Race Scoring (Ranking) - a new discussion

There are many flaws, for me the biggest one is that it only reaches a roughly accurate rank for very frequent racers. This means the lower rungs of the rankings are packed full of strong riders, which would ruin the race experience for anyone genuinely ranked in that area.

All solved with the ELO-MMR approach.

3 Likes

You’d think, wouldn’t you. But virtually every study shows that if you have to actively cheat, by clicking ‘yes my weight is correct’ then people are far far less likely to do so than if they can passively keep it the same

2 Likes

Fair point, and should be an easy change with no negative impact.

2 Likes

This thread has great ideas, but unfortunately, it should be recreated in the “Feature Request” forum, and then obtain the correct (which is an unknown value) number of petitioned votes.

@Robo_Bane_ABR

I think the ranking will help a bit with this issue, the only advantage with weight change in a ranked system is you go faster and that will move your rank up so you race against faster and better riders.

….until the top tier at which there’s no downside to cheating. Sorry, I realise I’d sidetracked this discussion on ratings.

Adding the current rank score to riders’ names automatically in races and TTs in-game might be a good thing.

Rob, that is one of the issues with the ZP ranking system. The ZP system works ok assigning Race Rank points but only when all those within the race are ranked broadly the same - within 10% of the average of best 5 from top10 finishers. I believe it is what it then does with that Race Rank score that lets the system down somewhat. Hopefully something else better is being discussed within this thread.

If your ranking is more than 10% lower than the average of the best 5 finishers in top 10 then you can not achieve a Race Rank score better than your current ranking - even if you win. (Except see note 1. below)

If your ranking is more than 10% higher than the average of best 5 finishers then you are bound to get a Race Rank score better than your current ranking even if you come dead last.

I think I have found the race you are using as an example (2021). In that race the average of 4 best racers was 337. ( race shows winner with no ranking, I think it was 600, so only other four were used)
Your ranking before the race was 361, so less than 10% over the 337 so race results should be ok. You received a Race Rank score of 371 so, as you expect for coming last, worse than your 361 ranking. So why did you receive a Gain of 2 in your rank? Because ZP system uses average of your best 5 and this 371 Race Rank score was 10 better than your worst score of the 5. If ZP system was using only your last actual Race Rank score you would have gone down in ranking BUT it is the fact that ZP uses average of 5 which resulted in you gaining in ranking even for coming last.

Hopefully this can be addressed by using a better option currently being put to Zwift.

Note 1. Similar to just explained (I hope above !) for coming last and gaining the same applies in reverse. Winning a race where you are clearly by far the best Ranked racer in the field. Your Race Rank score is almost certainly bound to be higher than your ranking but if it is better than the worst of your 5 counting scores then you will gain a ranking as your average will be improved.

Thanks Ian! Great analysis. I was actually referring to this race:

Though similar applies. The fifth rider was DQd for no HR data.

I’d entered the race with the hope of improving my 20 minute power numbers and not knowing at the time how ratings worked, had no idea I’d get a bump. I think I had a goal at the start of the year to get into the top…500? In the UK on ratings. And this was the race that did it for me, but it felt like the goal was a stupid one and since then I’ve largely ignored ratings as my 9 year old could’ve come last in that race and gotten the same boost!

Another aspect about it that feels wrong is that it’s also self-perpetuating, especially if you ride in small fields as I quite often tend to (I have weeks where I race at 5/6 am). By having a low rating, I make the race quality higher (number lower) by my own presence. Ratings systems shouldn’t work that way.

Yes this race was a clear example where you couldn’t fail to improve your ranking.

When a system uses average of best 5 race rank scores regular racers are almost certainly going to trend in a better ranking direction. Again I don’t think this is the fault of the Race Rank score being allocated in each race but how that score then fits into your best 5.

If everyone was racing flat out trying to get the very best ranking they could you would only ever need to use your last Race Rank score to be your ranking. Race well - ranking improves, race badly - ranking gets worse.

But what we all fear is race badly, race badly, win your next race. Rinse and repeat.

2 Likes

New ranking system coming today folks. We have a trial event where pens are split according to your rankings. Scheduled for 18:30PM UTC on Wednesday, keep it free!

9 Likes

Is a higher ranking better in this model? i.e. Is the top category “Cat E: Platinum 3 and Higher”? If so just wondering why it wasn’t called Pen A.

Why do you think? (not deliberately not answering, I’ll answer afterwards)

Well, personally I would prefer platinum and gold to bronze given current market conditions.

Pens are just pens. Pen A doesn’t mean any different to pen D. I flipped the pens around (so highest rankings go in Pen E) so you think before you click, and to get people thinking a little differently.

Is this Tims btw?

1 Like

Gotcha. How far back does the data go to put people into pens. I guess I was a bit surprised I managed to make it into gold pen only having 2 races this season, both being very lackluster results with ZP ranking of 450+. I’m thinking the data would have to go back a bit further than 90 days.

Allllll of the information will be shared within the hour, hold your horses :smiley:

2 Likes

Here you go: https://forums.zwift.com/t/introducing-zwiftracing-app-zwift-racer-rankings-done-right/

2 Likes

Deleted 8