Anti sandbagging and other areas that need development and communication

It doesn’t seem to get noticed or discussed much but I’m pretty sure there is a cornering effect. Do a steady (erg mode) interval round the u-turn in London if you don’t believe me, and keep an eye on your speed. It could be stronger.

Indeed there is, but I believe that point in London is the only spot.

2 Likes

Suggests it shouldn’t be beyond the bounds of possibility to roll it out more widely. Is there nothing on the u-turn on the crit course? Shallower bends should see something too of course.

1 Like

And only in one direction.

Yeah I agree completely, slowing for corners will add a lot - particular if larger groups slow more on tighter bends. Gives the breakaway more of a chance.

I link this one up with descending - a large pack descending accelerates to ridiculous speeds which means an attack over the top of a climb is a waste of time.

1 Like

Vote up Speed limit in corners

Bumping up this request, because I’m hoping that a bit of reflection and future directions can be shared.

If time and resources permit, I think there would be an incentive to continuing the Classics season and Autocat testing on Tuesdays also during ZRL. It would allow teams to rotate their riders and have additional riders on stand-by so that in principle all teams can start a race with a full 6 riders. Those that are not needed for the particular ZRL race could then enter (perhaps using the same course?) the Classics event. This side-by-side approach would hopefully allow further development and testing of the WTRL Autocat racing system.

This has been race 6, and although I have not participated in any race I did keep an eye on the threads, the signup list and race results for my category (and have seen how riders I usually race with perform). Based on this, I still think there is merit to using predicted(?) average course speed (though maybe based on solo rather than group speed) for initial classification, if only to take away the emphasis on rider weight in the traditional categorization equation. From the outside, it seems to perform decently, although some challenges seem to exist.

  • The Autocat seems to work brilliantly technically;
  • The far majority is probably in a (as much as possible) fair category;
  • Expectation management for 6 categories is that the tail end of each category will get split from the main bunch at some point during the race, and there will always be a tail end no matter how these categories are sliced;
  • Despite the above, some riders are probably under/over-categorized and examples of this have been given in the various threads;
  • (Some of) these riders do not seem to move categories in following races, even when they ride off solo to victory or are dropped very early on in a race (i.e. have ‘demonstrated’ their true racing ability).

I cannot fill in the priorities that you and your team have defined, nor am I aware of the many challenges that I’m undoubtedly not seeing nor knowing about, but it would be nice if performance/results-based movement of riders between categories (or any other clever alternative) could be tested as a way to refine categories along the way. Obviously a one-off bad performance or race win should not result in a relegation/promotion to another category straight away, but it has to be possible to find a middle ground.

It would also be nice to track if upgraded riders were at the sharp end of a race (as you would expect), and how these riders get along in the next category. None of this needs to be perfect straight away, the reference still is the silly w/kg system that is notorious for upgrading and maintaining the wrong riders. I suppose I am just hoping that whichever new categorization is used reduces the performance gap that most riders will have experienced after being upgraded, and that some never get to close (and eventually may drive them away from racing or the Zwift platform altogether).

1 Like

Well Marco, you’re seeing something different from the rest of us who’ve been racing the series. Take a look at the Innsbruck thread and tell us exactly how “Autocat seems to work brilliantly”. My experience has been remarkably less than stellar.

With technically brilliant I’m referring to: you click sign up, Autocat does some background magic in its databases and Voila you can enter only one category - the one you’re assigned to. All this does not take ridiculously long to decide either. As far as I know this has not been the case in ZRL nor TTT PL (it didn’t go further than closed events you needed a race pass for, no?)

To be clear, this is besides any discussion on which boundaries should be used and if they are right or not. I’m afraid such fiddling can go on endlessly, as I’ve also expressed before this experiment was started, and there will always be some percentage incorrectly classified. Just hoping to move to a next step and let race performance smoothen things out while further fiddling is done in the background to get things more accurate in the first place. I see the lack of rider movement as a bigger issue than how the cake is sliced and hope this can be tackled next, rather than putting things in the fridge until ZRL is done again.

FWIW, your examples are amongst those I referred to with “Despite the above, some riders are probably under/over-categorized and examples of this have been given in the various threads;”

That is nothing new we used that during the ZRL and the WTRL platinum league.

The testing in the CLASSICS is far from being ready to use. If you can enroll one day and the system put you in C3 and the next day it put you in C2 for the same event then there is something wrong.

