Autocategorization Test Events & FAQ

Hi Racers,

You’ve been asking for an update to enhance your competitive experience and we’ve assured you that we’re listening. Today, we aim to give you a peek into a feature in development by WTRL that could help auto assign your race category.

This post will be a bit longer and will feature an FAQ at the end; please read through it before asking any questions.

WTRL has been poring over data from Zwift Racing League and ZwiftPower results alike, and believe that within that data lies a possible solution to the oft-asked questions of “what category do I belong in?”, “how do I really know what w/kg means for me compared to other racers?”, and “is the category I’m in giving me a fair and fun racing experience?”

This means that there may be - at least for the near future - an answer to these questions as well as a way to deliver you a better racing experience. This feature is still in development and needs extensive testing but the development is a step in the right direction.

With that said, we are offering all of you an opportunity to participate in our initial pilot test of auto-categorization on 7 July at three different times for three different regions:

  • "DOWN UNDER" - 19:30 AET, WEDNESDAY 7TH JULY
  • "MIDDLE EARTH" - 19:30 GMT, WEDNESDAY 7TH JULY
  • "WILD WEST" - 19:30 EDT / 16:30 PDT, WEDNESDAY 7TH JULY

The test will offer 6 categories for men and 5 categories for women, with a cap of 100 riders per category on a first come, first served basis. This will serve to better monitor and gauge your feedback and will determine if the system is working as expected.

Instead of using the normal power-based categorization that you are used to seeing in Zwift or ZwiftPower, this system works a little differently: your historical finishing times for courses based on a difficulty rating of 1-5 within the most recent 30 days. This will be used to pair you with racers that are most similar to you in performance on that same difficulty of course. You are then placed in the correct category via a private event link to ensure you’re racing in a fair and fun event… Each event link is limited to one category only.

Here is a bullet-pointed explanation of how we expect the upcoming test to work:

  • The course for the test will be Watopia Hilly Reverse for 3 laps.
  • The system will analyze your historical performance on Volcano Flat within the last 30 days to determine who will be similar to you in terms of performance for you to race against on this course.
    • If you do not have race data within the last 30 days, please see the FAQ below
  • When you click the “Autocat Me” button, the data is analyzed within seconds.
  • Each category - Class 1-6 for men and Classes 1-5 for women - will be placed in a single event with a single subgroup, ensuring that only the 100 riders meant to be in that category will be on that course. This means there will be 6 separate event links events for men and 5 separate event links for women, with each link representing a single Class.
  • Once the races are run, the results will then appear on ZwiftPower.

If you wish to participate in this test, please follow these instructions:

  • Ensure you have a connected ZwiftPower account, that your Zwift profile is connected to ZwiftPower via your account connections page
  • Ensure you have a WTRL account.
  • To find the “Autocat Me” button, head to this link at the WTRL site. You will also see an overview of the system. Click on it!
    • If you click the “Autocat Me” button and you aren’t logged in, you will be taken to the Login screen where you can also select to create a profile (requires name, Zwift ID, location etc)
  • Once the above steps are completed, the software kicks in and after 15 seconds or so, the system will have your category and take you directly to the Zwift page (or Zwift Companion App) for your event.
    • If you are using your computer’s web browser, you will then be sent a private event invite link.
    • If you are using your phone, you will be taken to the sign up page on Zwift Companion.
    • You will then get an indication of the Class assigned from the number in the event title. Class 6 would be novice and through to Class 1 would be expert.
  • The results will be on ZwiftPower

We want to stress that this is a test and only a test. The goal of the test is to make sure this system of auto categorization is clear, works well, and is simple before it could be considered to be used more widely in future events.

In the spirit of fair competition and transparency in racing, we need to address one critical factor about the test event links: they can be copied, pasted, and sent to others, such as friends or teammates who may not be in the same race as you.

We kindly ask you to please consider what kind of repercussions this might have on the race itself, and to consider if sending a link to someone that didn’t receive the same one as you would be in the spirit of fair competition or not.

It would be counterproductive to attempt to circumvent this system by sharing links and would effectively render this new system useless. Thank you.

We also ask that participation in the discussion on this post is focused and civil. Overwhelmingly negative comments or counterproductive discussion will not be tolerated.

Check out the FAQ below. We invite racers of all abilities to the test events and see how this system works in your category. Once you’ve tried it, please come back to this thread to comment and ask questions.

Ride On. :ride_on:

_____________________________________________________________________

_____________________________________________________________________

FAQ

Q: Where do I sign up for the test events? Are there limits on the number of racers per event?

A: As mentioned above, there will be 100 riders per event on a first come, first served basis. You can sign up via the WTRL link. You will also need a ZwiftPower account.

Q: Why speed and time instead of using power to determine my category?

A: WTRL has a great explanation on their Autocat webpage.

Q: Why six categories for men and 5 for women? How do these categories relate to the current A-D categorization on Zwift?

