Forecast AI Beta: How is it working?

General
140

You wiil get fitter and fitter until you don’t.

Absolutely.

Could I ask then why XERT predicts when you need to increase your volume (rather than doing similar volume at higher load) by looking at the average person’s response to training rather than your response to training?

more likely due to out of band factors than training load

I’m just totally confused by this. 4 years of improvement with almost no increase in time trained or intensity distribution - and it’s just down to nothing to do with the fact that I am doing similar volume but now at a higher load?

141

I think what @xertedbrain is saying on volume (if you define in hours) is that it’s not directly used. XSS and training load are what drive the model, and for communication purposes that’s translated back to hours - am unsure whether that’s based on average intensity or personal intensity, but am not sure it matters - look at the XSS level and ramp, and if you figure you can do that in fewer hours, go for it. It is important in the initial plan set up though, I agree.

I think the whole ‘personal’ vs ‘average’ when it comes to responsiveness may be somewhere in between…, in your account settings there are parameters for training response which are personalized, but presumably have a degree of averaging / balancing against the population (think statistical credibility). I don’t work for Xert and don’t know, but it’s my impression from what I’ve read.

The accuracy of that for you will depend on quality of data, and one of the important things is presumably how frequently you have had breakthroughs in the past. That’s important so your signature is up to date (it decays unless you change the default) which in turn affects historic load calculations on one side (overstates training load a bit). And on the other side, with infrequent breakthroughs I expect it has infrequent personal data points from which to infer your response to training. E.g. you could be getting a lot fitter but if you never fully express it, the system won’t know. I have no idea of your training history or data, just mentioning a possibility. Maybe post your PMC (unless you’d rather keep private) to get a sense of some of those things. There may be other considerations from the Xert team too

Am not quite clear what this means - do you improve each year on constant hours and constant training load? Or constant hours but increasing training load. Per above, Xert models the latter ie XSS and training load.

Or by load do you mean absolute level eg your 100 TL was based off a 350w threshold one year, but a 380w threshold another year, in the same hours and intensity distribution. If so, very keen to know how you did that :blush: and actually, there is an interesting question of how / why fitness signatures track with training load (at least for me and presumably many), given training load is a relative measure…

142

I’m not certain if the Forecast AI is working well for long Grand Fondo rides. I registered to a long ride end of June and set it up here entering my time availability. The event itself is 270km long, has 5000m elevation gain, takes 11-12h of riding and Xert expects 800 XSS.

Last year I often rode 300-500km a week and during peaks more than 2000km in a month.
I do 4+ hours tours every week, 6-8 hours at least once a month (more during summer). Also towards summer, my rides contain more and more elevation gain with longer tempo/threshold efforts. I also ride on Zwift, where I do most of my VO2 / sprint workouts.

Back to Forecast AI: I configured Tuesdays for 4+ hour rides, but almost all Tuesdays are Rest and preceeded by a high intensity ~100 XSS on Monday. Only the last Tuesday before the event is a longer 400 XSS ride while all other long/intense days are below ~200. Hence I slightly doubt the plan as it doesn’t really ramp up the “per ride XSS” at all. Like just ride 2-3 hours, then once 6 hours to be fit for 11-12 hour ride.

143

We need to address some possible issues with 4+ hour availabilities. Try setting it to just 4 hours and see if the plan lays out better. Post the chart here so we can see what you’re describing.

144

A lot of other people have mentioned it but at the moment “event” just does “Minimum xss to make sure you don’t fail your event”.

You have to use the improvement option to maximise fitness within a given timescale.

1 Like
145

I think the problem you are describing is the same as described here:

1 Like
146

After switching to 4 hours, Forecast AI uses these 4 hours almost every Tuesday generating ~280 XSS recommendations.
But now it has kinda scrapped/ruined the taper week

image
image1602×289 43.3 KB

147

800 XSS on Sunday???

148

Click and drag over the last week in your plan, click on the -Form- legend and mouse over the lower red dot on your 800xss event. Show us a screenshot.

The workouts leading up to the event will have small slivers for high/peak and thus are simply keeping the engine revved up, so to speak, for the event. You’ll be fresh for the event and recovered in less days that what your Event Readiness is set to.

If you think you can handle more training, you can use the Program window (ciick Program on the top right) and manually input more xss and/or use the Settings tab and lower the Event Readiness and re-run a forecast to see final Training Loads are within reach. Ultimately, it’s the Event Readiness together with the event XSS, Focus and Specificity that define the target Training Loads you need to reach for the event and thus, the amount of training you need to do in preparation.

149

That’s the event. Rosenheimer Radmarathon, longest tour, 266km, almost 5000m elevation gain

I expect about 6 hours of tempo/low threshold climbing paired with lots of descending and some flat bits for a total of 11-12 hours.

150

I hope you mean this view of the forecast chart

image
image823×578 38.3 KB

I would start the event with a form of [17 17 -0.1 0.1]

151

Perfect. :smile:

Your status is blue at the start (blue dot above the red dot) and these will drop to red over the course of the event. It’ll take you 3.8 days for low intensity to recover, 3.5 for high to recover and 1.4 for peak to recover upon completion of 800XSS. 4 days to recover is considered “Poor” Event Readiness. For an 800XSS event, it’s hard to get to 3 or 2 Event Readiness since that would require a lot more training. You can move the Event Readiness slider under Settings in the Program window to 2 or 3 to see how much more training you’ll need (lots more hours!!!).

