# Can my power curve slope UP? **URL:** \ **Category:** TrainerRoad Software\ **Created:** [December 9, 2019, 11:39pm UTC](https://www.trainerroad.com/forum/t/can-my-power-curve-slope-up/25530 "2019-12-09T23:39:03Z")\ **Posts on this page:** 17\ **Page:** 1
**Author:** ![AgingCannon](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/agingcannon/32/48092_2.png) [@AgingCannon](https://www.trainerroad.com/forum/u/AgingCannon)\ **Post date:** [December 9, 2019, 11:39pm UTC](https://www.trainerroad.com/forum/t/can-my-power-curve-slope-up/25530/1 "2019-12-09T23:39:03Z")
![image](https://global.discourse-cdn.com/trainerroad/original/3X/1/4/1498e4fc79c7dd007054d0870a3856e8153ac427.png) My power curve shows two PR’s that I find very confusing: - Point A shows 259W for 2:53 - Point B shows 268W for 3:45 Isn’t it mathematically impossible for a shorter time interval to have a lower-power PR? Or am I missing something here? Thanks in advance for any illumination you can provide. ---
**Author:** ![SalvatoreGilbert](https://avatars.discourse-cdn.com/v4/letter/s/7ea924/32.png) [@SalvatoreGilbert](https://www.trainerroad.com/forum/u/SalvatoreGilbert)\ **Post date:** [December 10, 2019, 12:01am UTC](https://www.trainerroad.com/forum/t/can-my-power-curve-slope-up/25530/2 "2019-12-10T00:01:17Z")
30/30s at 1000w/0w repeated 30s PR is 1000w 1min PR is 500w 1:30 PR is 666w ---
**Author:** ![jpowers](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/jpowers/32/35855_2.png) [@jpowers](https://www.trainerroad.com/forum/u/jpowers)\ **Post date:** [December 10, 2019, 12:04am UTC](https://www.trainerroad.com/forum/t/can-my-power-curve-slope-up/25530/3 "2019-12-10T00:04:34Z")
It can, as an anomaly of doing intervals, mostly. ---
**Author:** ![Bryce](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/bryce/32/72_2.png) [@Bryce](https://www.trainerroad.com/forum/u/Bryce)\ **Post date:** [December 10, 2019, 12:18am UTC](https://www.trainerroad.com/forum/t/can-my-power-curve-slope-up/25530/4 "2019-12-10T00:18:58Z")
If you zoom in closely you will see why this is mathematically possible: ![image](https://global.discourse-cdn.com/trainerroad/original/3X/8/8/889565c61821efca6583ff4b5145ddc3d3154e53.png) This is the 3:45 interval with an average power of 269, and you can view it for yourself [here](https://www.trainerroad.com/career/agingcannon/rides/66512216-working-off-the-limp?startTick=2455&endTick=2677). In this workout, you set new PRs for 2:53 through 3:45. Click the link above and try creeping the interval shorter one second at a time and watch the average power. You’ll notice that the average power does not go up as you shorten the duration due to the spikes at either end of the workout. It trends downwards. ![image](https://global.discourse-cdn.com/trainerroad/original/3X/8/8/88b137f691536c16d4696b4388e43d4ccf07c4d1.png) This is the 2:53 interval with an average power of 258. As you can see, you lose that spike on the left side of the chart, which makes the average power for the interval lower. Does that clear things up a bit? ---
**Author:** ![AgingCannon](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/agingcannon/32/48092_2.png) [@AgingCannon](https://www.trainerroad.com/forum/u/AgingCannon)\ **Post date:** [December 10, 2019, 2:58pm UTC](https://www.trainerroad.com/forum/t/can-my-power-curve-slope-up/25530/5 "2019-12-10T14:58:45Z")
Thanks a million, @Bryce. I didn’t know that I could go from the power curve to the interval like that, and it hadn’t occurred to me that spikes at the start and end of a time period, with lower values in the middle, would yield that result. I had tried to look at it several different ways, but missed that by a mile. Learned a couple of things today! @Nate_Pearson, take note: THIS is the kind of spot-on, detailed, helpful, one-step-beyond educational SUPPORT that will keep me and many others paying for TrainerRoad long after open-source AI can design our training plans. The software is great, but it’s the human element where TrainerRoad shines and differentiates itself as a company. Whatever you do to reward excellence in one’s work, Bryce has earned another one. 👍🏻 ---
