Race ranking change request

Hello @xflintx

I raised this in the old zwiftpower forum. Now that you have worked through some other issues maybe you can have a look.

I really like the race ranking calculation but it has a major flaw in the way it is calculated.

Basically only finishing riders are included in the formula for calculating. This means anyone who is part of race but does not cross the line has no effect on ranking performance which I think is unfair.

Every race there are large groups starting but not finishing (various reasons) and they should be part of the way rankings are awarded.

Suggestion

Keep the race QUALITY calculation the same this is right and fair in current method.

Adjust the average ranking of finishers to be average ranking of starters in the points per place calculation. This can be anyone finishing first KM as an example.

Thanks in advance

I think that all of your comments are talking about ranking points for a given race and I agree with you.

Another topic for discussion is the assigned Cat of a racer and when they elect to race above that Cat. Racing up should be encouraged and is the opposite of a sandbagger. What if the racer only does sprint races less than 20 minutes though and is never assigned a higher Cat? He races and wins B Cat races but is still ranked as a C Cat. It is kind of riding below 3.2 on occasion to keep yourself as a C but then racing sprint B races to gain the overall #1 C ranking.

Here is a video where I looked into this a little while back. ZWIFTPOWER.COM CATEGORY C RANKING REVIEW - YouTube

Does this tactic effect the race points only for B since they race as a B or does it effect C racers as well?

Thatā€™s definitely an issue. There are a few other problems with ranking and rider behaviour. I would like them looked at individually so as not to complicate matters.

I first got alerted to the issue I posted when I raced wheel to wheel up the alpe only to have the rider quit in last 100m. He had his reasons (weird) but I was denied the opportunity of the improved ranking points by the action.

If you look at live tab on zwiftpower there are always people dropping out mid race and they are not all tech issues.

@Wattwell Yes, there is definitely an opportunity here.

I placed #3,500/4,200 in the Haute Route GC, which required completing 3 stages.

There were 23,000 stage completions reported on ZP (which wouldnā€™t have been 23,000 unique riders, and would unlikely have been less than a third of that unless people were repeating each stage multiple times. Letā€™s call it 12,000), so now my finishing place should be 3,500/12,000.

But that was 23,000 completions, which ignores the non-finishers, non-starters, and registered-but-no-shows. Adding those back in might double the total number of participants. Or more. Or less.

There were also several disqualified riders. They wouldā€™ve been captured in the 23,000 number, but not the 4,200 number.

So I know I finished 3,500 of the riders that completed all three stages without being disqualified, but I truly have no idea how many total participants were present.
3,500/4,200?
3,500/12,000?
3,500/23,000?
3,500/46,000?

I really do not know.

1 Like

The chess world uses the Elo Rating system to pair players. It is well documented.

iRacing has adapted this for fields of more than two players, and named it IR.

IR is a trailing average of finishing positions in recent events, and includes did-not-finish, did-not-start, did-not-show, and disqualified results in the calculation.

At the start of an event, the server assigns riders to the appropriate split based on their IR.

@Clive_Norton I counted up just over 1517 people completed all 3 stages in Cat C. Zwift seem to have counted people that have done 2 x Cat C races and one Cat B (or A) but that adds no more than 200 - 300 people to the total completed 3 Cat C stages.

Whilst they have worked through some issues, they still have a bunch of stuff they broke when they moved the servers that need addressing before they could/should look at dev work.

Having said that, I totally agree with you. And Iā€™m ok with people joining and then exiting early for whatever reasons, but when Zwift give us tools such as Race Quality that guide us and the Quality changes as people drop out early, I do feel a bit cheated.

But I dont think Zwift will put any effort into addressing this (or fixing the stuff they broke). They barely look after racers as it is and they have demonstrated over the past several months that even with very high visibility issues they cause to racers, they will still not put in adequate resourcing/investment to help us. Wish it was different.

Voted

@Dean Hi Dean.

In just the stages I was in, I was typically in ~400th place within a few minutes of leaving the start pen, and typically finished in the high 200ā€™s without passing many other riders.

Considering that happened 3 times, thatā€™s 300 missing entries right there.

