We hope everyone had a great holiday season and a happy new year! Here are a few racing updates to kick off 2025.
Pen enforcement will now happen when joining the race. Previously the score range was only enforced when signing up, but starting today, you must still be eligible for the race category when joining the event. If you are not eligible, a message will appear indicating you need to sign up for a different category.
Tiny Races now more accurately calculate scores. Because Tiny Races are short races in quick succession, there was an issue where scores weren’t updating fast enough race over race causing inaccurate results until the score finalizer ran 12 hours later. Now you will be able to see a more accurate score ahead of the final processing.
An issue was fixed with scores calculated from races held between Dec 30 - Jan 6. We identified an issue where scores weren’t finalizing as expected during this period and have fixed the issue (and our error alerting). During this window, you may have seen scores that didn’t correctly consider past race results and therefore were off by a few points. All results from this window are now fixed.
We’re looking forward to making racing even more fun and exciting this year and build upon the foundation for Zwift Racing Score that was established in 2024!
There are a few oddities around ZRS, one is that you can get much bigger score increases from small events, a second is that it appears the ZRS is not actually mu (in the OpenSkill model), but some form of mu minus sigma which could account for a lot of the confusion around why scores go up/down (is it because of actual mu going up, or is it because of sigma getting smaller etc.). Also, if you race up a category and get a bad result that should probably not tank your score as getting a bad result is expected in this case.
Are these questions ones that Zwift is looking into?
I managed to try a couple of the new Flat is Fast zRacing series last week and it appeared that the new ranges were working very well. Obviously on a flat race it’s harder to break away so it will flatter any changes.
But this was a very encouraging start. Looking at the entrants in the up to 510 pen on Zwiftracing.app there were very few Emeralds there, which is perfect. Narrowing the ranges and alternating them is sure to be a great success.
It will be interesting to see how it pans out on hillier routes as and when we get to those.
That appeared to be in operation from 23rd December, I signed up for some 0-180 Zmonthly pens during the brief window of v2 seed score being active (which put me under 180) and I couldn’t join the race pen when my Racing Score increased after v3 seed score.
On Windows PC, when I clicked the event on the home screen, the new window appeared on the right coloumn, but at its bottom in blue it said “change category” instead of being orange and allowing me to join the pen.
Don’t recall anyone else experiencing this and confirming my observations.
Last time I checked, I’m now at my 85% seed score Racing Score of 178, so this issue is now mute for me.
Good news regarding quicker score updates for Tiny Races, hopefully this means that a few might be excluded from joining the next race in a quartet, due to exceeding the sign up pen threshold in the previous race in a zone quartet.
Under the linked ZI article from @Eric_Schlange_ZwftIn the following shows as a completed part of the project:
Podium Bonuses: The top 3 finishers in each scored race are now receiving a podium bonus meant to help move racers up a category faster if they are regularly winning.
If this is complete, why when I recently won a race did I only see an increase of 7 points for winning a race with 130 other racers?
Either, I received a bonus and the system thinks that a score of 6 or less is appropriate for that result, or the podium bonus is not working?
The bigger picture is that scoring is currently backwards to how people expect it to work (well attended races show little change, empty races show large changes), but there is no recognition of this being either a fix on the agenda or direction to say this is working as expected - probably in the form of a ZI article as god forbid zwift staff engage with their user base beyond females who are impacted by a change or podcast hosts…
This one.
As you said, the system gives big changes for small fields, small changes for big fields.
Podium bonus is some percentage of the change, so it’s still small in big fields.
It’s backwards, and it’s the opposite of what the FAQ says. But the evidence doesn’t lie.
Winning a race against 130 would award a bonus to the top 13, which is added to their seed score for a limited time such as 28 days, to reduce the chance of successful racers trying to tank their score in the short term in order to not get a pen promotion…
If I had my way.
Winning a race against 130 others should score far higher than winning against four others.
Races with less than five racers should not be changing racer scores after the result, it certainly should not be changing scores when there are just two in the pen, as happens currently!
How are these top 5 riders in the 330-450 category? These guys are upper end cat B’s. Of course their Zwift score is below 450, but they clearly don’t belong in this category. These 5 riders left the pack within the first 5km of the stage 2 Flat is Fast race. We had 30 riders in a pack for a large portion of the race and we couldn’t catch them. The top finisher produced 4.2W/kg in a Cat C category. Clearly the Zwift race score is severely flawed!!!
The one I looked at didn’t start back racing on zwift this season until mid October, I suspect the others are similar.
People who were active before this came into the ZRS system at their seed score based on their best 5min power at the time, which was likely a representative effort.
Anyone who came in after ZRS went live would start at their true seed score if they do a maximum 5 minute effort in a training ride before their first race, otherwise they will come in at their floor score which is 15% lower. They’ve essentially tanked their score as low as it can go without even trying to do so.
The floor being 15% below seed was necessary in the first iteration of ZRS, where the sprint power or lack thereof, was skewing things massively. With the current seed score algorithm such a big drop to floor is no longer needed. Hopefully that’s on the list of things to fix in the near future.
Right now, it appears they base results on the number in the pen.
Some may not start, some are DNF and some sandbag.
None of these riders will affect their own score but their presence affects others score.
From what I’ve read about the Microsoft methodology for grouping players using their True Skill method, the score used is “mu minus 3 sigma” so that the score used is 99% certain to not overstate the player’s ability. Over time, sigma, the uncertainty element, will reduce to a residual small value (and so the score is essentially mu) as the system has more results to work on.
I don’t think the uncertainty element really works for racing, as there are so many reasons why a rider may not be performing to their true ability, so the sigma element of the score could remain large, reflecting variable performances.
Were you in cat for both or did you race up for the 4 of 4?
I think it is ok to race up but I don’t think that racing out of cat should affect a racers score and out of cat racers should not affect other’s score.
Most of the issues concerning non starting/finishing participants can be addressed by counting qualified finishers but that will miss out of cat racers.