**Author:** ![Nate\_Pearson](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/nate_pearson/32/117463_2.png) [@Nate\_Pearson](https://www.trainerroad.com/forum/u/Nate_Pearson)\ **Post date:** [December 10, 2019, 3:04pm UTC](https://www.trainerroad.com/forum/t/can-my-power-curve-slope-up/25530/6 "2019-12-10T15:04:10Z")
Thank you! Great job Bryce! ---
**Author:** ![foldingbikebass](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/foldingbikebass/32/101743_2.png) [@foldingbikebass](https://www.trainerroad.com/forum/u/foldingbikebass)\ **Post date:** [March 17, 2023, 11:26am UTC](https://www.trainerroad.com/forum/t/can-my-power-curve-slope-up/25530/7 "2023-03-17T11:26:06Z")
This has been bugging me only this season. It took a while for me to independently come to your conclusions. I have been doing a number of Threshold workouts in quick succession, so I get very structured points in the PR curve for the interval times and then anomalies either side depending on how much of rest recoveries and adjacent intervals are included. Must not have been so jarringly explicit in the past. I was going to say that this wasn’t repeated in other software. A quick scan - Intervals.cc exhibits the same problem. However, I think Golden Cheetah and Training Peaks take a smoothed curve or fewer data points and display that. Logically, if you can achieve an average of Y for time X, you must be able to hold \>=Y for X-n, even if it’s not been demonstrated. Is there not an algorithm tweak in the graphing of PRs that would draw a horizontal line left from each power PR until bested by a greater average? @Nate_Pearson, a hard or easy maths and coding problem? ---
**Author:** ![WindWarrior](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/windwarrior/32/84679_2.png) [@WindWarrior](https://www.trainerroad.com/forum/u/WindWarrior)\ **Post date:** [March 17, 2023, 3:29pm UTC](https://www.trainerroad.com/forum/t/can-my-power-curve-slope-up/25530/8 "2023-03-17T15:29:26Z")
Logically, your PRs are your PRs. Its called a Mean Maximal Power (MMP) curve. That needs to stay because its what you actually did. A modeled Power Duration Curve (mPDC) is what you are requesting. Thats a nice feature of WKO, GoldenCheetah, Intervals, etc. ---
**Author:** ![jbakkane](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/jbakkane/32/21181_2.png) [@jbakkane](https://www.trainerroad.com/forum/u/jbakkane)\ **Post date:** [March 17, 2023, 4:22pm UTC](https://www.trainerroad.com/forum/t/can-my-power-curve-slope-up/25530/9 "2023-03-17T16:22:21Z")
Think of this in real time. Scenario: End of a race, you sprint for 30sec at 400w to get in position, then you hold 200w for 30 sec to hold your position, then you sprint for the finish for another 30 sec at 400w. Your 1 min power will be either end of this scenario. 30 sec at 400w + 30 sec at 200w = 600w / (2 sections at 30 sec) = 300w OR (30 sec at 200w + 30 sec at 400)w = 600w / (2 sections at 30 sec) = 300w It can’t be both the 30 sec at either end added together, because it has to be continuous. So your 1 min 30 sec power will be: (30 sec at 400w + 30 sec at 200w + 30 sec at 400w) = 1000w / (3 sections at 30 sec) = 333w ---
**Author:** ![foldingbikebass](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/foldingbikebass/32/101743_2.png) [@foldingbikebass](https://www.trainerroad.com/forum/u/foldingbikebass)\ **Post date:** [March 19, 2023, 7:22pm UTC](https://www.trainerroad.com/forum/t/can-my-power-curve-slope-up/25530/10 "2023-03-19T19:22:09Z")
@WindWarrior, interesting the distinction. I haven’t seen the PDC on Intervals.cc - I see a drop and rise similar to the TR PR curve - so it looks more like an MMP. On GoldenCheetah, the modelled PDC is a bit frustrating as I can’t seem to investigate it - actually extract the plot points as a useful reference. Hovering over the MMP for any given time gives the wattage, but I can’t get this from the curve. @Jbakkane I get your point. The TR PR curve is mathematically exactly what has been achieved in that timeframe. However, intuitively, if I can hold 333w for 1:30, I can clearly hold the same at least, for 1:00. For those of us who don’t do enough maximal efforts, because we’re busy doing intervals, the PR curve becomes increasingly less helpful. However, I appreciate that if TR “filled in the blanks” with a horizontal line, this would cause problems when referencing back to the original power file. ---
