I’d like to be able to teach the TrainerRoad AI that I simply loathe some workouts and will nearly always choose an alternate if they’re scheduled.
For example, I can’t stand the [AI favorite] Cloudripper family. It is much easier for me to have the shorter recovery period from the equivalent Wheel family with the endurance block instead of the tempo block that Cloudripper prescribes.
I’d also love to be able to choose duration over intensity for my shorter recovery rides. I was prescribed King today (30 minutes with a 6-minute block at 78% of FTP); I know I will feel better if I do Neige or Mediane (45 minutes with 7-minute blocks closer to 66% of FTP).
I’ve often thought this would be a good option, plus a “I bloody love this” option.
As I’m sure the outcome between a few workouts would be the same, but some I love to see and some I detest - any mixed just makes me happy and any z2 sessions that are a single static power for the whole block will cause me to swap workouts quicker than you can say Dylan Johnson.
I think this would go over well in my household—my spouse hates King specifically, but gets assigned King for easy days all the time, and will literally do anything else lol
It’d be great to be able to tag that level of hatred/love for the AI to learn from
Obviously, if you had free rein, you could trap yourself in a program that was all Pioneer-3, all the time and that didn’t help you become a stronger athlete.
But if you had a limited selection, or you had to select very similar equivalents (like favoring Wheel over Cloudripper, in my case), or even if you said “I struggle with this type of workout because of this reason,” maybe the software could help pick workouts that you would do rather than avoid.
Maybe we could have a limited amount of “hate/love this” tags. Enough that something you will never do gets removed from rotation but not so that @Joe is free from all the over unders for ever!
I wonder if we will inadvertently get there. Do we subconsciously rate workouts we like easier? Ones we dislike a bit harder. Overtime, the workout picker might find itself picking ones we like more because of that.
Some time back I suggested that AI emphasize using “favorites” when making a plan… but I really like the “I hate this” function. I know this is not the case but some of the “attack workouts” feel just as unstructured as my mountain bike rides. I just can’t mentally do them. Old dog, new tricks and all.
To a degree. But take a SweetSpot workout. I’d pick a solid 4x15 at the same power every time over 4x15 that’s something like 1min at 88, then 2min at 91, then 3min at 90, then 27 seconds at 92, then 83sec at 91.7%. I hate the undulations. I’d rather just do the same power. I highly doubt that one is better than the other in terms of adaptations. But I hate one and I love the other.
For anything endurance, when I did my own program building, I’d just do them all as free ride as I like to adjust them a bit different than what TR offers, often still getting the same IF.
Now that I’m doing a TR program again, I had mostly been doing the TR recommended endurance rides. I think I’m going to go back to more free rides with my own set schedule and generally targeting a similar IF as the workout that TR suggests (endurance rides only). With how my bike is set up on the trainer with my zwift cog and controlling resistance through my garmin, my cadence grouping is a bit wonky. Adjusting it a bit on my own and I can keep out of the certain narrow ranges where I have to be at like 99 rpm or 83 rpm just to keep a certain power.
I haven’t found that the above method impacts my AI FTP prediction. Granted, my upcoming prediction increase this weekend (300>317) is way too lofty as-is.
I’d prefer that TR would be working towards individual workouts dynamically created on the fly to best serve the needs for the athlete.
In that case that feature request would not be necessary as much (besides maybe defining which interval type you don’t like…for example O/U or constant vs. fluctuating power).
I’d definitely be using this. I’d tag the Eclipse, Monitor, & Fang Mountain families of workouts with all their sloping intervals as never-doing. Being that I don’t use erg mode I get natural variation anyway. Chasing a moving target is a nuisance.
That was @hlab, not Joe.
But I could get onboard with this. Maybe a limited number of tags within a particular zone & workout profile combination. If you meet your limit in threshold over-unders, you still have tags for sweetspot mixed intervals.
Would be super cool if the AI could also search within the athlete’s custom workouts & teams’ workouts to broaden the scope of what’s available.
This sounds a lot like Pandora (music) thumbs up and thumbs down for tailoring a playlist / radio station. I’m all in!
My pet peeve / dislike: ramped warm ups like a lot of the new endurance rides have. 5 min is too short, especially if that first interval is >65% FTP. I also like starting at 50% so I can extend and manually adjust…nearly impossible to do with the ramps.
my pet peeve aside, I’m very much in favor of this feature.
This 100%. I’ve made requests about this in the past to no avail. If doing an endurance workout alternating between 55% and 73%, what not start with the 55% for a better warm up?
If you’re talking about workouts like Ophir that take the form of five equal periods: high, low, high, low, high, between 56 & 74% of FTP as the high level & 55% as the low level, you already can extend & run it at 55%, which is the level the warmup reaches at its end.
Though I have seen some other workouts where the sloped warmup ramps you straight into the first interval.
As I understand, if the white ball (for interval average) under the power output isn’t showing yet, it doesn’t matter, so just run it in resistance mode until you’re ready to hold whatever output it’s wanting. A bit clunky though.
I prefer the ramps rather than the “steps” personally (I do my warm up in erg, workouts in standard) but definitely would like at least a 3-5 min start at 50% so you could extend as needed before ramping up intensity.
