That’s great stuff. As to your question why weight matters in your model, it’s well known in zwift physics that a heavier rider with the same W/kg goes significantly faster than a lighter one. eg 80kg at 320W is a lot quicker than 60 at 240. As to the underlying reason why this is the case, it seems that zwift uses weight as a factor in calculating frontal area and therefore aero drag.
1 Like
The autocat system as currently implemented does not appear to adequately account for rider weight, therefore I suspect it will (unless changed) provide extremely challenging races for most women, as they are generally lighter than men. The current ZP categories address this problem with the 250, 200, 150W thresholds, which aren’t perfect of course.
1 Like
Another good race in Western Australia C4. One strong rider got a 5 second gap to win it (looks like a possible mis-categorisation, not their fault so a fair win regardless) and then there were 18 riders contesting the sprint for 2nd.
Another group behind that of about 22 riders which I think got detatched during the lap 3 prime, then a few smaller groups following. It’l be interesting to see how many riders from the 2nd/3rd groups played the game and picked up points for fastest laps.
All-in-all a very competitive race, so thanks to the organisers for putting these tests together and getting us closer to a proper race system.
Totally disagree with you Sandy !! I was in the same race as you and us Cat C’s were there to fill the numbers. The majority of the B’s prob had great fun but us lowely C’s got whipped. Not 1 Cat C has made the top 10 any any C4 race today.
last weeks C4 seemed spot on with a few stronger B’s but more C’s, this week was way out imo.
I’ll see what next week brings
Hah, yeah you have a very valid point, I guess it depends on where you saw the race from! Given that we’re working with what are essentially still static categories then we’re always going to have people nearer the top/middle/bottom of that category, which isn’t a great place to be. I expect that if I was moved up to C3 then I might have a similar experience to yours today…
Ideally we’ll get to something more dynamic as has been suggested already where we all get thrown into a pot and then categoriesed just before the race based on who has actually entered rather than what performance boundaries we fit in.
Tangent, but that’s not the reason.
More raw power = more speed (more energy to overcome air resistance, friction etc).
W/kg is only relevant once there is a gradient, as then you need to overcome gravity. More mass = harder to overcome gravity. So at the same w/kg, a heavier rider will be faster on any gradient simply because they are overcoming gravity in the same way, but putting out more power. They will be significantly faster on lesser gradients, a smaller difference on steeper gradients.
If the raw watts are the same on a completely flat course (e.g. 60kg @200w and 80kg @200w) the 60kg rider will be faster on Zwift - due to an assumed smaller CdA.
If somehow in real life the CdA’s of both riders were the same, watts the same, all other variables the same, but weight different, the lighter rider would still be quicker - due to the force required to accelerate the larger mass initially.
This is really cool btw, bravo 
1 Like
Yes you’re right I completely garbled that. What I was explaining is why raw watts doesn’t work as a speed predictor on flat ground, which some riders are a bit surprised to find, especially if they have used sites like Steve Gribble’s power/speed model.
I suspect (back on topic!) that the current autocat also doesn’t make much allowance for weight, there were some very lightweight (high W/kg but low absolute W) riders put into fairly challenging categories in today’s races. Basically, D cat (<150W) riders in B/C cat races. I wonder how they got on.
1 Like
Got it, I figured it was probably just a typing mixup!
I did the Crit City Slam (Western Australia) C4 today, my first foray into the autocat system.
I didn’t actually realise I was a “C4” until after the event, but in any case it wasn’t what I’d expected. For me personally, it was a worse experience compared to a normal C race.
30 seconds into the race I was pretty much dead last. 60/60. What? Huh? No way! I’m not fast, but normally I’d expect to be at least mid pack in a “C” race. OK, I didn’t go out hard hard like I should have, but I was expecting a closer range of abilities.
Honestly, I assumed it was a mixed category race, and a bunch of A and B riders had gone off the front. 
Anyway, winners were doing 3.5 - 3.7W/kg which made for much less of a competitive race for someone who can do 2.9W/kg these days.
Finished 49/56, 4 minutes behind the winner which for me was disappointing. I’d probably just stick to standard C races if that’s the norm (but of course I need to do some more to be sure).
On the up side, if I order the result by 15s power I’m 8th. 
