actually managed to find the bin possibility.
There was an example of that and I did see the cogganclassic binning method.
The multi-select time period is cool though. I learned there. thanks

but I was trying to do something like this. It it possible ?