INSCYD is a bit out of my intellectual league but isn’t this what Adaptive Training is all about?
Yes, there’s an easy* way to model VO2Max and VLaMax using Mader’s 1986 paper on anabolic threshold. Hauser 2014 also describes this process. Mader 2003 is an updated version, but it’s a bit more complex.
I just finished coding it in Excel and it’s pretty nice. It provides several of the graphs found in Inscyd.
It’s also a fantastic tool for training since you can estimate lactate generation and clearance for interval training.
It’s also great for estimating fatmax based on the lack of pyruvate curve.
For me, the model estimates my threshold pretty closely using my VO2Max from my metabolic cart and my estimated VLaMax (I haven’t tested this yet).
7 Likes
Could you link to the Masters Paper. I’ve googled it but it doesn’t seem to come up
Sorry, that was a typo. It’s Mader’s paper.
1 Like
The 1986 paper is a good start. I recently stumbled upon the patent for Inscyd which describes how the models are implemented, references for the model, and ranges or preferred values for the model constants (magic numbers). Here’s an early version of my calcs. I still have a lot of work to do, but this is a start.
6 Likes
Looks good! To be honest, there is not a lot more magic behind the model besides what you’ve figured out.
Just wondered though, you got Running Power from stryd?
1 Like
That’s good to know! I know the curve values are off for the older model, but the other numbers seem to line up
And yep, that’s power from Stryd. My pace vs power correlation is for a weight setting of 88kg, so I need to revise those predictions based off 100kg and scale the results. Stryd is actually a really nice device for pacing and VO2 correlation.
I need to start getting people into my “lab” to collect data now.
1 Like
Very interesting. I basically did exactly the same. Using a 10min, 30min all-out test to calculate Vo2max/MLSS and then “guesstimate” VLamax that fits to those values. Also tried to calculate similar to the inscyd protocol the VLamax from a 20s sprint but that didn’t work at all for me.
Had a good laugh that Inscyd actually does this even without a Stryd. Amazing cash-grab. 
However, what metabolic card do you own? Aren’t these ridiculously pricey?
That’s kind of how I approach things…with time trials. But I also use WKO5 to see what their vlamax estimate is. To be honest, using WKO5 vlamax (which I’ve heard is lower than Inscyd) I get a MLSS prediction that’s inline with FTP from WKO5 and CP from Golden Cheetah. Using WKO5 vlamax+15% changes the MLSS prediction by 10 watts, or 3-4% overall power, so no biggie to me. It’s a good ballpark.
I have a Pnoe portable metabolic cart (and a few Moxy’s and Lactate Pro), which is a whole other conversation in and of itself. It’s a decent cart, but it’s consumer-level accuracy no matter what anyone else tells you. It’s good for field testing though and has fewer limitations than VO2Master. I’m looking to purchase a refurbished lab-grade cart (not portable), but that $10kUSD and I’m not sure I wanna spend that money unless I start making a little side money with my lab testing and numerical modeling hobby.
I used mostly Mader 1986 and Hauser 2014 equations and constants, and my graphs look the same. The crossing point vor MLSS seems reasonable, too. However two things seem off to me:
- there is no leveling of for Ox-curve
- the slope of Vlanet seems too high beyond MLSS (working at Vo2max gives infinite/crazy high value)
Have you seen Webers unbublished paper of 2003 that gets mentioned by Hauser?
Those two papers helped me create the graphs, but I also read through Mader’s 2003 paper, which has some different constants.
For 1, what do you mean by not leveling off?
For 2, my spreadsheet does the same thing. VLaNet is very large. I haven’t seen Webers paper, even though I’ve spent several hours scouring the internet for it. I’m pretty sure he adjusts the model, as in some Inscyd plots, the curves look different than from Mader’s 1986 model. Not really the shape, but the values.
Inscyd can do simulations, so I think their model is a bit more sophisticated than just predicting MLSS.
The function for oxidation of lactate should not be a linear function, imo. It should level off towards VO2max.
The slope of Vlanet should be a bit more gradual.
I wonder if the exponents might be the cause and was hoping to find a hint somewhere (basic kitchen biochem tells me that there should be some “rate limiting effect” for both glycolysis and fat oxidation towards VO2max as a function of ph). Maders 2003 paper was over my head 
hence: maybe the exponents “need to scale” as one approaches VO2max?!
1 Like
Hello, it is very interesting what you are doing (similar to Inscyd) however, to determine the VLaMax, it is important to know the VO2max precisely. What formula do you use to estimate it? Thank you!
What do you think of this notebook ?
Seems like everything is there, I don’t think this ppl will have a business model soon.
There are several ways to estimate VO2max from power only testing that give reasonable results. I measure mine, but the VO2max estimate from WKO is pretty good. My Garmin gives a good estimate, too. I would need to find the equation, but there are correlations for VO2max from a 4-minute effort, too.
Measuring VLaMax is the tricky part.
1 Like
I really like that notebook. I’ve been sharing it with people already. It’s only the easy part of the Mader model, though.
However, Incyds model is safe, although ridiculously expensive for the average athlete. I’d even say the free notebook is a much better value than a one-off Inscyd profile.
What the notebook can’t do are the metabolic simulations. With Inscyd, you can input your metabolic test data to tighten up the model. And then simulate your race event or training sessions.
1 Like
Thank you for your reply and sorry for my late reply.
Do you know how WKO5 estimates VO2max?
Regarding Garmin’s equation for determining VO2max, Firstbeat (their partner) has published a lot of white papers on this subject, they are easily available. If you wish I can send you a copy.
What test are you referring to to determine VO2max from a 4 minute effort? Do you take into consideration weight which is a fundamental metric?
Thank you for your answer.
Hi everyone,
Last week I did a new Inscyd test and improved during most of the efforts I had to complete, however the report I got does not match what I feel during efforts/intervals in training and does not match for instance eFTP in intervals.icu.
At 68kg (I’m still 2kg too heavy 
):
18" 810W
3’ 419W
6’ 366W (disappointing compared to the 12’ effort)
12’ 340W
I did 20" and 12’ one day and 3’ and 6’ the day after. In the report they gave me an eFTP of 281W while yesterday I relatively easily maintained 284W for an hour on Zwift at an average heart rate of only 146BPM… Intervals.icu set my eFTP at 314W which is a lot more in line with how I feel in training/racing. In the report they also mention my fatmax is at only 183W which for me is a heart rate of only 105BPM. I also did an 8’ LT1 step test about 2 weeks ago where I kept breaking down lactate even at 200W (so Z1 limit in a 5 zone system) and my first infliction point was only at 240W…
The only difference I can see is that the last time I was 1kg lighter but that does not cause a large difference in W/kg… And my trainer weighted VLaMax 1, VO2Max 10 and anaerobic threshold 1, while my first test was done via Inscyd themselves (online competition of Team Jumbo-Visma) and they weighed everything 1…
If the Vlamax result is high it will drop the threshold and fat max. I’ve seen this many times. If you crack a really good sprint off and one of your other tests (6 or 12mins) isn’t great it will under report FTP etc. Same is true for the opposite.
I’ve got a INSYD coaches account and not used it for 2 years. I think the aerotune one is much better if you want to do these.
But is the vlamax stuff even relevant? A lot of people much more intelligent than me don’t think so.
1 Like
INSCYD test inside on the same bike/trainer as Zwift? Or outside?