Jason: lighter riders naturally have higher watts/kg since your wattage does not scale proportional to your weight. So if the flat ride is about watts/kg they have an advantage
But again, your data set shows exactly that higher raw watts is a higher speed.
Again calculate your R2 and you see you have a linear progression.
You are trying to start a discussion, while your dataset shows exactly that on flats, raw power is most important.
You have a scatter-plot because you have height differences and you use average watts. Also how those watts were distributed during the tt is determining average speed (fast-start-dying in the end, negative split etc)
This is a non discussion, because your dataset proves the exact opposite as your title suggests
I guess it would be interesting to be able to compare to some online speed calculator to see what the differences are at each weight/height vs Zwift.
There seem to be a bunch online, is there one that people think is pretty accurate? I think what weâll find is everyone is going faster than they would IRL because Zwift assumes ideal CdA and drafting conditions, perfect bike set up etc. but it would be good to see if there was some bigger difference depending on weight or height than should be there.
zwift have more or less said what model they based the physics on, i canât remember which, or if there even is more than one viable model. i just turn the pedals, but it seems to me that any major differences are in the nuances, rather than the fundamentals
no wind
static preassigned cda values
no elements
no danger
a more controlled environment
etc.
Ehm no? The watts/kg graph has a much, much tighter relationship than the absolute watts one.
Colorcode or something. That the high wkg are in the low watts area
Your second graph proves exactly that raw watts makes speed. Perhaps not as beautiful as the wkg graph.
But with the 2 graphs I canât tell that a high wkg is putting out low power. So in your hypothese the first graph doesnât add anything.
You say raw watts arenât determing the speed. If that is the case, your second graph should be a circle shaped cloud, which it isnât
For a full theoretical relationship the dots could be more on the line. But as discussed before in your second graph, height or power distribution during the tt wasnât taken in the equation.
Put a linear trendline in your graph and the R2 shows that speed is scientifically dependent from raw watts.
Yeah, I was looking for it but didnât see it, if someone finds that info it would be good. Zwift Insiderâs analysis is just for a specific rider power etc, so if we know the model that would be nice.
1 Like
As opposed to�
Your test where we donât even know if a rider was going up or down a hill when the data was taken?
Your test where height is a variable?
Your test where bike/equipment was a variable?
The known fact that literally none of this was true steady-state.
Again your scatter is âprettyâ to look at, but it also is littered with an extreme amount of variables as far as anyone here is concerned.
I didnât do a test, I was asking for more/better data.
oh shoot appologies, I didnât realize I wasnât quoting OP!
6 hours and 120+ miles and up ADZ today, Iâm clearly wiped 
1 Like
Known by whom?
This whole conversation is completely pointless without some alternative model for CdA.
Which part do you disagree with? The basic equation for speed of a cyclist is tuned to watts and CdA on flat surfaces and w/kg on hills.
Iâm not stating raw watts donât add speed. Obviously they add speed else there would be no reason to rotate your pedals. Iâm stating they donât add enough speed. Because as you say, the watt/kg fits more nicely.
Since raw watts and relative watts are not independent variables- one contains the other - I donât believe your solution would cut it. But curious to hear your hypothesis on how it would look (I can also send you the dataset if you want).
what this graph suggests imo is that the relationship between absolute watts and speed is mainly explained by the fact that more absolute power causes higher relative power as well.
If absolute power was the main explaining variable, than you would expect that variable to have the strongest relationship with speed, as opposed to the other way around which you see in my graphs.
Of course this is not conclusive evidence, that would require a complete alternative model fitted to zwift data.
Both graphs are based on the same data and have the extra variables that are indeed not controlled for.
Still the watts/kg explains almost all the variance in the data and shows a much tighter fit on wattopiaâs waistband. You can see that because the individual points are much more tightly concentrated around the linear fit. The unexplained variance (which is represented by the distance between linear Fit and the dots is very small). A high fit also implies that all of those variables you suggest have a marginal role.
Hereâs the point: it is clear from the graphs that watts/kg predicts speed better than absolute watts, which is opposite of the expectation on a time trial with 95 of elevation over 25.5 km.
Sorry I wonât make a phd level analysis if thatâs what would convince you. Itâs my feedback to zwift that this needs to be looked at.
It should be tuned that way, but with the limited information we have, it sure doesnât seem that way right?
then what, you just want other people to agree with you, or to do the hard work and confirm your bias? what exactly is the point of contention? i might have been ready to agree with you, and i even gave you some suggestions as to what factors might influence certain body types more than they influence others outdoors that are largely nullified on zwift for you to consider, but now i just have to ask you this: do you just not like light people? thatâs not a loaded question either, because i happen to meet a lot of people on zwift who donât, but canât, or wonât elaborate.
CdA depends on rider size in real life and also zwift. Zwift models CdA as a function of both weight and height which seems entirely reasonable in principle to me even if their particular function may not be perfect.
Do you not accept that itâs sensible for zwift to model CdA as some function of weight and height? The âAâ in CdA is quite specially cross -sectional area.
Lol, obviously I donât dislike light people, please donât make this personal.
People donât have to agree with me, this forum is just to discuss, maybe provide supporting and/or counter evidence.
Itâs just infeasible to account for all variables, since it would mean I have to fully reverse engineer zwifts physics engine.
no, itâs really not obvious. hey, discuss away. i really donât think a scatter graph and some sentences that start with the word âclearly,â are going to go anywhere though
You are changing the discussion topic from zwift physics to attacking my intentions, which you suggest you know better then me. This is a classical ad hominem fallacy.