**Author:** ![WindWarrior](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/windwarrior/32/84679_2.png) [@WindWarrior](https://www.trainerroad.com/forum/u/WindWarrior)\ **Post date:** [March 19, 2023, 7:30pm UTC](https://www.trainerroad.com/forum/t/can-my-power-curve-slope-up/25530/11 "2023-03-19T19:30:38Z")
Check the “Show model power curve” option on the Power menu in Intervals: ![image](https://global.discourse-cdn.com/trainerroad/original/3X/9/0/90f76f41865f8ac33134d2e0a34be55801e9c81c.png) Model is light green dotted line. Actual MMP in dark green. In GoldenCheetah, first go to trends section (latest beta), then CP chart, then bring up the Chart Settings ![image](https://global.discourse-cdn.com/trainerroad/original/3X/0/a/0a71851f07a410e3942d889e3f0526863aa11ec3.png) these are my settings for looking at ~90 day window in 2017: ![image](https://global.discourse-cdn.com/trainerroad/original/3X/d/0/d0bf3285026d0469670173e3dd12345b6f5c21a5.jpeg) Data Series = Power Current Activity = Activity MeanMax And my Model sub-menu: ![image](https://global.discourse-cdn.com/trainerroad/original/3X/b/f/bf9a31ed6942bb2bc9bb8ae0d1e3e1bee9ba2971.png) are the defaults. Hope that helps, ask questions if you have any. Changing timeframe to last 3 months: ![image](https://global.discourse-cdn.com/trainerroad/original/3X/6/e/6e798aac0a06e36d18ea63413a401d1d5208cbc7.png) Changing to last 4 months: ![image](https://global.discourse-cdn.com/trainerroad/original/3X/7/a/7a00713d8fdaf2dc4271479c2d8b18186ee894a2.png) etc., etc. ---
**Author:** ![foldingbikebass](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/foldingbikebass/32/101743_2.png) [@foldingbikebass](https://www.trainerroad.com/forum/u/foldingbikebass)\ **Post date:** [March 19, 2023, 8:10pm UTC](https://www.trainerroad.com/forum/t/can-my-power-curve-slope-up/25530/12 "2023-03-19T20:10:53Z")
Thanks for all the effort. Much appreciated. On Intervals, I was being a bit thick. I had the “show sub-maximal efforts” ticked, which removed the modelled numbers from the floating labels and replaced them. I can see the Modelled numbers now. On GC, on first glance, the only difference I had was Unticked “Filter Unique bests” and ticked “Show sustained efforts”. But whatever I’ve done, I can now see the numbers for the mPDC floating as I cursor the curve. Thanks for taking the time. Incidentally, do you know what the “filter unique bests” and “Show sustained efforts” does? Interestingly, I can also see dips in my MMP - clearly they have been subtle enough in the past, or I’ve been further into a season so I haven’t noticed them before. ---
**Author:** ![WindWarrior](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/windwarrior/32/84679_2.png) [@WindWarrior](https://www.trainerroad.com/forum/u/WindWarrior)\ **Post date:** [March 19, 2023, 8:42pm UTC](https://www.trainerroad.com/forum/t/can-my-power-curve-slope-up/25530/13 "2023-03-19T20:42:33Z")
> [@foldingbikebass](#): > > Incidentally, do you know what the “filter unique bests” First off, all options appear on that menu… it doesn’t only show those that apply to Power curve. I’ve only seen Filter Unique Bests do something if you are looking at CP curve (1/power): ![image](https://global.discourse-cdn.com/trainerroad/original/3X/f/c/fc25469a643a4acbbdd6d6b7534abf6d26776e14.png) > [@foldingbikebass](#): > > “Show sustained efforts” The sustained efforts appears to make a difference on the power curve: ![image](https://global.discourse-cdn.com/trainerroad/original/3X/f/b/fb2768b0345355a5f79b656fe5b04b42c8170c82.png) and appears to show other efforts below your MMP actual bests. That is all I can tell without looking for doc (wiki?), or looking at the source code, or asking on Google Groups. ---
**Author:** ![foldingbikebass](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/foldingbikebass/32/101743_2.png) [@foldingbikebass](https://www.trainerroad.com/forum/u/foldingbikebass)\ **Post date:** [March 19, 2023, 11:38pm UTC](https://www.trainerroad.com/forum/t/can-my-power-curve-slope-up/25530/14 "2023-03-19T23:38:11Z")
Ironically, I played with those settings again, and when I clicked on sustained efforts, I lost the ability to investigate the mPDC. There’s something buggy. Will eventually crack what combination I need to click to get it back… ---