Then add in all the riders that didnā€™t complete all three stages, as their ratings were not counted.

Then multiply it by 24 for all the hours in a day.

Then multiply it by the 4 (or 5) categories.

Itā€™s a large number.

I grabbed all the ZP data and have it in a spreadsheet.

23000 stages completed by Zwiftpower racers (thats a different number than you see on screen for start as you are counting non-ZP racers). Not all events were heavily attended - 1st event for example had a total of 439 riders across all Cats.

Remove:

  • people that entered wrong cats (DQ)
  • people that exceeded ZP thresholds
  • people that only did 1 stage
  • people that only did 2 stages
  • people that did 3 different Cats

Other stats
533 Completed 3 all Cat A stages
1309 Completed 3 all Cat B stages
1514 completed 3 all Cat C stages
586 completed 3 all Cat D stages
91 completed 3 all Cat E stages

So thats 12000 of the 23000 activities accounted for right there.

Thanks, @Dean
Youā€™ve been able to grab the data from ZP in a way that Iā€™ve yet to work out - I can only find the 3 stages in which I competed, and the final results which show as empty.

You and I are analyzing the data with different goals.

If, hypothetically, 30,000 riders participated, and only 20 riders finished all 3 stages, it would be very wrong to describe the rider in 20th position as being in last place. In my mind, that rider finished 20th of 30,000.

Iā€™m trying to work out for this event (and any other), how to obtain the number equivalent to the 30,000 in the hypothetical example above.

If you would be so kind, can you tell me from your spreadsheet how many uniquely different ridersā€™ names appear?

If I correctly understand what is in your list, that will provide the number of riders that completed at least one stage.

There is another number I would like, that counts unique riders when including the DNF, DNS, and DQ results.

I did the same as Zwift - just grabbed all race data published by Zwiftpower for all races, and all Categories.

I get that, and that is the problem with ZP ranking that Wattwell describes and wants fixed (as do a bunch of us).

But for racing results, only Zwiftpower data is ever used and what I dont have is non-Zwiftpower riders and non-finishers (Zwiftpower and non-Zwiftpower) - both the numbers you are after.

Zwift will/may have total numbers but wont have total per Cat as non-Zwiftpower users arent categorised so weā€™ll never know how many were genuinely in each Cat.

4033 unique names. Zwift has 4249 which includes riders that only completed in 2 of one Cat and one of another (which I didnt count)

Thanks for the explanation, @Dean

You and I are counting differently. Iā€™ve found a (rather tedious) way to get data out of ZP, and so far have done so for the final 10 start times of the total 71 start times.

From those last 10 start times, there are already 4,475 unique names, including DQs, and not counting DNFs.

Iā€™ll keep scraping away at the data. If thereā€™s an automated way to do this, Iā€™d love to know.

Sorry yes, I answered incorrectly. 4033 unique that finished 3 stages in the same Cat.

No easy way I am afraid, scraping is what I did - took about 5hrs, so not the most :slight_smile:

Hey, @Dean

Finished - that was tedious.
Including the disqualifications (DQ), there were 31,216 finishing records, involving 16,736 unique riders.
So now when I tell my wife I finished 3,500/16,700, itā€™ll sound better than 3,500/4,200 :rofl:

Not included are
1/ The results from Saturday at 1500 EST - they are absent from ZP.
2/ DNF - Riders that started a stage and did not finish one.
3/ DNS - Riders that registered for a stage and did not start one.
4/ Riders not on ZP

I should have given you all my data as I have zero spreadsheet/database skills so it took me a while longer than you!

Well done @Clive_Norton :beers:

Thereā€™s an email (presumeably to all) saying there were 34,000 participants in Haute Route.

Iā€™d tell my wife that I was 3,200 / 34,000, but I think she doesnā€™t care.

1 Like

haha - I can just visualise the false smile ā€œyes, thats very good dearā€ while she goes about her business.

I donā€™t know what is wrong with some people

I told my wife that I have over 100,000km on Zwift and I got the same reaction as telling someone what day of the week it is without being asked.

1 Like

Iā€™m not allowed to mention bikes at the dinner tableā˜¹ļø

1 Like