I have tons of race data and currently in C. It appears that I will have to race in A?
These are my current numbers in ZP and I race at least once per week:
3 Likes
Well…as a ZP B I’ve been put in A, and my lightweight wife who is a mixed C has been put in B. So…either you think we are sandbagging below our abilities, or you’re doing something to change the categories quite significantly, I think.
My first thoughts:
It rates me as a C. I’ve been a C for some time. This seems right!
I’m looking forward to going “proper” racing!! 
4 Likes
That helps a bit, thanks. However without the formula, it’s pretty hard to test and feedback. VO2 max calculations without lab testing are an absolute minefield. I can see this calc causing issues, but the proof will be in the pudding.
For anyone interested, MAP is also generally lab tested, but at least you can get a bit closer with estimates by looking at power in the 4-8 minute range. I assume they are doing this. Paper here for those that like to deep dive. https://www.fredericgrappe.com/wp-content/uploads/2014/07/MAP.pdf
6 Likes
I’ll point out again, that using success at racing, via a ranking system, eliminates trying to guestimate how well people will perform in races based on guesstimates about their power curve based on guestimates about things like FTP and VO2Max.
A ranking system can be well defined, and documented so that there is little guessing to be done. And everyone will know exactly how their rank was calculated.
9 Likes
Your 15s, 1min, and 5min blows mine away and ZP had me in B before this change. So my guess is the VO2 part of the equation is what pushed you up.
Edit: Though “A” seems wrong, maybe there’s a bug there.
Is this suggesting that category availability will also be driven by the type of race? That is, the same person signing up the same day for 2 very different race profiles could be allowed to be in a lower category for one race type than the other?
2 Likes
Yeah, I could understand B, but A?
Sounds like the formula is using all rides and not just races. Do you have any Zwift rides with ‘better’ which weren’t races?
I rarely do Zwift rides that are listed on the calendar. It’s either a structured workout (primarily sweet spot training), a race, or a short C pace partner ride.
@xflintx Is there no compensation for female (or light weight riders)?
I feel you might get some negative feedback on this.
Just looking at me and my other half…
Me - B as expected.
Other half - Currently zwiftpower shows C mixed Cat, B female.
She can only enter B cat for both flat and hilly races.
She has 187w 3 race avg at 3.6 - True FTP is probably closer to 4.0.
Initial feedback is she would be hesitant to enter those events.
3 Likes
Unless these races are 5km long, I see no reason to sign up in an A Cat race if I’m a pretty solid C racer/lousy B racer.
2 Likes
If the proof of the pudding is in the eating, considering the phrase “eat your own dog food” let’s hope this doesn’t turn out to be a right old dog’s breakfast. 
3 Likes
Autocat used a secret formula to put riders in to categories, but it was horrible because no-one knew the formula and it clearly got a lot of people very wrong.
Other than the fact you can choose to race in a higher category, how is this different? For all we know, it is the exact same formula as autocat?
4 Likes
well autocrat had 5 or 6 pens IIRC so it’s not quite that…but it still seems…unwise…to invent something so radically different from the ZP categories, when we had been assured that this was not going to be the case. I mean, there will just be a bunch of people saying they’ve been put in the wrong categories, for starters…
4 Likes
Thanks Lee. Please fill out the feedback form linked above to let us know! This will be the most helpful thing for us.
The ftp value that’s being used, is that taken directly from the users zwift profile or from a value stored from previous rides? Essentially asking if a user was to deflate their ftp value in their profile how much effect would this have or is that where you then use these artificially determined vo2max numbers? Also does the algorithm take ftp decay into account?
This isn’t the same formula at all. What I said we’re using - VO2 and FTP - are the only factors we’re using to open or close categories to a given Zwifter for this particular test.
Autocat as it was tested with WTRL last summer looked at many other things simultaneously, including average speed, course difficulty rating based on climbs and their intensity, etc.
We also aren’t placing riders in a category when they sign up, we’re simply showing them what categories they can choose to join in a more limited fashion than we traditionally did it.
I recognize that you’d like exactly the numbers we’re using as well and while we’re not going to show you the exact formula, the information I provided above coupled with the link you posted shortly after is pretty spot on.
It’s my hunch - and it could be completely baseless - that if we show everyone how we’re calculating their categories that defeats the purpose of having test events designed to limit folks from intentionally cheating. It’s not hard to spoof numbers in Zwift right now as we all know, and if the exact info were made public, what’s to stop folks from cheating in a way that more closely matches them with a category they want to disrupt on a fresh account?
What we’re going for here in the long run are three things:
- A system that is easy to make changes to relative to how we’re choosing categories (ranges/limits in terms of VO2, etc.) for future events under future definitions of categories, whatever those may be;
- A system which we can easily apply different category rules per event to if we want in the future;
- A system that we can easily provide transparency with about the different rules we apply, though we decided not to do that in this first iteration to collect feedback (which I see was an error).
5 Likes
Those who will spoof their numbers or game the system probably won’t enter a test event testing auto categories…
It’s chicken and egg scenario…
You need to apply this testing to something like the Giant Crit crusher series etc… I’m aware this may come down the line.
1 Like
The point is, it could be the same formula. A secret formula is a secret formula. We can’t know if it is the same or different.
Do you really think there is a significant cheating risk from people knowing how certain duration/power values are used to place them in a category? If so, would this not be a good thing to test?
I’m mainly frustrated because this is the opposite of what we were told, in no uncertain terms, we were getting.
I am all for using more of the power curve, but the calculation needs to be shared.
This is what you will get. Without knowing what the formula is, who knows if they are right or wrong.
2 Likes