I’m still learning about the platform (new premium user), and since I’m currently sick, I have some time. If I understand correctly, PP refers to the highest possible power or sprint power. Why, then, does it differ so much between max 1s, 3s, and 5s power (see examples from intervals.icu)?
If PP is too low, would Xert overestimate my threshold power? That seems to be the case.
XERT calculations:
TP based on 5min/376 W (max ef. 29s) = 317 W
TP based on 21min(342 W (max ef. 2min) = 312 W
Based on the 20min test and Ramp test TP should be lower and just above 300 W.
Should I just do all out sprint and recalculate the numbers? (last time I was doing sprints was in Q1 or 2 2024)
It’s also strange to me that both the Adaptive Training Advisor and Forecast AI are predicting lower 10 minute power in April 2025, despite around 20% more training hours compared to my numbers from two months ago.
Is this a part of the same problem - too low PP - wrong MPA ?
I hope my questions aren’t too basic - I’m really eager to step up my performance next year using Xert. 5 w/kg for 20 min in 2025 seams reasonable
Unlike other platforms Xert uses MPA to derive your fitness signature and power curve and does not rely on MMP sampling alone. The emphasis is on calculated values rather than power points achieved in the past. That is a fundamental difference if you are accustomed to real-world numbers to define your curve.
You can see this in action on any activity but particularly on HIT or hard workouts/rides that include periods of maximum effort under fatigue. Open the activity details and select the Power Duration tab. Note the three curves (Current Fitness Signature plus Signature and MMP for This Activity).
A quick way to validate your calculated PP is to perform (in Slope mode) a sprint workout like this one or a “fitness test” workout as described in the links above. Or a few 20-30 sec sprints outdoors.
Another option is to view details on your last BT activity, enter a higher PP value below the chart, and click Extract. Watch what happens to TP. You can also tweak HIE and enter a median value from this chart then Extract.
If you like the results, click Save which locks the signature so that future adjustments apply to that reference point signature.
An XFAI plan configured for “will likely achieve” is going to be a conservative prediction.
IME I have reached or exceeded predicted numbers prior to target date and I’m good with that.
Another factor is Xert is applying their hybrid polarized model to the forecast which does place an emphasis on more hours than you may be used to. You can tweak that by using the General Settings sliders and accumulate more strain in less hours or purposely select tempo/SS workouts on endurance days for a more time-crunched approach.
If I understand your question, you are right in your observation: peak power (from your signature) should not be lower than max power in principle. What could be going on is that Xert requires effort longer than a certain duration (maybe 3 or 5 sec) to remove power spikes / PM errors. Im not sure if those rides were breakthroughs, but if so, as @ridgerider2 says Xert estimates the PP parameter from that (slightly longer) effort rather than simply taking the highest instantaneous power.
Relative to what’s in intervals for 2024 at 5s, where your PP is lower than observed 5 sec, could it be different time periods? The rides in your screenshot are from April while intervals looks like the whole of 2024 so it could have happened after? (Anyway there is not a huge difference and there could also be some ‘decay’ since the last PP BT if it was a while ago)
PP needs to be ballpark to get a valid signature, you’re right, but it’s not that sensitive to smaller differences. Also, 20m at 342w would give a 325w FTP using the 95% rule (not wanting to start a debate on validity of that) so I don’t think TP is overstated by Xert. If anything it’s lowering your TP vs simpler approaches.
Is that lower 10 min power than today or lower than 2 months ago? And 20% more volume than today or than 2 months ago?If youre ramping volume from today your signature values should be increasing from today… that would be the first check.
Also agree that if you’ve achieved a level in the past with X TL you should be able to exceed it with 20% more volume… intensity could be a factor as @ridgerider2 suggests, and aside from that, it’s good to check confidence in the numbers as the model and predictions are as good as the inputs… e.g. it’s important to have a recent BT before generating a plan / future estimate - when was your last? And you may also want to check how ‘good’ your signature was 2 months ago / were there weird changes in between? Did you actually achieve something near the 10 min power 2 months ago or was it the modelled 10 min power? How do the three signatures and TL breakdowns (all 3 components) compare (2 months ago, today, and April 2025)? From that we may be able to see something interesting
5s activity was in April, also Xert has data for my past 3 years.
If we compare the same activity in intervals and xert:
So, if I understand correctly, Xert is filtering potential power spikes caused by PWM issues, resulting in lower power readings.
When I get better I will validate PP with “SMART - Adventure of a Lifetime - 90”, thank you for recommendation @ridgerider2. Also if PP is least sensitive number when it comes to the signature algorithm ± few precents shouldn’t be a problem.
I’m sorry i made a typo, it would be nice if it would be for 21 min but in reality it’s for 10 min
XERT calculations:
TP based on 5min/376 W (max ef. 29s) = 317 W (end of August)
TP based on 10min(342 W (max ef. 2min) = 312 W (October)
Based on the 20min test and Ramp test TP should be lower and just above 300 W. (September, October)
When I did my 20 minute best (an 8% increase from my previous best - yep I’m rarely doing max efforts), Xert didn’t recognize it as a breakthrough because it overestimated it based on my 5 minute and 10 minute breakthroughs. It would be nice if those estimates were accurate, but unfortunately, they’re not.
So, my question is: if Peak Power (PP) is the ‘least sensitive number,’ why does Xert overestimate FTP by about 4%?"
I agree the model is only as good as the inputs. Maybe a lack of max efforts is leading to FTP overestimations and underestimation for my anaerobic energy system. Low HIE / W’ seems reasonable, but maybe 18kJ is a bit too low.
My last breakthrough was one week ago. I’ve been using Xert for two weeks, so my past 10 minute power is actual data, not an estimate. If I understand correctly, Xert is modeling my future 10 minute power for April 2025 based on the recent BT. If the recent BT wasn’t a true max effort, then the current modeled 10 minute power is too low, which means only the numbers from two months ago are achievable, even if I increase weekly hours from 11/12h to 16h.
Current estimation:
From 329w to 340w by April 12, 2025.
Decay is set to “slow” because I’m in a base period, and my hard efforts are tempo/threshold low cadence workouts, not max efforts, just “maintaining sessions”. I’m also very consistent with my training hours (though not following a progressive overload approach - that is wy I want Xerts help), so it’s a bit demotivating to have a plan with more hours just to achieve the same numbers.
Probably need someone from support e.g. @ManofSteele to give the detail re what data is cleaned by Xert for the power trace, vs what is done in BT detection… I meant that to trigger a BT and update PP it needs to be around 3s or so… the image you pasted shows that the max power is below even the 5s power (assuming that’s the effort referred to in the intervals.icu max efforts), so there may also be cleaning of spikes in the data itself
How do you know that Xert estimates are wrong / why do you think ramp and 20 min tests are more accurate? I’d also note that Xert’s TP is not the same as FTP - probably closer to CP which for most is a bit higher than ‘FTP’… have you tested how long you can hold Xert TP for?
More broadly, you mentioned you don’t have a lot of max efforts - the more you have, also across a range of durations with ‘natural fatigue’ (not just stopping after - minutes), the more accurate your signature should be… it can go wrong though, so it’s good to ask the questions
Correct. Your TP is down vs September and the question is whether this is a true reflection of lost fitness, or due to decay and a lack of max efforts. I would therefore aim for a new BT which is truly max (keep pedaling till you can’t hold TP) and then see what Xert predicts. You haven’t said how your training load compares at those points, which would also help e.g. if TL is much lower than at September (natural due to off season if you are in the northern hemisphere) it’s natural to have lost fitness and need to work to regain it)