A simple IF statement in the event sign up would do better. That we have been asking for since 2017

5 Likes

I’m already signed up to Zwift.

Then I have to sign up to Zwift power and connect that account.

Then I have to sign up to and link a WTRL account.

Yeah…I wonder why only 8% of people race…

10 Likes

And in order to sign up to Zwift power and connect the account you still need to jump through the loops of having to find your Zwift Id and adding it to your Zwift name for the initial connection despite Zwift having taken over Zwiftpower. One would have thought that part of the process could have been automated after the takeover.

5 Likes

The categories are so broad that this statement is pretty much meaningless. Had I turned up yesterday I would not have been quite last in my assigned category of C1 but also I certainly would not have won C2 or C3, though I should at least have been in the front group of the latter race.

When you can see stronger riders being assigned two categories lower, something is going ridiculously wrong. If the organisers think this is working correctly, they have some completely crazy ideas about how zwift races work. If the categories are not mean to be grouping riders roughly according to ability, then what are they for?

3 Likes

Do you have that many people on teams? We try to put as few people on a team as required to have 6 per week.
Why don’t you just enter more teams if you’ve that many riders?

Also, as far as autocat goes - I’ve only tried it once. I think it’s only a very small difference overall from the traditional 5 A+,A,B,C,D categories. C3 definitely didn’t feel any different than B2/B3 category ZRL I’ve been doing.

Really - in my opinion what’s required is to lock people into entering only the right category when they enter any race through the Zwift client or companion app, mucking about with adding ‘speed’ to the w/kg categories is minor, unless autocat is ALSO the mechanism to add results-based classing, it’s probably a dead end.

2 Likes

I thought I could select any Penn I’d want once clicking the race pass, but TBH I haven’t ever tried to click any other Penn than I was supposed to - my bad memory then.

This is exactly why I’m asking for reflection - to hear their point of view on this and roughly if/how they would want to proceed.

I was going by Martin’s statement that every rider has been placed correctly. My interpretation of this is they were assigned the category they should have been assigned to based on their defined cat boundaries. This is different from having everyone placed correctly to provide everyone with fair racing. Solving that last bit is the key, obviously. I’d personally rather not see that the project gets lost in endless fiddling with Cat boundaries, but - irrespective of the underlying categorization - would prefer that over/under performers get sorted out based on their performance instead. According to Martin’s earlier comment this is the easy part to add on top so let’s do this, no?

Your conclusion might equally be true, in which case we may end up with nothing for the foreseeable future and the project put in the fridge again. I suppose I’m just hoping for a bit more.

Unless they changed the boundary settings in-between these two data points. We won’t know unless communicated.

I sort of agree and your comments provided another good example of under/over-categorized riders. However, the W/kg system is not exactly great in keeping riders of equal ability together either. It may be better, it may be worse, but does this really matter if race performance could effectively sort it out after one or two races provided it works well?

Primarily, I suppose, I’m not here to defend this testing process, but just really hope to hear Zwifts/WTRLs stance on it all. It will also be my last ask for this, I’ll just sit back and see things unfold (or not) from here.

1 Like

If they did that then… well I don’t know. That mean that someone signing up on Sunday would be categorized different from signing up on Monday. That will skew the data completely.

I don’t think you will get WTRL feedback on the forum. Martin made it clear they are not interested in the forum they get feedback somewhere else. . " I will be referencing those sources and leaving this particular one where it stands."

It wouldn’t surprise me if there was some kind of percentile magic in there that makes your category depend on who else is already signed up when you do, meaning that you could get a different category on different days…

Because I’ve ridden in the same event timeslot every week, I’ve had almost exactly the same cast of characters each race. Completely upended my Nemesis/Victim roster…

It’s Zwift. The first thing we should do is stop floating the pretense that WTRL is some separate entity. Not gonna address the reason the other sources seem preferable except to say I left the FB groups way back in the spring.

But in any case, this is a Zwift initiative. Subcontract between ZHQ and WTRL is their business, we do business with Zwift.

5 Likes

I’m not sure why there is still a discussion, WTRL have said they have all the answers to all the questions before they have been asked, they have thousands of people providing feedback somewhere in a place we will never see and that feedback is remarkably similar to the expected results so time to move on and accept what is thrown this way as it is all an improvement apparently.

7 Likes