I am sick of Zwift race scoring system. Seems that every race I have entered since November I lose racing points. And then here is the results from today.
Set a PR up the Alpe yesterday, fastest time ever. Then beat at least 10 other racers in this category where I was supposed to be at the bottom. You can see from the results I beat at least 5 other guys with scores higher than me and Zwift decides to take one racing point away.
Like it cost me a race point to race today….
depends on the size of the field and the points sometimes never make any sense because their interpretation of the formula used is a bit ■■■■
Yeah, ZRS is not working the way it should. I looked on zwiftracingapp, and it had predicted you would get 30th, and it showed you getting better (at 26th), and gave you a small boost to your ZRA vELO score as a result.
ZRS has never worked decently as presented to Zwift racers, it’s been broken since release in summer '24.
Why did the person in next to last place lose 12 points? Last place 0 points?
Have to presume that Ryan in 34th place has done ~400W+ for 5mins recently, but not for an event shown on zwiftpower, so was deemed to not try hard enough. 313W at 99.8Kg would be a seed score 282 and he’s at 459. Or he has lowered his weight to set a better 5min W/Kg.
My current theory is your race 5mins has to be more than ~75% of your 90 day best, if you are to get a score decrease for finishing in bottom half of total signups for a race.
Ian in 32nd place did ~76% of his 90-day best 5mins on zwiftpower, so he got his -12.
So, you get punished if you have done a five minute max test in the last 90 days?
I wouldn’t call it punishment, you are expected to try in a Zwift race compared to your recent ability.
Pretty sure the minimum effort is higher on Tim’s Zwiftracingapp for a vELO decrease.
75% of your maximum effort is far too lenient and is helping some racers to manage their ZRS far too easily.
In fact, Ian our -12 guy is flagged for not reaching minimum effort on Tim’s site!
Full agree with you that the minimum effort level is far too low. In addition IMO power output for 20 min and also race average should be used.
However on this occasion the 0 points movement doesn’t relate to minimum effort, it reflects the fact that rider is already at Floor and can’t go any lower.
At least my vELO score went up. That system seems to be reacting properly.
VELO is great
ZRS is horrible. A random number generator wouldn’t be much worse than this. Pick a random number. Then a standard deviation and few would notice a difference.
ZRS is using the same power metric as vELO for initial seed score. How they adjust the score is far worse (decay is too fast and too much, scores don’t move enough per race, don’t take your expected finish into account enough, etc.)
Velo score 1 and 2 are far better, but it’s not like this really matters, right? You got 22nd and lost 1 point, virtually nothing has changed. You lost the race that you entered, so even if the scoring system sucks, its still vaguely indicative of your skill.
I disagree with this as a general statement. Sometimes ZRS scores move far too much per race, with small field races getting much bigger movements even though the system hasn’t gained much new information.
I think this really relates to how ZRS used to start some racers at their floor, and a new 5 minute PB would only increase their floor score rather than move them to seed. Moving up to seed with a new PB was a good change, unfortunately somewhat cancelled out by the reintroduction of decay.
But the main problem with ZRS is that it has no terrain adjustment. I think this is the root cause of the score movement problems, because very different riders will win climbing races or flat sprint races, so a system based on probability must have a high level of expected uncertainty built in if that isn’t accounted for. That’s why we see score movements seemingly ignore relative scores when the field has a ~100 point spread, because it treats those as 50/50 probabilities. When the score difference get big enough the relative scores do have an impact.
That is one problem with ZRS.
But it isn’t the main problem.
The main problem with ZRS is the fundamental simplicity of the whole thing as implemented…
Finish in top half of pen compared to number of signups, ZRS increases.
Finish in bottom half of pen compared to number of signups, while doing a very low bar of ~75%+ of your 90-best 5mins, ZRS decreases.
A small number whole finish around 50% of totoal number of signups get no score change.
Absolutely no consideration of abilities in pen and where racers should finish relative to each other if they put in a decent effort.
No consideration of a racer purposefully using a sub-optimal frame/wheel choice for a race, from what they have available, so they can put in decent effort and will fall behind many others and get a ZRS drop.