If you want to dig into this deeper, if you think the event 800XSS is too high/low (or not sure exactly how much XSS this event would really be), you can upload something similar that you’ve done in the past and use the XSS from that in order to get a better gauge on how much XSS the event will have. You can also manually adjust the Focus and Specificity although that’s even more advanced. This will help determine how much training you’ll actually need to do to prepare. The key training outcome being that you’ll be fresh before the event and fresh after the event by a certain number of days. The optimizing algorithm will create a taper that ensures you’re fresh before and will help you reach the training loads necessary for the Event Readiness. The less recovery afterwards you need, the more ready you are for the event.

2 Likes
152

Is there any way to get Forecast AI to recognise something as an important event within the program?

e.g. I have a 5 hour race on Saturday. I put it in the planner and the engine is spitting out 2x intensity days before it.

image
image2219×633 77 KB

153

It definitely should. Can you send this into support in case we need additional info?

1 Like
154

Sorry, it’s me again. Maybe my now 600 XSS target event throws the Forecast AI off? I don’t know.

It is generating some really hard training targets, especially concerning High and Peak energy. For Tuesday (my long 4h day) it suggested a Puncheur workout with 30+ High Energy and AI Generator made this:

753c3f74-231f-4233-8249-a574f06f0bdb
753c3f74-231f-4233-8249-a574f06f0bdb1578×823 169 KB

2 minutes @ 522W is very challenging for me - I managed to do two intervals before I had to turn it down during the 3rd set to 90% (470W), skipping the 4th set to recover. I afterwards reduced my HIE by 10% generating a manual false breakthrough. Good idea?

I also burned almost 25g of Carbs during a single 520W interval, which would mean eating a gel every 9 minutes to keep fueled! Yes, 350g in 3.5 hours is a lot of Carbs to eat (and digest). I’m no pro cyclist who is used to taking 100+g Carbs per hour. Does Forecast AI (and the AI Generator) consider Carb intake at all? I’d guess no. For 30-60 minute workouts this might be fine to rely on stored Glycogen but not for 3.5 hours.

I’m also not good with short / hard intervals. I can quite well ride for 20 minutes @ 105% TP. VO2 is decent, but I’m not that good at anaerobic. In the past I preferred to do two hard, short workouts (VO2 up to sprint) per week each with 60 to 90 minutes including warmup and cooldown. I see no way to make the Forecast AI split up a big chunky workout into two or several smaller ones. Is there a way?

Tomorrow AI Forecast want’s me to to a 195XSS [7P 23H 165L] workout with “Zone: 6 - Anaerobic. Target Interval Power: 481.9 watts.” in the description. Using that I can come up with:

image
image1609×790 68.3 KB

It’s “Good, Optimal Focus” . Why is it “Good”? Close enough match in Peak-High-Low XSS target would sound reasonable. Why is it “Optimal Focus” when it’s Rouleur and not Pursuiter?
Doing 80 seconds @ 130% TP (about 3MMP) seems manageable. It’s only 13.6H but I don’t see a way to increase it on such a long workout without blowing Carb consumption through the roof. On a short workout, I can probably reduce the recovery to like 2 minutes and start the next interval before MPA has recovered. Should be possible for a set but not for 3 sets.
By the way, the AI Generator made this one:
image
image1597×764 85.7 KB

Not sure I can follow that AI workout, seems to be too close to 5 times breakthrough territory in the first set.

155

The 80:20 polarization can make workouts very challenging. If you think this is too much for you, reduce the polarization level to 75:25 or 66:33 and re-run your Forecast. This should result in less challenging workouts.

156

Also, do send any discrepancies you see in workouts to support so we can have a look at your account and see why these are there. If the generated workouts are not matching your forecasted targets, there could other issues that might need some attention. Thank you for your understanding.

1 Like
157

Training Suitability – Xert (baronbiosys.com)

Focus type will vary depending on duration and specificity of the workout along with where you’re at in your plan (Periodization).

Are you relying on Autogen because there aren’t many 2+ hour workouts in the library to choose from?

158

Yes, and I also wanted to try it out. But now I realized that it goes way beyond the target power for the day to generate a specific type workout. Here’s an example:
image
I would use the textual guidance to generate a workout using the 3MMP. AI Autgen will generate a “Mixed Pursuiter” focus type workout by using basically anything and going up to 1MMP.
Hence I have to generate the workouts myself:

  • for doing the correct power on a certain day
  • not blowing through the Carbs
  • not doing too short intervals like the Criterium series (my trainer ramps power up in 7-10 seconds)
1 Like
159

Workouts in the ~2-3 min power range can be tricky for a few reasons - either you achieve the strain targets in a reasonable amount of time (and have a relatively high difficulty score - like Ronnestads), or you aim to keep difficulty score low, which really drags out the duration of the workout if you’re trying to do repeated ~1-3 MMP efforts.

For example, to achieve ~27 High XSS, you could either do a short(er), hard(er) workout, like one of the Crit series and be done in 75 minutes:

image
image1920×905 123 KB

Or you could do a (relatively) easier workout that takes considerably longer - which isn’t an option for some athletes with availability constraints… something like this would take over 2 hours to complete:

image
image1920×905 99.5 KB

You also bring up a good point about glycogen depletion on longer/harder rides… It’s something that we can look to address in future updates to the autogenerated workouts feature.

Might depend on the trainer you have - some are faster to ramp up than others. Also, the ramp back down usually offsets the lag on the front end. You might even consider doing the workout in Slope mode and using your own gearing & cadence to control the efforts. I hesitated a long time before trying one of them, but just did one for the first time a few weeks ago… wasn’t nearly as bad as I had anticipated:

image
image1920×745 145 KB