**Author:** ![jbakkane](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/jbakkane/32/21181_2.png) [@jbakkane](https://www.trainerroad.com/forum/u/jbakkane)\ **Post date:** [March 20, 2023, 2:27pm UTC](https://www.trainerroad.com/forum/t/can-my-power-curve-slope-up/25530/15 "2023-03-20T14:27:36Z")
You might not be able to hold it for 1 min straight. ---
**Author:** ![enki42](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/enki42/32/68042_2.png) [@enki42](https://www.trainerroad.com/forum/u/enki42)\ **Post date:** [March 20, 2023, 3:43pm UTC](https://www.trainerroad.com/forum/t/can-my-power-curve-slope-up/25530/16 "2023-03-20T15:43:32Z")
Yeah, you might need rest between higher wattage outputs. Though I wonder if there was a way to make a chart that represents the longest power that you can sustain and not need rest. Maybe the max average for each time length but throw out the ones where x% of the data is over a standard deviation lower then the average ---
**Author:** ![WindWarrior](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/windwarrior/32/84679_2.png) [@WindWarrior](https://www.trainerroad.com/forum/u/WindWarrior)\ **Post date:** [March 22, 2023, 12:31am UTC](https://www.trainerroad.com/forum/t/can-my-power-curve-slope-up/25530/17 "2023-03-22T00:31:23Z")
> [@foldingbikebass](#): > > “Show sustained efforts” does? Did a quick search and found this: > > > \#1. Extracting intervals from rides (v3.2) > > This will mean automatically identif…ying parts of a ride as intervals. The key here will be what rules or types of intervals are extracted. At present we could have; > \- \*\*Route\*\* - user labelled routes or segments that represent a route you ride often. it may be a particular climb or col you regularly ride, or perhaps a 10mile TT or training loop. Its identified using GPS. > \- \*\*Climb\*\* - extracting intervals where you climbed/descended; we will need to identify short-steep ramps as long as longer shallower grades. Some kind of scoring metric might be useful for classifying these. Likely to overlap with a Route but do not need GPS, they are extracted based upon distance and altitude. > \- \*\*Peak\*\* - peak durations; for power but also cadence, hr, speed/pace. > \- \*\*Efforts\*\* - a sustained effort. This is quite tricky to describe but basically represents a power interval see comment in post below with annotated CP chart and the wattage discussion. Using the zone and CP config we will identify efforts that represent an Anaerobic, Aerobic, VO2Max, Threshold effort. These are likely to overlap with Peak power intervals. > \- \*\*Matches\*\* - A particular kind of sustained effort where W' is being consumed, so will span all zones above CP. > > These intervals will be identified and associated with the ride in which they occur but will not be added to the ridefile. Which therefore leaves us with one last interval type: > \- \*\*User\*\* - user defined intervals; the ones you added to the ride using the find intervals tools. Likely to also overlap with the automatic ones. > > Since there is an overlap with these intervals we will need to think about how we remove duplicates in the UI to ensure you don't see lots of intervals all the same. > \#2. Augment the Intervals for an activity (v3.2) > > Pretty obvious, but once we have extracted all these intervals automatically we should asociate them with the rides so you can plot and select them in the activity view. > > This will basically mean that the interval sidebar in activity view will list all 6 types of intervals to allow you to select them or drag and drop them into the compare pane. > > If we do this then no other coding is needed to allow you to plot an interval sample by sample -or- compare intervals sample by sample. > \#3. Fatigue Score (v4.0) > > Using the CP, W' and W' bal concepts to define a fatigue score for an activity. This will take into account the relationship between performance and fatigue during intermittent exercise to provide a metric to quantify peripheral and central fatigue. > > some of the more relevant > http://www.ncbi.nlm.nih.gov/pubmed/21487156 > http://www.ncbi.nlm.nih.gov/pubmed/25688762 > http://www.ncbi.nlm.nih.gov/pubmed/23019318 > http://www.ncbi.nlm.nih.gov/pubmed/24492634 > http://www.ncbi.nlm.nih.gov/pubmed/24509723 > http://www.ncbi.nlm.nih.gov/pubmed/24068048 > http://www.ncbi.nlm.nih.gov/pubmed/24021781 > http://www.ncbi.nlm.nih.gov/pubmed/9243165 > > and curious about.. > http://www.ncbi.nlm.nih.gov/pubmed/24740653 > http://www.ncbi.nlm.nih.gov/pubmed/24728680 > \#4. Interval View (v4.0) > > Select and chart 'sets' of intervals. To be able to track and compare intervals. It may not be a new view, rather it may be a 'mode' for the existing views (TBC). > > Like the trends view it is about selecting and charting; where the trends view is focused on sets of rides over time, the interval view is about sets of intervals over time. Of course there is a big difference here, we are looking at specific sections of rides that are interesting for a \_particular reason\_. > > This \_particular reason\_ means we are going to be looking for indicators and correlations that go a lot further than just quantifying stress or tracking improvements (or lack of them). These \_particular reasons\_ are likely to warrant us to offer different types of charts and features within the charts, some that immediately come to mind are: > \- \*\*Ranking and Ordering\*\* - Showing intervals in some form of ordered / ranked chart. Possibly just a bar chart or histogram. > \- \*\*Scatter / Bubble\*\* - Plotting intervals spatially to show any correlations or patterns e.g. gradient and peak power (from discussion on the list). There are possibly other new metrics we will want to add for analysing intervals but we have some useful ones now like fatigue and pacing index. > \- \*\*Radar Plot\*\* - this is actually likely to be equally useful for rides; but aggregating/averaging multiple dimensions for a set. > > In all cases we are also likely to want to: > \- \*\*Compare\*\* - Sets of intervals with each other; showing each set in a different color > \- \*\*Query/Filter Sets\*\* - The sidebar will need to be designed to select intervals into a set; almost certainly using the same approach data filters and autofiltering ux we use in the trends view, but adjusted for intervals. For example; where in trends we allow the user to autofilter by a metadata field, we might autofilter by interval type (e.g. an autofilter for all 5s peaks, all routes called 'Fave Loop'). and this: > **[Finding TTE and Sustained Efforts in a ride](http://markliversedge.blogspot.com/2015/05/finding-tte-and-sustained-efforts-in.html)** > > Defining the problem Any given training ride or race will contain periods of sustained effort, sometimes to exhaustion. Perhaps during th... and doing a quick “sustained” search in the GoldenCheetah source code, I pulled up and spent 5 minutes trying to inspect the code in SustainMetric.cpp, CriticalPowerWindow.cpp, RideFile.cpp, and CPPlot.cpp. Didn’t spend enough time to follow all the code, just got a sense of it to try and following along with the blog post I quickly skimmed. Going off the blog post, it appears the code looks for sustained efforts within 85% of MMP or modeled curve, not really sure of details. But it looks for top sustained efforts and plots them. Looking at the charts below the sustained effort points have a lower boundary - I tried doing some quick calcs for of some in Z7 (zone 7) but couldn’t figure out the details without spending more time. So for example when I was hammering away like crazy and doing a lot of sustained near maximal efforts it looks like this: ![image](https://global.discourse-cdn.com/trainerroad/original/3X/9/f/9fa782b36fc76396670f797ea3b0ea1f0b7426a5.png) versus last 90-days when the focus was on slowly and sustainably building the engine: ![image](https://global.discourse-cdn.com/trainerroad/original/3X/e/4/e472725662f92b4e92ca567addd157d2ca9f4297.png) You see how many more near max efforts there are in the plot, when I was hammering away like crazy vs last 3 months? Longest sustained effort I’ve found is 58+ minutes - easy to find by setting date range of chart from October 2016 (bought a power meter) thru today. Hope that helps.