2021 Zwift Classics Stage 8 London International Classique

The final week of the 2021 Zwift Classics! It’s going to be 4, 5 or 6 laps of Zwift’s London Classique course depending on your category. The race starts from the start pen and finishes when you’re done.

Starting from a makeshift parking lot opposite the Tower of London, the 5.1km lead-in follows the north bank of the Thames and under Charing Cross station. Be sure to prep for a sharp right up Northumberland Avenue (400m 5%) into Trafalgar Square and then another right on to The Strand. The course doubles back on its self, back through Trafalgar Square, through the Admiralty Arch and on to the Mall where the race starts proper.

The start finish banner on The Mall marks the end of the sprint segment (all passes will contribute FAL and FTS to your points tally). Then it’s on to Buckingham Palace, passing The Canada Gate (presented to London by Canada in memorial to Queen Victoria). The route makes another U turn and passes right in front of the Palace before circling St James Park on The Bird Cage and on towards Westminster Abbey. The race then swings left on to Parliament Street and Whitehall passing the Cenotaph (The National War Memorial), Downing Street, Horse Guards and back on to the Trafalgar Square roundabout. Now its back on the Strand and just a few kilometers to complete lap 1 at the Start/Finish gantry. That’s one lap done - just a few more to go!

What category were you assigned and how did you experience the race?


I’m in C3 again. I am in such a state of anticipation to see if WTRL has finally figured out that having 5 to 10 riders with 20 minute watt output 100 watts higher than the majority of the field of 40 results in a less than stellar experience for the little kids here in these YAC events. Please feel free to astonish me WTRL.

This of course ignores that massive inbalance of the individual pens. C4 in the West Oz events has had about 8 riders each week. I think C2 had 15. I had 40 of my closest friends in C3. Spread us out otherwise this is going to be another bloodbath for us low B and high C racers. I don’t anticipate astonishment happening here. We’re all “PERFECTLY placed”.

1 Like

Highly technical course…



Daz Carter again in C1. Good to see they’re really learning from this experiment.


C2 again. This will be my 78th time for this venue, of that, 48 times for 6 laps. So I’m looking forward to it :slight_smile:

The top of the Whitehall climb is the “normal” selection point. You always want to be in the front half of your group when you get to the top.

On the odd occasion small groups will also attempt to get away by accellerating off the front after the U turn in front of the Palace (where the speed drops to 33kph.)

Literally zero evidence that anything has changed as it’s gone along. If they’re tweaking things, it’s not showing either at the high or low ends.


I think it was explaining quite black and white in another thread that there is no requirement to tweak things as its all rather splendid and working as expected. The fact it has resulted in zero significant change really , and certainly clearly visible similar levels of outlier issues (like existing process) should be therefore quite easy for the retrospective to come to the correct conclusions (you would hope !)

Personally I don’t have a problem with what they are doing , because I don’t have a problem with what was there before. I just think its all been a bit of a gigantic exercise in getting no where at all , since no feedback was really wanted (or at least entertained) . The evidence was clear from probably the first two weeks that this was not really going to solve the fundamental issues and as such we should have moved on to tackle those ( and by all means continue with experimenting with predictive categorization as improvements however small are always welcome )

The value here is not really a good return for relative effort , the only people who will see benefit from it are new riders who will get a marginal improvement in there initial categorization. We really really need to start tackling the bigger issues , for those of us who already are well established zwift customers and event participants.

The good news is that this series has pretty open and shut shown this .

  1. The important thing is to enforce pen sign up not necessarily what predictive math is used to get there
  2. That any predictive math is just that , its not the same as actual racing . I think even this week the zwift community live guys joked about it being like we don’t have to actually ride any more , why not let the computer decision who will win . There is clear evidence week after week that categories once set on sign up to zwift/racing have to be governed by some sort of results modifier . (i.e its the only way the “** Carter” issues can be addressed )

I hope Zwift are listening to that because if they get those 2 things in place 90+% of the noise just goes away…


Onto the Race itself . Clearly not going to be much good for me , YACR , too short and no longer climbs to hamper the sprint types , That said I have won on this circuit more than once before by going off the front early , but not with the level of people I will be racing in this one , and the likelihood of a very small field again if they don’t change anything . None the less great workout and enjoyment expected and event trying to stay focused at least to hang in long enough to go for a long one to see if I can get a big enough gap (no chance) .

