davidtinker

Hmm. I will have a look. If you change it on your most recent activity or in settings it should apply to new activities. It doesn't track Strava changes (yet .. thats planned for people with scales that update Strava),

I have added the estimate for the ride to the help info displayed when you click on the "Est. FTP" number. As you can see I have some work to do before the PPA Sportive #5 on 1st September! If I don't get my FTP up to 310w its not going to go well.

Wow I didn't know about the TR podcast. Tx! Do you have an idea how far in? I see the latest is nearly 2h long. I will add an option to show the estimated FTP even if it is lower than the running estimate with decay. This chat stuff has been sucking a lot of time. Once thats out the way its back to implementing analysis features. Still need to do group chat + sharing activities. Hopefully this weekend for at least group chat.

Decoupling % is only displayed for rides with more than 2h of useable data. Short rides also don't break out the 1st half vs 2nd half info.

Intervals.icu now has a chat system. You can search for your friends, share training tips and ask to follow them. If you don't find them on Intervals.icu you can invite them to signup. Coming soon: Group chats and sharing activities.

It works over time. If you maintain your current training load it will stay the same, if you reduce training it will decay. You need to do something above the critical power curve for your current FTP to get it to go up, specifically a max effort of 60s or longer. I am going to make the min time max effort configurable as 60s might be a bit short. It actually estimates for every ride but only uses the result if the new estimate is higher than the current estimate (with any decay applied).

I had a look at your ride on the 5th of July. The intensity is 116% for 1h13 and that means your FTP is set too low. You shouldn't be able to get intensity of more than 100 for anything longer than an hour. Low FTP will also cause the power spike detector to "fix" things that aren't spikes. It uses your FTP to spot suspicious power. Setting the power spike threshold to 99 effectively disables it. I think your FTP is probably something close to the Intervals.icu estimate. Hope so anyway

The algorithm picks out all the intervals starting with a highest power point working out looking for a sudden drop on either side. There are heuristics to figure out if something is really an interval (long enough for the intensity) and others to combine intervals with short gaps between them. Everything in between becomes recovery. It hasn't done too well on your race which was basically 2h51m of flat out eyeball popping riding with no intervals. Its picked out a few points but the in between stuff is almost as hard and not recovery at all.

intervals.icu now support tablets. Touch and drag on all charts to move the marker and expand/pinch to zoom in/out on the ride timeline chart. The activity calendar changes its layout in portrait view to fit everything in. Testing has been done on an old iPad 3 and brand new iPad air so hopefully it works on some others as well. Otherwise let me know!

First up I have fixed the zooming bug. Tx for the reports guys. On the power vs HR chart and decoupling. The model fit is just the r^2 value for the line vs the points it was derived from. Poor model fit doesn't mean you should ignore the chart, just look at it closely and see if the lines look good. I need to do some more work to automatically exclude "bad" data from the model e.g. warmup when you can do 250w at 140 HR because you are still freezing (literally at this time of year). Same from descents though it does get rid of some of those. Then model fit will be better and the decoupling value better. @Frosty your power vs HR is all over the place so the model fit is bad. You did 350w for more than a minute lots and sometimes that was 140bpm, sometimes 170+. Thats not going to make a good model. From what I have read people look at this more for 2nd half of long endurance ride compared to first, not so much racing.

It seems to break if you drag out the zoom brush from right to left .. if you go from left to right it works. Anyway I will sort it out tonight. I have been doing a lot of work getting Intervals.icu to work on touch devices (tablets, big phones). You can now touch expand / pinch to zoom in/out.

Thanks for the report. I will sort that out tonight.

The Intervals.icu fitness chart now has a "Refresh" button that will fill in any gaps on the chart. New activities will get your current FTP and weight. If this is incorrect you can use the "Bulk Edit" button to fill in your historical FTP and weight.

You can now bulk edit activities using the "Bulk Edit" option on the activities page. Use this to fix up your history. It can take a little while if you have a lot of activities to update.

I'll take that Good to see "using max effort over 60s to fit you on FTP curve" gets close to a ramp test.

She says she is "fine"

Intervals.icu now has power profiling on the /power page (thanks to shaper for the suggestion). What kind of rider are you? I still need to do a bit of work on this. Distinguish between climber and TT guy? Under 70kg = climber?

You can get it from the /power page, its listed at the bottom of the chart for each power curve plotted. It represents how much work you can do above your critical power / FTP before blowing up. You can see how it works by adding the W'bal chart to the ride timeline view. When you are riding below FTP then it recovers. The model isn't perfect as it doesn't consider fatigue at all.

You can now see normalised watts and training load for each interval. Click the "Fields" button under the ride timeline chart to add them. These are only visible for newly analysed rides and rides you re-analyse by editing your FTP (you don't have to change it, just click it to edit and choose ok).

It will disappear from the calendar. You could just mark it private in Strava instead? As far as I know I am required to delete data obtained from the Strava API when a user does so.

If it doesn't find more than 2h of data points for the chart (1m of avg HR vs avg power) then it doesn't show the decoupling. The ides is that is more something to look at on longer rides.

I think I am going to have to dial this back a bit and bump the default power spike threshold up from 8% over your curve to 20% or so. Lots of people have FTPs set very low, easily 8% or more under what Intervals.icu thinks it should be. People with dodgy power meters and correctly set FTP may need to tweak it down to eliminate enough of the spikes.

Please mail me links to those activities and I will take a look (this is new stuff, could be buggy or need tuning). Probably (at least according to the power curves in Intervals.icu) your FTP is a bit low. If you don't want to change it you can increase the spike detection threshold for the ride and in settings for future rides.

The fitness chart should progress. I just checked mine now. The activities chart does not, I will fix that. Please mail me your Strava ID (david.tinker@gmail.com) so I can have a look at your chart.

Intervals.icu now detects and fixes power spikes in rides. So if your power meter sometimes acts up it won't make a complete mess of your training load, estimated FTP and other stats. It compares your power output to what you should be capable of based on your the FTP for the ride. If your FTP is 300 and your power meter says you did 450w for 10 minutes (ha ha, nice try) then that is likely to be a measurement error. If large chunks of the ride are missing then your FTP is probably set too low and you should consider bumping it up. You can see the unmodified power trace by clicking the charts button and adding Raw Power. The screenshot shows this in action on a ride. The series of spikes to 1875w have been eliminated (the orange line on the chart) and the numerous drop outs have been corrected. There are some really ropey power meters out there! The detection threshold is configurable per ride (click the 'Power spike' message) and in settings for future rides. But usually you should consider raising your FTP instead of raising the detection threshold.