I realize this topic has been around probably since the first week TSS was ever used by a coach, but now we have AI to help. I acknowledge that TSS/ATL/CTL are primarily an endurance metric, yet those of us who do a lot of both know that even a relatively easy day of squats and deadlifts affects the next couple days disproportionately to the HR Avg/calorie burn/time.
I also know TR has the Working Sets concept - which I have been finding is providing a reasonable estimate into the Yellow/Red day recommendations. My lifting days are all more like HIIT session because I do super sets with minimal time between lifts, typical “lift to rest” ratio is 1:1 or 1:2 at most (so 20’ under load in a 40’ leg/core session)
Recently I started putting my actual sets, reps, weights, 1RM baselines, my Garmin heart rate graph, and the total session time and time doing reps. Google AI is kicking out some interestingly high TSS. Last Monday I did a near max effort leg/core day. with 5 sets of squats building to my 3 rep max, 3 sets of RDLs, and then 4 sets of deadlift building to my 5 rep max. 49’ of total lifting, 20 mins under load, and I hit a 160 HR on my last deadlift. AI TSS = 71. Given how I felt the next day that seems about right.
Is anyone else trying this (or be willing to) - and I wonder if TR could use same thing in the back ground for lifting - assuming people had their 1RM (or more safely a 3RM test)
Update:
I spent a couple hours collaborating with AI on the pros and cons of this. Also, it does need the heart rate graphs, and 1RM along with the actual sets and reps. it’s then using a MET conversion. It looked at 4 options: report zero, report 100%, report 50% of each weight session’s acute TSS, or a Dual Track option.
1.Zeros Hard lifting today -does- excessivly tax your CNS & damage muscles which affects the next 48hours (acute TSS). Thus ignoring acute TSS from a session (like a 70 TSS leg day) will significantly understate your ATL, causing TR AI to program a hard day, despite metabolically being RED.
2. 100%: Adding the full “acute” TSS of a lift, will over time hyper-inflate your CTL such that you appear much more aerobically fit than you are. This creates its own problems with properly programming workouts and the AI understanding it’s affect on TSB.
3. 50%. This sweet spot does account for some of the TSS in your ATL/CTL, but wont hyper inflate your CTL. The downside is on any given day, you may still be more fatigued than it appears.
4. Dual Track. Report 50%, but “outside” TR do 48 hour look at your ATL & TSB using the full 100% acute TSS to allow you to make educated plans for those day’s.
Since you can forecast workouts, including the dual track option there may be some useful meso-cycle planning here. I think it is what I’m going to try for now.
Thoughts?
would you be comfortable summarizing your lifting program?
I understand why he recommends it - here is what my week looks like RN (3 week build / 1 deload) as I’m trying to get to the 1000 lb club (recent lifts have me around 920lb at 159lb BW, 14% BF & a not great 178FTP or ~2.5 w/kg).
Mon:
75-90’ ride with VO2 reps
Tue:
5 rounds of heavy bench press (225+) & 6 rounds of upper body/core supersets w/ no rest (~ 15’ under load with HR over 120); easy elliptical
Wed:
60-75’ Z2 ride
afternoon 20’ walk
Thu:
5 rounds of heavy deadlift (315+) & 6 rounds of lower body/core supersets (same high HR density); easy elliptical
Fri:
5 rounds heavy squat (225+) & 6 rounds of lower/upper/core super sets (see above)
Sat:
75-90’ with Threshold intervals
Sun: 20’ walk
IMO - the high density super sets replicate anaerobic efforts on the bike enough to have -some- TSS added, just like I would if I did sprint drills on the bike. With AI being able to view and measure the “area under the curve” of my HR info post workout, it is getting a better view than traditional methods of using hrTSS as an estimate.
Interesting - I just wing it based on effort - but I agree a short 45min sweetspot workout might only 50 TSS, but if you got after some over-under intervals the next day I think it would impact me. Similar to a hard hour in the gym… So ignoring TSS of lifting…is a bit short sighted, regardless of who says it. Similar to sauna work - if you did 30minutes a day , plus lifting, plus some hikes - its TS …
Personally have had some fun with Heat Acclimation Load tracking in a Gemini chat - where it tracks it for me over time. lol give it some feedback from outdoor training in the heat as well to supplement closer to reality based on what my Garmin “heat acclimation score” is at. Plus have it ramp the acclimation down at a half rate because of all of my indoor training sometimes with fan off, etc. Cool to see long term graph lol
Nothing too exciting, a simple A/B full body split, and nowhere even close to your lifts. I certainly don’t think that fatigue from strength should be discounted. You seem to be primarily a strength athlete, isn’t there a fatigue/load metric in that world? Admittedly I’m mostly familiar with cycling training, truly curious what an alternative would be.
I still think using AI to come up with a random TSS for a strength session seems like misguided precision. Why not have a set number for a hard, medium, easy session, or use one of Training Peaks strength TSS recommendations, hrTSS (Coggan is cringing right now) and call it a day?
apprecieate the comment.
I’ve always been a runner & lifter due to 30 yr in Army. but about 20 years ago got into triathlon and thus the bike. did a couple full IM so have done big cycling blocks over the years, but TR suggests (probably correctly) that my power profile is more of a sprinter/puncher. at 5’7" that 160 is pretty compact and I’m pretty lean.
What the AI says its doing is using time under load and a MET conversion, based on my recent max rep info and a screen shot of my HR graph from the lift. So it sounds like a very objective look at “is this close to a good TSS estimate.”
What I found really interesting is when I prompted it for “pros and con’s” of trying to estimate and use TSS for lifting in a cycling program - it cited Coggins et/all and made compelling arguments for both sides, and then spit out “here’s a possible middle ground”. The last prompt was "what if we use 50% to keep the CTL in check, but off-line (outside TrainerRoad) do a 48 hour look based on the MET full estimated TSS (especially those deadlift and squat days). And when it looked back at my last couple weeks it was dead on about when I felt wrecked despite having a Green/Yellow day.
I wouldn’t worry about an algorithm number matching up with how you actually feel. After heavy strength training, the endurance training is going to be compromised. So you adjust accordingly by feel. If that means the color of something in a program is a little “off,” so be it. The AI models are not coaches.