Zwift Racing Score - Seed V2 Release [December 2024]

Racing
1

Tomorrow’s the day for the new seed formula to roll out! We’re excited to share a bit more information about what to expect and how we developed the new formula.

At 6a PT tomorrow (Dec 4), we will be starting to reprocess historical seed values, floor values, and race results for all racers and events since the last reset on Sep 25, 2024. The release and reprocess is expected to take a couple of hours. Once it’s complete, we will verify the data and notify this thread that processing is done. At that time, you will be able to see your updated score. You will also be able to look back at historical race results in Zwift.com and Zwift Companion app to see how your score changed in each race.

To reiterate a few points from our last update in case you missed them:

  • Most Zwifters should expect their score to change. Once the reprocessing is complete, we expect scores to change less than 100 points for a majority of Zwifters. Since one of our main goals of this change was addressing inaccurately seeded racers, some may see scores change even more due to being inaccurately seeded with the old formula.
  • A score increase or decrease doesn’t mean racing will be harder or easier. Racing subgroup difficulty will be relative to where others landed, but the main goal here is to ensure the groups feel fun and competitive.
  • We will be revising our standard category ranges in Zwift events starting January. The overall score distribution is not changing drastically though, so we are comfortable releasing the V2 seed formula ahead of that adjustment.

What does the new formula look like?
The new formula is based on 90 day best 10 minute power and rider weight. The main difference from the old formula is that we removed the 30 second power interval and we leveraged ZRS-based races instead of power-based races to map these values to OpenSkill scores in our 1-1000 range.

How did we develop the new formula?
We considered many options with the new formula - we actually had 10 different candidates that we analyzed (including ZwiftRacing.app’s customized compound score). These candidates varied on which power intervals they contained, how much weight each factor had and whether they were trained with mass start events or ZRS ranked events.

Our main criteria for ultimately deciding which formula to implement was based on how strong it could predict a score after x races*, what the resulting distribution of racers would be, how much scores would change, and how many people would be at their floor.

What else has been going on with Zwift Racing?
In addition to the new seed formula development, we’re excited to share the following ZRS updates

  • The change to require only one 10 min activity (vs. originally requiring 3) within 90 days to start racing was released on Monday, December 2.
  • The mechanism to require a minimum effort in order for a score to drop (anti-score tanking/sandbagging) was released on Monday of this week and will take effect for any races Dec 2 and beyond.
  • We’re still on track for updating score subgroup ranges on Zwift owned races starting in January.
  • We’re still on track for enforcing subgroup rules at the time of event join starting in January.

* For the stats nerds like us, we used correlation coefficients, R^2, RMSE, and residuals to measure model fit

23 Likes
2

Nice! (Also, first!)

12 Likes
3

Reserving judgement until we see how racers’ scores are adjusted and how it allocates them relative to their old Category Enforcement pen.

But disappointing that 10mins is still in the seed formula. That would require a proper full effort up at least 3Km of Ven Top for below average pen E racers and closer to 6Km for above average pen As.
5mins would cover most racers doing an all out effort up Petit KOM.

4 Likes
4

It would be nice to have a 10 min metric on Zwift Power.

Good timing with the Keith Hill climb in ZRL next week, which will force a lot of people to put in best 10 min efforts.

3 Likes
6

Nice, I’m interested in how well the anti-tanking measures work. Will be interesting to see. Not super thrilled on 10min power as the main metric vs. compound score, but I guess it’s better than 20min power (which is easier to cheese), and probably better than 30s power for determining who will be able to make it to the end of the race.

My guess is no matter what, folks who are at the bottom of whatever strict cat range is standardized on will be complaining, people who are in the top half will be more happy :slight_smile: I still think Zwift needs to adopt very flexible cat boundaries for the major events by default so it’s much more likely you will have a varied race experience within a season and to make sandbagging more difficult. Hard to target a moving number.

3 Likes
7

it is there in ZP you just need the check every profile/power

1 Like
8

For those dummies like myself how does the Race Score work from race to race, seed to floor etc?

So I get a seed score. Floor is determined by % of seed score. I begin to race. My race results alone impact my Racing Score or does my LOE also impact my Score? Does the quality of opponents impact my score as well? How does the floor change over time? How often does the seed change over time?
I enjoy racing and all but with each race seems like almost everyone’s score increased, so something seems amiss with my understanding.

9

I think many aren’t bothered by where they fit in the category so much as the scores reflecting competitiveness. I know several people who aren’t bothering racing at the moment because they are not competitive against people of similar numbers or worse, aren’t competitive when they’re at the top of a category and then get upgraded to the next one and become even less competitive - doesn’t feel like rewarding progress or providing good racing which is what I think people want.

Keeping fingers crossed that v2 will help fix that.

2 Likes
10

Very roughly, finishers in the top half increase in score and those in the bottom half decrease, unless they are limited by 85% of their seed score eg. right now I cannot drop below 190. Or they improve their seed score during the race.

Race 1 Tiny Races zone two from last Saturday (pc web browser will show more than 50 racers per pen)…

1 Like
11

Missing in the update is the option for organizers to use in different scenarious for Leagues, Tours, etc., please think of this to be added ASAP. :pray:

  • Max 30 day ZRS
  • Max 90 day ZRS
  • Seed Score
6 Likes
12

Are the category limits for community-organized racing going to change as well in January?

There are several clubs already using several different category schemes all different from the default, and don’t really want to go back to the default.

I’d prefer we adjust ourselves after we see what happens with the re-seed.

2 Likes
13

My guess is those same people will not be happy if the new seed puts them bottom of a cat because the folks bottom of a cat are not really competitive with the folks at the top of a cat. I guess we’ll see.

1 Like
14

I will run some test with different rider weights tommorow after Seed V2 is out and i can report some of the findings so that organizers can get some idea on what the custom pen limits could be for there races.

3 Likes
15

Disagree. If you’re bottom of a cat and improve then you should get competitive. Pretty much everyone I know is fine with that. The issue is if you improve and still remain uncompetitive which has clearly been happening in some cases with v1.

2 Likes
16

The issue is the cats are so broad that most folks will not have that level of improvement through a season. The people you know could also have “improved” to be competitive in the earlier cats but they just gave up.

17

They aren’t changing in-game physics, only the ZRS seed formula. There’s no reason to do any testing with different weights.

18

it is not a speed test…it is a test to see how 50, 60, 70, 80, 90, 100kg riders are seeded at same Wkg and what is the Wkg difference at same SEED.

6 Likes
19

Not really. People started mid cat or near the top and weren’t ever competitive before getting upgraded as their score increased.

Essentially we’ll know it works when riders at the top of categories are consistently competitive.

2 Likes
20

Their score would only have increased pretty much only if they were top half of the field which means they were competitive in Zwift’s definition. That will be the same issue with the new seeding. So I don’t see the difference here.