For all the carping about the various things that have and have not been done (Muir and the Mountain being dropped ) this series has been relatively speaking good “fun” for me , and after all that is all that counts. I think it probably caters well enough for the limited audience who ride it.


Yeah, poor ol Daz, I think he would prefer to race with us in C0 actually

Here we are again a week later with virtually the same cast of characters riding YAC that’s pretty much a pure power course. The little hill isn’t going to create much selection and rest of the route is going to be Sprinterville. I think we can ignore w/kg this time around but that’s still a huge range for C3: 3.2 to 4.5!

Here’s the salient points for the WestOz C3 event:

  1. As of this morning, our A rider, the redoubtable Mr. Evans has not signed up. In this case though, he wouldn’t be the highest 20 minute power rider at 371w but another rider with 381w!
  2. Still the range of power in C3 is 202 to 381. I’m in the bottom 10 or so at 231.
  3. There are 40 riders in C3. 5 riders in C4. 5 riders in C5. 20 riders in C2. So what’s with the ridiculous spread of power in C3? Surely we can spread this around a bit? Isn’t that part of what Autocat does so well?
  4. As a friend of mine who has very similar power to me (and is in WestAm C3) said, “I’ll be riding sweep again this week.”
1 Like

At this point I’m starting to think that C3 is supposed to be the middle placed dumping ground. The vast majority of riders will be autocat into C3 and then Zwift/WTRL will rely on the upgrade/downgrade system to move people out of C3 into their proper category.

How else can you possibly explain the huge spread of power in C3 (from 3.2w/kg @ 210W to 4.3wkg @ 342W in C3 Eastern Australia) and why C3 usually has double the number of starters of any other cat?

Hopefully this is the end of autocat testing and they can start work on the far more important results based ranking system.


Except that never happened. They said that was going to happen and it never did: C3 still has 5 or 10 riders that should be in C2 and 5 or so riders who should be in C4. It literally makes absolutely no sense whatsoever. I have personally lost respect for WTRL and their events. I’m not interested in participating in any of them after tomorrow and I don’t imagine they’ll miss me. That’s fine 'cuz there’s lots of other things going on and I’m still hopeful Zwift will get this straightened out eventually. Autocat in its current implementation is not the answer.

1 Like

For better or worse I’ve been enjoying them. Large starts with a reasonably narrow range of power. My only criticism was possibly too much climbing :slight_smile:

And I do look forward to a test of a results based ranking. Which could be ZwiftPower ranking to start with. Just deliniate some boundaries.

It would (IMHO) also be nice if Zwift just did not let people join and event in a category lower than their minimum suggested category as specified in Zwiftpower. That coupled with a way for us race organizers to say that our races are open to only Zwiftpower registered racers.


In hindsight it becomes clearer what really happened, what really happens right now and what will really happen next.

The assignment given to/the ambition of this team of consultants, you know the ones with the vaaast experience of medical research and whatnot, the ones that were supposed to do what Zwift themselves could not, was just to create and test a model for categorization before races. And only that.

Categorization is never complete without promotion/demotion, but that was apparently not the scope. Early on we wondered whether we would see quick upgrades during season. To some it seemed there were some since some people moved up and down in categories a little. But there never was. It was just them tweaking the knobs of their model a little from week to week, resulting in different starting pens for one and the same rider sometimes.

We have been used to potentially very fast upgrades, for better and for worse. In the ZP system you can theoretically go from e.g. a solid B to A in a single day looking at the mechanics of upgrading. That’s one way of doing things. But it wouldn’t be unreasonable to keep people in their categories over a very short (half) season of 4 weeks, 4 races. Hence you wouldn’t necessarily see upgrades during this season anyway. Only then you’re stuck with seasonal upgrading, which I doubt they want because it is ill-suited to the wide spread of zwifting patterns among subscribers. But that was never the focus anyway.

The focus, the assignment, was just the initial placing of riders in suitable pens in the best possible way, using moronic performance measures. And they failed of course. Like we predicted (knew rather) they would. Because you can never do it right, it is doing things the wrong way for the wrong reasons. That is bound to fail.

I think they have realized that too now. I doubt they are even tweaking any knobs this week. Well, maybe a little, just for show, so they can report hours and get paid. But I think they have given up and I think Zwift has too (they must have or they are even dumber that we thought).

So what happens now? What is the next step? Well, even if they had got it right, or right enough, there would still have to be testing of upgrading in the system, which requires another season. And since they don’t take control over their own product, it would be thrown to WTRL again. Maybe test that later. In a coming season. In coop with WTRL. Some time.