A: 6 categories for men and 5 for women aims to provide much closer, fine-tuned classes with less of a ‘performance’ gap between classes, regardless of if you’re racing on the high end or the low end. It also aims to ensure races still have healthy attendance. No one wants to race by themselves or with just one another person! This is not final, however, and may be changed if this system is used more widely in the future.

Q: How is my category determined?

A: WTRL has a great explanation on their Autocat webpage.

Q: What if I don’t have the data in ZwiftPower because I am new to racing?

A: For the upcoming test events you will need to have some race history. This is to compare categories and to gain user feedback on how the races feel. Zwifters with no race experience will not have an enjoyable experience during the test events.

If WTRL’s AUTOCAT is used for more races, it would be beneficial for Zwifters to have at minimum 1 race under their belt, with the ideal being 2-3 for best category placement. Ideally, these would be races of a maximal effort for you. Provision might be able to be made for a Zwifter with no race history by using their power-based data. If using this data, the system will calculate roughly where you should be. Please note that the system will be focused on placing riders in this way in higher categories, otherwise known as “catting up.”

Q: What happens to my category if I have a great race? What if I have a bad race?

A: During the test events and any future events this system may be used for, if you are performing outside your assigned class, you will be automatically moved up or down for the next race depending on your performance.

Q: What if someone joins the wrong race?

A: For the test event, we ask that you only join the race you are assigned and do not share the race link with others. If this system is used in future events, we will have a 2 strike system, where riders will be disallowed to sign up for events after joining the wrong race twice.

Q: How did the anti-sandbagging feature influence this system? Is it being used in this system?

A: Anti-sandbagging is still undergoing work, and is also separate from this system. It is not being used in these races and did not influence the system in any way.

Q: Do I need to be registered on ZwiftPower?

A: Yes! Please see our support article on how to set up a ZwiftPower account if you do not have one.

Q: How do I get put into a category to race? What’s the registration process?

A: We’ve outlined the registration process above, and WTRL also has a great explanation on their Autocat webpage

Q: What if my category feels too easy or too hard?

A: This is what the test events are designed for, and not every racer may have the best feeling category during the test. If this system is used moving forward, the more race data you have will ensure a more accurate placement, and if you are placed in a category that is too hard or too easy, this will also be taken into account prior to the next race.

Q: Where do I see my results?

A: Test events will be on ZwiftPower only, so make sure you have a ZwiftPower account. If this system is used for future events, it is possible that results will be on ZwiftPower as well as on WTRL’s site, though that isn’t yet confirmed.

Q: What does this system mean for the future of Zwift racing?

A: If this system is used for future events, the hope is that it delivers a consistently fun and fair race experience based on your performance. WTRL also has a great explanation on their Autocat webpage

Q: Do I need any special type of account to get access to these races?

A: You will need three accounts total: Zwift, ZwiftPower and WTRL. This ensures all data is shared and accurate.

Q: What type of Zwifter is this system designed for? Do I need to be a racer to use this?

A: The test events are designed for more experienced Zwift racers, though everyone could participate. If, in the future, we use this system, it is designed to give every Zwifter a better overall race experience.

Q: Can I use this new system for my events?

A: It would be great to make this possible in the future, but not right now as it is still being tested.

Q: I was sick/ill/injured the past few weeks, how will that impact my auto-categorization?

A: Provision will be added for this if the system is used. It does not exist for the testing phase, though there are several potential ways this could be addressed later.

Q: How do you select the course data you will use if I haven’t been on the course I am signing up to race?

A: The system does not use course specific data - all courses are ranked 1-5 currently in terms of difficulty. Performance is compared based on course rank.

Q: What if someone hasn’t raced a course in the past, but has completed a social ride? Is that data used to determine their category? How is that fair?

A: Rides like fondos, group rides, or social rides are NEVER used to determine race categories. Rides are usually speed/power controlled in order for all riders to remain part of the group. While this will likely push a few people, it is not generally a competitive experience for the majority, and as such does not give a fair indication of efforts that might be seen in racing.

Q: I’m trying to sign up, but I am getting this error: UNABLE TO LOCATE A RACE FOR YOU IN LAST 30 DAYS. CONTACT WTRL (racecontrol@wtrl.racing). What do I need to do?

A: Since the test events are designed for Zwifters with some sort of recent competitive history, the system is looking for race data within 30 days. If you wish to participate in these tests, please do at least one race as soon as possible and then try the “Autocat Me” button again. In the future, if this system is used for future events, there will be a way to use other data to place you in your category.

Not much to say, other than that I have been the first to complain that Zwift have been slow to develop anything for racers, and also complained that we need better comms on development progress. The above post is a great bit of both progress and comms - thanks!

I do think 30 days is probably too short a sample, especially during Summer, however it looks like this is (or will be) addressed by the last paragraph.

I doubt I can make it, especially if England beat Ukraine on Saturday, but I hope you get some good rider numbers and data!

12 Likes

Lovely stuff. This is really encouraging.

5 Likes

Not a fan of requiring yet another Zwift-related account to take part, but promising steps forward nevertheless. Be interesting to see the outcomes.

3 Likes

Is the “Autocat Me” button deactivated for others or just me?

2 Likes

Also for me.

I think it is only activated a few days before the start. Not sure though.

This is awesome :clap:t3:

1 Like

Great to see this!

1 Like

What about racing in a more demanding category than the lowest one you are allowed to race in? Based on the description of the process and the FAQ, there are no accommodations for that at the moment.

I have no interest in doing a maximal 20 min performance in a race except by accident (what do you think I am, a triathlete or something?) and prefer to do races where the 1/kg component is less important. As a consequence, my assigned ZP category has been C except for a short while in B (and the current D due to lack of races), yet I find racing in A or B more interesting. (Heck, I placed 4th in the one A+ race I have done to date…)

It’s good to keep in mind that the aims of the category system are almost the opposite in IRL vs e-racing, in that IRL it is at least as much a matter of protecting the higher categories in terms of safety considerations and space constraints as it is about removing sandbaggers from the lower categories, whereas in e-racing a strong but inexperienced racer can’t harm others, space is not a real issue either, and sandbagging is much bigger a problem.

1 Like

It is a maximal performance (i.e. do the best you can) not maximal power. So you can still be tactical, draft and sprint. If this system matures and develops, would there still be any reason to want to cat up? Other than maybe joining a friend or a team for a race?

On the WTRL AutoCat page, they mention that they will test out:

…an auto-categorization system with the following factors in mind:

  1. 6 Rider Classes based on average racing speeds.
  2. 20min, 5min and 1min power ratings.
  3. Race Results and course difficulty based ELO system that influences category upgrade/downgrade as well as providing a ranking system.

By adding #2 to the calculation, your sprint will maybe be picked up by the 1 min if you start the sprint early :slight_smile:

A few questions…

How do you move between Cats? It doesn’t seem clear.

If it based on previous time, then you are governed by the number of racers and the blob effect & not power. Larger races will tend to have quicker times especially if you get dropped by the front group - There will be multiple blobs in larger races as opposed to TT effort in smaller fields. Is this accounted for or weighted against?

If based on time & everyone is of a similar nature, surely the people with the biggest 1min & 15sec power are going to win - I.e. Sandbaggers will continue to win?.

As I mentioned in the other thread, a split of 1Wkg across the current cats gives a good range of opponents - Racing faster people means you get faster (you might fail a few and get dropped whilst doing it) but over time you get quicker - If you are all expected to finish in the same time range and you cant form breakaways where does the improvement come from?
Looking at the test race, there was .5wkg split over the cat of racers (with a few outliers) - That seems like it is too close to racing the same types of people at the outset…

1 Like

0.5 W/kg spread (with a few outliners) within the category seems to be pretty normal in races with the current category system.
Looking at cat 3 in the test race, there is a minute between first and last. The field is split in two.
Cat 4 has an even greater time difference from first to last, with about 5 minutes. The winner of cat 4 has about a minute up to the last rider in cat 3.
Such differences will not happen if all riders in the same category are of the same type.

I think that is inaccurate
Not in A, B, C - You get the full range from low end to the very limit before being upgraded. B’s wil have 3.3 through to 4.1, C’s will have the full range…

Even with people of similar nature you will get splits - Its an odd statement to say you wont… Especially on volcano flat with the downhill in the dust… Doesn’t take much for a gap to open and it wont get closed due to the game dynamics.

I commented on your statement that the racing was “too close to racing the same types of people”. The time differences and splitting of the field shows that that is not the case.

With the test races you do not upgrade as such, the categories are dynamic (calculated each race). I commented on your statement that .5 W/kg is too close. I looked at some of my previous races and found that 0.5 W/kg (with some outliners) is pretty common. W/kg is irrelevant with categories based on speed. It does not matter how many W/kg you produce, just how fast you go

Again, id suggest your statement is wholly inaccurate\conjecture and you cant show it isnt… Similar riders will get split due to game dynamics and terrain… If it was perfectly flat, there would be some weight in your comments… the course has some key points where the group could be split, such as climb out of the tunnel, volcano climb, the dirt road and the esses.

I just had a quick look at a few B cat events for today… And whilst numbers are very thin, they do seem to have an even split from 4.2 through to 3.3. There will be a bell curve of users who sit in the middle, but the numbers at either end were very limited (outliers) in the test race - Id rather see a even split across the whole (original) cat range…

Edit to add - Lets not derail this thread… I have asked some specific questions, id rather Zwift provided some actual answers than your conjecture now we have gone back and fore a little…

2 Likes

I can’t show that there was a significant spread in time? Maybe I’m misunderstanding you.

I agree. Let Zwift answer you questions :slight_smile:

It’s worth clarifying that the times shown for the race are GMT, not BST, so it’s at 20:30 BST (I think)…