That’s correct. ZRS is not ELO-based
4 Likes
DNFs won’t show in results but perhaps they’re counted in determining the halfway point.
Yes, E-cat doesn’t follow the pattern. Too many coming off of ZRS seed scores of eg. 2 or 11. (not sure why, but a score of 11 seems to be really popular 
). Actually the lowest of the Es in a Tiny Race are taking more than 10 minutes to finish, and by that virtue alone are supplying more ZRS data than everybody else finishing before 10 minutes have elapsed. So their score can go up, because they indeed are registering an improved 10m power.
You are much closer to Zwift so if they have informed you that is the case thank you.
So signifies a pb both before the race, since the last race, and also during the race?
The up arrow with a line under it would signify setting a new 30s or 10min power PB during the race. That’s all it would signify. A power PB set elsewhere (say, when free riding) might change your Racing Score, but you won’t see any arrows anywhere telling you that. You’ll just see a changed score!
I’ll keep an eye on this, but for now and for whatever reason (possibly it’s just very coincidental), the 50% +/- rule as a function of signups seems to describe score changes pretty closely amongst the established racing cats (A thru D).
The primary point though, is that there isn’t any apparent relative score adjustments based on what the scores of those finishing around you have.
From the top vs bottom of the 430-550 cat.
For these top 8, 8th place gets an increase, but everybody he lost to has a lower score. Winner has lowest score of any of these 8.
In the bottom of the race, the 343 RS racer loses points, despite beating riders all over 400. I know time isn’t considered, only came 4 seconds behind a 500 point RS rider.
1 Like
I’m not sure I see a good mix of up and down arrows in the 0-300 group. The down arrows start at 20th place, from there two of the 4 up-arrows in the lower half were due to PBs (as indicated by the ceiling icon) so we can ignore those, and the absolute last place person is one of the only other two up-arrows I see below 20th spot, which doesn’t make a lot of sense given they were… last. - why did they go up?
The last and only other up arrow that does not fit the ‘once things go down they go down for everyone without a PB’ was for someone that ended with a RS of 13, so maybe they do something different to folks who are at such a low RS?
So yeah there was one up arrow below the start of the down arrows that might have made sense, the other up arrow that was not a PB floor increase was for the person in last place which doesn’t make sense unless maybe they are counting people that do not show up in the list?
But what I believe is happening is that Zwift are showing you the movement since your last end of race score [not since the start of race].
End the last race with a score of 500.
Then do a good pb and gain 20 points.
New ZRS score on your profile = 520
Start next race and come bottom third and lose say 5 points.
New ZRS = 515
Your race arrow should be down if it covers race only
But I suspect the arrow you will see is up to signify you moved from 500 to 515.
i believe this is wrong…up arrow with underline will only show if your SEED is raised more then 15% from your current score.
If you get a small Seed increase you will not get any info…you can only know when you are at your floor. ZHQ should show Seed and Zwift Racing Score…this way we can follow how people’s scores are moving.
Example of a problem we have at the moment…returning rider has 3 activities with Robopacer and gets a score of 100. He enters a race and gets new PBs. His new seed is 500 but his ZRS will not move to 500 but it will move to 425…this is his new floor. The arrow will be up and underline. Why do we want this Discount on his score…it just doesn’t make any sense.
Ok, then this would explain the person at the bottom of the list in that 0-300 group that I discussed above.
They were last in the results list but showed a bump up-arrow without any “increased seed” ceiling icon. If they got a minor bump in their seed to 13 as a result of their effort and if Zwift only shows that icon for larger bumps then that would explain it.
Eric or zwiftHQ,
Are you able to provide a rough count for how many people or rides were used in the racing score seeding, and what percentage are likely skewed by returning riders for the winter season?
The issues seem to be that of data analysis, not of lacking data. With all the data collected from every second of every ride that ever happened, I’d bet somebody could write an algorithm that would predict my or anyone else’s placing in any type race. And I think that is a good indicator of how seeding should work.
It should be something like calculating results for returning riders based on past data. And for new riders, matching them via the many power metrics to their “twin” existing user.
And then lock weight changes to some percentage over time so people can’t game the metrics by losing or gaining 5-10 kgs over a few days between different races. It’s the weirdest phenomenon racing all of these 150# 40-50yo Americans.
That said, having done a few mid-C tempo rides after being off all summer, I see I am seeded in the bottom pen of tiny races……
i would guess the reasoning is because big spikes in score = angry/confused forum complaints… the idea seems to be to have power PRs and seed scores and stuff not getting in the way too much, in the long run
probably
though as stated, zrs isn’t an Elo or system or vElo, one of the aspects of vElo i always thought would be hard to “sell” people on is the minor/major zrcs adjustments
Thanks for sharing, Tim. This explains a lot about why top half goes up and bottom half goes down in most races.
what in the world is ZHQ doing with this system? @Nick_ZHQ when will you share the algorithm as promised so we can figure out what you’re trying to do?
making us nervous 
zr.app isn’t Elo-based either, is it?
Yes, it is. 
So, if I’m understanding ZRS at the moment (and I hope I’m not). Then if you are someone who is just barely above average in your seed category (i.e. you’re always getting dropped, but finishing just above mid-pack each time), then you will eventually get promoted to the next cat at some point, and that will not be based on who you beat (assuming you get about the same position each time), but more about how many races you do and what your starting seed was. You might never get even close to any podium, or make any final sprints at all but as long as you’re over halfway above mid-pack you’re on your way to the next cat.
Once you get into that higher category you will be below half of the pack, lose even worse, and then move back to top of the previous category as your ZRS will then drop a bit, and at that point you will basically bounce between the top of your seeded category (where you were mid-pack) and the category above (where you’re getting wrecked).
2 Likes
Thanks, and sorry, I misunderstood what James said earlier then.
1 Like
The ZHQ description (beat people ranked higher than you to move up, lose to people lower than you to move down) is distinctly ELO. Why are they describing it that way when it isn’t? What IS it?
Edit: It seems to be top half moves up, bottom half moves down, with an occasional exception. Since ZHQ is refusing to explain the system (as promised) can you fill us in?
2 Likes
I believe they have recently updated their Racing Score FAQ. Possibly earlier wording was misleading?
“Your Racing Score will go up or down based on your finishing position, your competitors’ ability, the participant field size, and there’s a bonus for a podium finish.”
And
“Pro Tip: A higher finish against stronger competitors in a larger field will increase your score more significantly.”