Forecast AI is broken

I’m rather confused here on the “Achievability” of a certain goal. To me it sounds rather achievable to increase 5-min power by 5W in 5 months with 5-6h of training per week. Unfortunately, according to the forecast, I can’t increase my 5-min power by more then 2W in a 5 month period. Sounds pretty absurd. Screenshot for proof :

How many actual hours per week is needed in order to increase 5-min power? Don’t get me wrong, but I’ve been able to reach higher results by not training seriously in under a 5 month period. Normal bike season of under 2.000km per year mostly “chill” rides with some climbs. My 5-min power for that season was 375W. Now with training over a period of FIVE months, AI tells me I can’t reach what I’ve already done without training? … Let me doubt.

1 Like

achievable: “Increase 5-min power by 5W in 5 months with 5-6h training/week”

Those are absolutes that don’t really mean a lot without context. I’m pretty sure that you can take a lot of cyclists and not even see a 1W improvement in 5 months with 5-6h training/week depending on how far they’ve already gotten in their training career.

You’re not exactly starting from rookie numbers either if your current TP is 294W.

I am not starting from rookie, see edit I just posted. I’m not a professional either. I have uploaded the trainings from this season in XERT and was expecting for it to see the progress I’ve made and learn from it. Considering it’s just the beginning of my training, there’s room for improvement.

1 Like

My experience is it tends to be uber conservative on predicting gains.

1 Like

Post screenshot of your XPMC under Progression tab set to Year.
It’s recent history that matters for the conservative prediction based on where you are currently and how much you can improve on a TL progression.
That’s not to say you can’t bump up the results along the way. But 5.5 hours as a cap over a 6 month period when you are already riding 4-5 hours is going to be a major limiter.
You can, however, tweak other settings to squeeze more TL into those hours if you can handle the intensity. For that consider shorter blocks such as a 60 day goal.

Here it is. Started working out more seriously in January of this year. Last year was mostly riding outside without really training. Also, this is for a Goal I’ve set for end of May, so that doesn’t really relate to the initial question. I was playing around for future plans.

Goal for end of May :

Increase the maximum power you can hold for 5 minutes. From 354w to 365w by May 23, 2025

The reason why I decided to set a longer timeframe then this 1st Goal was to see if I could get the “Achievability” to switch from yellow to green. I was successful but with a power increase of 2W (lol).

That looks like a reasonable goal to start with.
Based on the steady TL increase between now and your target date you will likely reach or exceed your target power increase along the way.
A quick way to determine that is midway through toss in a fitness test workout from the Library on a HIT day preferably when you are feeling especially fresh.

It’s also possible to come up with a viable forecast plan when you push Achievability into the red text level. You won’t know until you try it. It may say something along the lines of – doable but no wiggle room.

The crucial element to a viable forecast is estimated changes in Signature (TP-HIE-PP) and TL (Low-High-Peak) values over the course of the plan.
You can see this visually by viewing the forecast chart and isolate each key value one at a time.
Here’s an example 60 day goal where Low strain remains steady but notice the increases in High and Peak XSS.

XFAI_Goal4minPower9wks

This also happens to be an example where 5 weeks in I exceeded my target goal by a couple watts. :slight_smile:

Your very first ride (with the open circle) could be an error since the signature if far lower than all the rest. The one thing that’s very hard for the system to estimate is your starting signature and training loads. It’s all guess work.

It’s important to get good data analysis on breakthroughs and what your training loads were throughout your training history. Your circles and the black line should be colinear in an ideal situation. This then helps generate a good relationship between training and performance, i.e. “Training Responsiveness”.

Your current training load is 43. At roughly 60 xss per hour, this amounts to 7*43/60 = 5 hours per week. You can confirm your recent weekly hours on the Progression tab.

Investing the same number of hours as you current have been doing may allow you to change the make up of your fitness signature but there’s not a lot that can be done to generate big improvements. Improvements don’t come without investing in your training. This means both more time and using the time for more quality training. You can attempt to do more within the same time you’re putting in but this has limited improvement value. If you think there’s some type of workouts that you respond better to that don’t show up in increasing your training loads, then by all means do them. Perhaps you are a high responder to training and your Training Responsiveness values should be higher. The system isn’t seeing this with the data you provided yet.

Suggestion: create a plan and even if it demands more hours, try and fit in the XSS by simply doing more within the available time you have. At 5.5 hours per week, you don’t have a lot of recovery getting in the way of improvements so should be ok to simply do harder workouts within your available time.

1 Like