But they didn’t get it right. So what now? Now they have no idea what to do. Complete failure. Arguably worse performance than ZP itself, apart from the cat enforcement. Can’t go ahead with that. Where to go next? Well, you can’t give in to the pressure from an aggressive little possee on the forum, can you? Where would that take you, if you handed over business strategy to the whims of a loud minority among subscribers. So openly experimenting with results based categorization next is out of the question. And there really aren’t any more leads to follow in performance categorization at this point.

Hence they most likely default to what Zwift does best, a tried and reliable concept that I described back in March:


Welcome to Zwift. Fun is Unfair.


Relative newcomer to the forum and was reading through the initial topic with regards to Autocat. As you all know, that had quite some interesting feedback but was closed as the topic according to WTRL was not providing “enough value” and plenty of different and more positive (less critical?) feedback was received from “other sources”.

Interestingly, the number of riders in each and every category have gone down in a straight line since the first 2 weeks or so. So whatever that “different feedback from other sources” looks like; every week an increasing amount of people seem to stop signing up to the Zwift Classics. That MUST raise some eyebrows in the final evaluation, assuming that the overall goal was to attract more instead of less people to racing…

Hopefully feedback in that initial topic will be appreciated after all and taken into account for further development. Marketing-wise, the “Autocat” terminology has probably done some harm to a significant part of the user base (= those not signing up anymore…) which will make it difficult to attract those users again.

Anyway, good luck all with the final Classics race!


While some people did say they stopped racing because they didn’t like the cat system, it’s important to be wary of making such confident conclusions from statistics like those.

It’s common for a series to have more people join the first events and then have numbers gradually drop, because few people want to start a series which is half way through/nearly finished, so that’s a factor we should take into account. I do agree though that it’s unlikely the feedback on this forum helped attract new people.

As for me, I wasn’t a fan of the first few races (except the fantastic 3-person race when I was uncategorised) but stuck with it and enjoyed the last races, especially when I changed from the European to New Zealand time zone.

Hmm… looking at the start lists I realize I might be (slightly?) wrong. The lists look vaguely different now. Might be because of series dropouts but…

Maybe the consultants haven’t actually given up just yet, like I thought they would and should. The reason is I almost get the impression they have included weight in the model now. Which would implicate that they did listen to the forum (that would be a first in Zwift racing history!)

It could be a mere coincidence and wishful thinking but weights are fairly well spread across within the various cats. It’s not the usual scenario of standard Zwift races, when you sort the start list on W/kg, where you typically have a bunch of heavies who can do the known upper limit for the entire race at zone 3 and who you know beforehand are going to take the podium.

On the other hand you get a few lighter riders with scary W/kg in their cat to counter these heavies. These lighter guys look out of place at first (“why is that guy allowed in this cat??”) but then you’re stuck in ZP thinking, a thing of the past that must die, whereafter we drive a stick through its heart, burn it and bury the ashes deep. If you are to make the heavies beatable you need to present lighter guys with somewhat higher W/kg who are allowed to do their best without getting DQ’d (it just sucks when you’re not one of either but that’s a different story and no valid critique against any cat system).

Still, without a known upgrade system these races are absolutely pointless, meaningless of course and we’re just guinea pigs (we sort of knew that though).

And AutoKitten still doesn’t address cruising, I can see that in the lists. Then again, you can’t address cruising without a known upgrade system. So a moot point perhaps although I don’t particularly like the… how to put it… involuntary presence of known cruisers (they never picked their cat this time, it’s just that they always cruise when they race which is an advantage).

As initial seeding for a new or returning racer AutoKitten might not be worse than any other performance categorization. But the next step is a results based promotion system to go from there and hence an idea of what it takes to move from one category to the next. Then I’d be satisfied.

Still a lot of potential problems with cats with skewed sizes, wide spread of W/kg, non-comparable cats between time zones etc. But it will be interesting to study the results later tonight nevertheless. I’ll hold back on the negative tone for a little bit just yet.

I was sick for several days and planned to sit this one out, but now I’m tempted to sit in as pack fodder… hmm…

1 Like

Looks like tonights race might be a bit quiet as Down Detector says Zwift has an outage and I certainly cant connect to it .

Outtage seems to have happened when C1 (Central Europe) was finished. Can’t see my fitfile, but according to Zwiftpower I finished 3rd (was 34th or so).
I’ll take Podium every day :stuck_out_tongue: