Autocategorization Test Events & FAQ

I was in C4 at 18:48 UK time and, with the joint lowest FTP figures before the race, I didn’t expect to do that well, and I ended up about 95th out of 120ish, which wasn’t too bad. What felt different was having the knowledge that something was being done to make it fair, even if I was being fairly beaten! I’ll hold out judgement until I’ve done a couple of these races.

Biggest disappointment for me was that Zwift crashed after I’d finished, and I lost the ride, which is something I haven’t had before. No ride on Zwift Companion, no data in Zwiftpower, and I’m not even on the results in Zwiftpower yet. Fingers crossed at least the ZP result will appear.

I think one guy has slipped through the autocat cracks. In C4 Eastern Australia the winner averaged 4.2w/kg 294W for 42 minutes and managed to finish 25 seconds ahead of the rest while riding most of the second lap off the front solo.

Another brutal start to the race with the first 5 minutes at 4w/kg. I lost a bit of time on the KOM but somehow managed to get back with the leaders before the sprint. Stayed with the front but the second time up the KOM was too much for me.

Looking at the results the w/kg for the riders in the top 49 are mostly around the 3.1 to 3.4 range but there is a 6 minute time difference between 2nd and 49th (not sure why ZP only shows the top 49, I could’ve sworn there was 70 odd in the event, was Zwift counting the mens and womens fields for its numbers?). It just shows that if you can’t stick with the leaders during those critical points your whole race can be over before you know it. It will be interesting to see how the flat races play out.

1 Like

I was C4 and usually a 2CV B at lower-middle end of B. I found the racing mostly as hard as usual, but the main difference was I managed to stay near the front and finish top 20, so in my case anyway the categorization seems quite accurate.

I did wonder if WTRL had an oversupply of buritos from other races though, as I kept getting them through each KOM and sprint. Trying to get rid of them without disrupting the group you’re in is a bit like a social ride after eating too much garlic and beans the night before.

2 Likes

There’s a button On the companion app to drop/throw away the power up. I did that with all my burritos.

Thanks, I had not realized that (funnily enough despite having done 86 races according to ZP, this is my first race that had buritos - possibly because I haven’t done any criteriums yet).

2 Likes

Hi @Dave_ZPCMR, I think I’ve found a pattern/bug which might show why you and I aren’t appearing on ZP results.

In both your C4 race (at 19:48 UK time) and mine (18:48 UK time) the ZP times show a large gap between 50th and “51st” places, which is because lots of riders have been omitted from the results. According to Zwift Companion you finished 59th, putting you in the “twilight” zone, while people who finished behind you and me in each of our races were included.

2 Likes

Good spot, that’s a bit weird. ZP bug?

The system looks at more than your 20 mins power. If you were to cruise around then smash 1/5 min power like a lot of riders used to do in zp cats in theory the autocat system would detect this.

C4 race today in Central Europe at 19:48
Much better experience than for the race test in C5. This time tough opponents much better than me to challenge me. I recognized the names of a lot of top C’s in the pack at the start and also some old C passed B not long ago. 120 racers at start line in July, that happiness.
Big problems in terms of rankings on ZP or WTRL … Many unclassified on ZP, all the riders who arrived between 50th and 51st place are missing. And the points awarded on the WTRL are false.
Very satisfied with the direction taken by the autocat. For me it’s a step forward

C4 race (America East).

Not bad. I was ranked 15th in terms of ZP racing ranks prior to race and finished 17th, 1:36 mins behind the winner. Improved by 1min power by a good margin because of the punchy nature of the course and trying to stay with the front pack on the first KOM climb, which is where the first split happened. After the split, I fell back to a strong group of “C riders”. There were 4 guys in a team who would overcrowd me in the sprint, so I decided to attack early once we turned into the straight to the banner. They were not expecting 700W bomb up that short hill, so it worked. All in all, I haven’t pushed myself this hard in a while and had tons of fun doing so. I don’t know if it was the autocat or the course that made this happen, that remains to be seen.

1 Like

Catching up after a very good but chaotic day!