2 Likes
C4 appears to be a mix of mostly mid-low B riders and high C riders. So the Bs are now in a race where they are more likely to be nearer the front instead of getting dropped in a regular B race, and the Cs are basically getting an experience of the upgrade from C to B where they go from the front to the back of the race.
In my opinion Autocat largely fixes the problem of sandbagging, but continues the legacy of fixed boundary categories. That’s still better than what we have today, and perhaps is the limit of what can be done within the limitations of Zwift’s current event system, but it still leaves room for improvement.
1 Like
Seems like a good example of how hard it is going to be with so much trial and error and heavy lifting to get ourselves into a good place categorising racing by numbers .
How much easier would it be to get us moving towards more competitive racing if those who won the race get enough points to eventually get moved up to C3 and if you continue to find yourself in gruppettos at the back will over time potentially be moved to C5 … where you will either start winning regularly and get promoted again or find yourself in a more competitive enjoyable event .
Note : I am not suggesting instant promotion for winning a race etc etc . Everyone knows how bike racing categorisation works in most national and other cycling organisations . I think we need to move to more focus on that approach to improving this situation sooner rather than later .
Zwiftpower already has a concept and handling for race points . It may or may not be usable as is (probably not ) but most likely just needs good product requirements and refinement work to be done .
1 Like
How’d it go for you? That field was one of the strongest C4 fields in those time zones front group finished almost a minute faster than all other races.
On this again i think I’ll start looking for more once ranking points and/or wins are included in the auto-classification. Today in that East Aus C4 race felt like a spicy B race with a large field.
FWIW, I checked ZwiftPower before the race, and it had me listed as a “D (nearly C)”. By that estimation, maybe I should have been in C5.
It does look like you were either mis-classified (not many races in the last 90 days) or were on the edge of C4/C5. Given that someone has to be at the cut-off point, what would make the race experience better for those at the lower end?
For example, if at the start of the race you were told that you were ranked 60th out of 60 then perhaps that would adjust expectations, so instead of being disappointed to have been dropped, a rider could be happy to finish in 50th place?
Maybe we should all be given race numbers to identify our predicted finish position? Then you’d at least have targets to aim at to try and beat, even if it wasn’t for the overall win? Are there other things that Zwift could do to improve the experience for those not at the front?
3 Likes
Could it be that you had a bad day or were not sufficiently warmed up?
How was the race at the back? Did you have riders of similar abilities to ride with?
It could be fatigue. I did a decent ride on Sunday (5.0 Aerobic benefit on the Garmin, 20.1 strain on the WHOOP - both telling me I was overreaching). Did a gentler recovery ride yesterday but still had a headwind for a lot of it.
But a bad day would still only take me to 3.0W/kg at most I expect. I was averaging 2.9W/kg in the ZRL a couple of months ago.
The group at the back was pretty good yes. We had a pack of about 6, which swelled to around 7 or 8 at some point.
And to be fair, although my average for the whole race was 2.8, I did less than that a lot of the time to stay with that pack after about 1/3 of the race. I think we all accepted we were the group and there wasn’t any point going harder to try and get off the front - there was no chance of getting on to anyone ahead, they were miles away.
The cobbled climb made for some decent digs.
But my 2.8W/kg is still about typical for races I’ve done this year.
(places in the ZRL aren’t all meaningful as several were TTT with staggered starts).
My point being that I’m a 3.0W/kg rider at best at the moment, and all my race results this year support that. Being in a race with people doing 3.5-3.7W/kg is worse than being in a race with winners doing 3.3W/kg.
If I was misplaced, I’d say the autocat should look further back in history and maybe it would have had a better sense of my ability.
And, to be perfectly fair, it could just be wholly a case of expectations. I went into the race expecting a better, more competitive race from my perspective. If the boundaries have been changed, that expectation might have been misplaced.
Still at least I took the field sprint out of the 5 of us left.
1 Like
I agree, this would be brilliant - it would mean everyone has a realistic target (improve on your seeding) and would create more “winners” who are pleased with their result. The race numbers idea would take it to the next level.
1 Like
This is a fairly flat course and absolute watts should be more import than W/Kg.
I don’t know. There are many factors in play here.
It was interesting to hear your feedback.
Was allocated to c4 but this felt too easy (with all due respect 
) There’s a way to promote yourself, says the wtrl website. Couldn’t find how though?