Haven’t tried this (not at home) but doesn’t the manual entry override the WOTD numbers?
For example, I select my Road profile on the Garmin as I’m heading out, Buckets syncs with XO and the dot goes green, then I use GC to send new numbers to the Edge.
I’m pretty sure once the dot goes green the app doesn’t check again for the remainder of the ride.
I put the numbers into GC then started my computer when i started my ride and let it sync. Initially I saw the numbers I inputted but then my Garmin Synced and replaced the numbers with the WOTD numbers. I’ll try your way next time. Like I said it’s not a huge issue, I can still see my numbers but it would be nice not trying to read the numbers out on the road, easier to see if the buckets are full.
Did a longer training ride yesterday (7:50) with efforts in the first 2:30 of the ride. I wasn’t going all out on the efforts, just doing the power that the AI told me to do but not to the point I was blowing my self up. I got a Gold breakthrough though only TP was significant. I had not done a true breakthrough for TP in a while. I blew past what the AI said would be my number in mid March by 15 watts already. I likely could have got another 5-10 watts had I done a true breakthrough effort. Now I look at the forecast and it is forecasting that my numbers will be exactly what they are today in the middle of march. It makes no sense, is this useful at all? I’m just going to use the workout plan and the targeted numbers/work and ignore the forecast, it seems to be 100% worthless.
I think the problem lies in the fact that we as users have specified a target and if you reach it far sooner than the target date it still remains the target. There either needs to be a way of incorporating further progression in at the beginning in case you do this, or what I would do and that is modify the plan to feature an increased power output for the targeted date. You don’t need to start a new one, though that is always an option, simply modify the plan.
We could look at it that the AI training recommended by XERT has been that efficient at aiding you in increasing your fitness that you have surpassed what it thought you could do at the start.
I’ve also played around a bit with this and no matter how I picked the parameters, at best I would only end up where I was (which is not close to my yearly bests). Anyhow, eventually I gave it a shot and while I haven’t been that strict in following the plan I’m ahead of the forecast in outcomes. 
So the prescriptions seem more sane than the the projected outcomes.
Except for all the fun bells and whistles the main reason I’m a long term Xert user is that you are not force fed a plan. My training is rather opportunistic. With the forecast you retain your freedom, while getting some better insights when it comes to planning, so kudos for that!
It would be nice if the outcomes could be modeled better though. Of course at least in part it has to do with “garbage in, garbage out”, but if my TP is down 40W it doesn’t seem reasonable that I could never get it back up a single watt. But hey, most training software has thought that power spikes of a few thousand watts is something to keep, so maybe it’s not that bad.
If it was actually AI it would recognize this and just keep providing the training needed to improve to the best power you can have at the target date. If you think about it, you really shouldn’t be putting in a target power at all, nobody is going to want less than the highest numbers they can have at that point in time. I’ve stopped even looking at what it’s predicting, just looking at my load, fatigue and the amount of high and peak work at what wattage it currently wants me to do. The prediction is laughably bad.
I’ve been using Xert for over 2 years now, I’m pointing out the issues with the AI but I’m not going anywhere. I’m a former cycling coach myself so I can fill in the gaps and fix the errors when needed. Xert helps me train myself, and I expect it to keep improving to the point where hopefully we aren’t having these types of conversations in the future. For example, I finally had to cut back on some volume this last couple of days, I was exhausted. My load is currently 130, Xert says I’ll be low 150’s by target date of mid march. Highest I’ve been since using Xert was 140, but that was 2 years ago and all I did to get there was a huge amount of volume, zero high intensity work.
1 Like
The prediction is based on your historical data. It doesn’t have some opaque prediction method. Just extrapolates your training load and fitness signature using the responsiveness you’ve shown historically to provide a value. If it’s laughable, you might want to review your own data.
Does that mean that lack of breakthroughs/all outs in old data will give poorer predictions? How old data is used?
Yes. Not just lack of breakthroughs but also breakthroughs from faulty power meter data. Having a robust and up-to-date historical data set provides Forecast AI the necessary information and starting point to prescribe a future training plan. Remember that Forecast AI doesn’t follow some training philosophy so it can’t make something up if the information isn’t in your data. It can use defaults however but these may or may not best represent you as an athlete. You do have access to these directly so that you can control how the system is working directly. If it’s not coming out with an outcome that you believe is achievable, you can make adjustments yourself as needed.
For those that truly understand what Xert does, they could be a very valuable coach to help those that may not have the best data and need help.
3 Likes
I can only assume then that Xert doesn’t look back very far in the data? All my numbers were meaningfully higher 6 months ago than it is predicting I can attain in the next 2 months, and I haven’t really trained 100% since signing up for Xert. I’m not trying to be negative here or put down the work you guys are doing, I’m just explaining the experience I’m having with the system and how the prediction is wrong by quite a bit. I’ve decided to just ignore the prediction and do the work, in the end it doesn’t matter what the prediction is it only matters what reality is. When I used XATA (I think that’s what it’s called) 2 years ago when I first signed up for Xert, it was predicting my TP would be 40 watts higher than I had ever seen. Over time as I rode and my training load increased, that number slowly came down to reality. I assume the same thing will happen with the new system.
2 Likes