@Sam_Splash - not sure if you are familiar with WTRL and how we work but when a new series comes along with some significant twists to a format or completely new (In this case Classics is new to WTRL and this was the first real main stream test of Autocat by WTRL), we tend to hold back on results allowing some to go out, feedback to be collected and of course a small crew of avid calculator lovers to check lots of elements… There are usually several ‘bugs’ (errors as they are known as here) that have to be chased out and I do believe that less than 24 hours after race 1 started and over 5,5500 racers later - we are just about there…ZwiftPower has encountered a few issues today with its processing but that is to be expected - it is having some quite substantial work done to it currently (better in the quieter summer than in the rush of the fall/winter season!). From what I have seen of ZwiftPower, it has for the most part caught up. Many of those people upset that their result wasnt on ZwiftPower did manage to find themselves on our results and while its taken a bit of time to fix the points system, everything looks ok now.

6 Likes

That explains why I am also missing from the 18:48 C4 results in Zwiftpower as well, finished 59th.

Just done my first race with this new system. Im towards the top end of Cat D and ended up in C5. Blown away at the start (I usually manage to stay with the lead group). Winner pushed 4.7 w/kg. I PR’d both laps and yet still came almost last.

I confirm everything is back to normal. What a day it must have been! Thanks

1 Like

Women Top D, was put in C4 Eastern Europe, mostly C riders.
Luckily apart from 2 riders that were true C, the rest was not so quick. I am a very heavy rider, so I understand that it is difficult to find the right category if many racers are half you weight and a good 25 cm shorter ;-).
Maybe to check C5 Eastern Europe women, the winner is a C almost B lady with a very high output (3.74 wkg).
Here Autocat missed her, but probably it will settle over time.
Not sure if 5 cat for women are not too much. 10 people in my time zone yesterday in C4…
But thanks for trying to improve the system even for low cat races like me!

1 Like

I’m a low-to-mid B (~3.5wkg) and this is universally what seems to happen in races. High Bs sandbag and ride at the low B pace until the first significant climb, then ride off at 4-5wkg 5min output, which is beyond capability for almost all low Bs, except true lightweights who have high power/weight but low absolute power. In most races, my experience has been that heavier low Bs get dropped about 1/3 through the race and form a second group. True new Bs are dropped even earlier and ride several minutes back, and the high Bs and climbers are conserving energy for the first 10-15min in order to then ride away from everyone else in the rest of the race.

2 Likes

This seems to have been about how my C4 race played out, too. Some high Bs who cruised up all the climbs at 4-5+ wkg, some low-mid Bs who formed a second group that finished about 3 minutes back, and about 15 stragglers who were likely C racers that finished 5-15 minutes behind them. I do not believe the autocat feature treated those folks fairly – no race of <30km distance should have 20min delta between first and last place.

2 Likes

Thanks for the feedback. I think this will all help the team make this even better.

I had a quick look at the last 3 places and it look like all 3 was a lot slower than the usual race pace that they can do. So I assume they gave up or just tested the waters. If you discard these 3 you get a spread of 9.5 min. That does not seem to be uncommon. That was not an easy course once you fall out of the bunch you loose a lot of time.

1 Like

Some feedback on the autocat - it would be nice if we could see our classification without having to sign up for an event (similar to how you can see your category on Zwiftpower), unless the intent is to hide this from people. I’m fine if we can see the class without the underlying metrics (unlike ZP where you know 95% of your 20 minute power defines your category), but it’d be nice to see if you upgrade/downgrade after a race. Similarly what’s not entirely clear to me, does the course you sign up for influence your class? I.e. is the classification dynamic and I’m, for example, a C3 on a hilly course but a C2 on a flat course?

4 Likes

Yeah, I did America West C4 and have not raced or done any interval training in well over a month. Fitness is at a pretty low point for me right now.

I’m pretty sure the auto-cat picked up my “almost B” stats from 45 days ago or so to lump me in with this group, since I haven’t been zwifting this summer. I had no chance and was dropped within 90 seconds.

2 Likes