News Ranked Championship - Season 1 2023

So this is not allowed then?
"To prevent 1 driver from winning a series in all regions, you are not allowed to join the same series in multiple regions."

 

What do you think if splits work more like this?



The lines dividing potential splits are more vertical than in the chart above and it reduces the impact of reputation. Otherwise I'm not sure it's a good idea if the same server has drivers with 1900+ rating and 1500+ rating competing against each other. Not even in terms of who has better chances for high results, but also in terms of atmosphere. 1500+ drivers would be too slow for the fast guys and 1900+ would be too dirty/aggressive for the clean/safe guys.

You can change the slope by adding a power to the formula. In this particular chart the power is 0.3:
Rating * (Reputation/100) ^ 0.3

It's just an example.

 

I found a picture from 2021 plotting rating vs. reputation among all drivers with activity score >0, and I highlighted the area of 70-100 rep and 1500-2000 rating on it.


And here's this area flipped upside down to make it easier to compare with the "splits" charts above:


You can see there's actually very few drivers with high rating and low reputation, so my worries are probably not justified.

 

in blue what the formula does upon a sample data excluding ghost riders (0 activity till an event to retain rate rep)
An elongated and straighten curved let to more easily decide the splits.. well thought.

btw, the formula is straight forward:
having 1750 one gonna retiain in percentage the rep of the score ( 1750 rate, rep 80... 80% of 1750)

the cleaner, the better

 

That's only when cleanliness is the goal. But the goal here is "fair" splits so if some adjustments make it better, then it makes sense to use them.

This is how the current formula divides the 70-100/1500-2000 region. Look at the slope of the lines; where the lines actually end up depends on how many people sign up for a race. (These aren't actually straight lines, but close enough.)

My question is, does it make sense to put these two drivers in the same split? They have almost the same "weight" in the formula, so they will always be on the same split no matter how many splits there are. But one is 98/1570 and the other one is 82.5/1860 - do they really belong in the same race?



With an adjusted formula the lines can go something like that, and edge cases end up much closer to each other in terms of competition on-track.

 

to state the difference among the simple rate rep... or the formula i've these 2 :

rate / rep:


the formula :


the point wich cuttings probably not state well the effective laptime and pace pilots can keep into the track, is true.

probably a mandatory pre-race time attack on the same race track, to better split splits, would be nice, if also useful for fairer splits groups.


P.S: rainbow gradient purposed over and over, is used just cuz is very simple using HSL color to state a diff having numeric values... also, i'm traditionalsit, at most for me is a peace flag if not the rainbow itself (max extend of visible light gradient)

 

I don't know why you compare these two. Your first picture is simply division by rating, this is not going to be used in the championship anyway.

A formula that includes both rating and reputation makes sense if you want to also promote clean driving, not only pure speed. This is understandable and I'm sure many will agree it's a positive thing to have. The question is what formula would be better for that.

 

The engaging part of a race is having similar pace drivers around that add bit of variability in following perfect laptime solo.

So a perfect formula would try to pick a smooth gradient of similar pace laptimes, to promote both driving fights for the position, and the spectating engagines

None of this take in consideration pace, so is more about a first coarse split the one group (the all drivers) in sub-groups to eventually sub-split again and again till 32.

For the purpose of first split in many sub-group and promote a bit fairness (reputations) the formula is good as is.

The comparison simply display that the basic rate/rep would have bigger sub groups in lower rates, and smaller in higher rate.

The official formula sub-groups/splits in more well distributed groups, heavly sub cutting lower rates, and almost not touching higher rates.

So in first cutting is good as is.


Edit:
the both graph with even 32 cuts:

without formula + groups of 32 cuts


formula distribution + groups of 32 cuts


the formula, as is, well works.

btw, are distractive decoys, the core fun is in race with similar pace drivers, as long as possible, thanks to having around also similar fairness drivers

 

Probably a mandatory pre-race, 5+ laps, recorded time attack laptimes, the more the merrier.
Should help in having min, max, average, 25% and 75% percentiles of drivers effective recent skill in the combo.

should fit better for multi-race events, than generic daily races ratings.

In general, without over-complicating things, is good as is.

 

@Andi Goodwin , can you please cut posts from #22 to this included, and create a new thread with them, if also @Maskerader agree pls ?

We got a bit Off Topic.

 

I can't shake off the feeling that you don't understand what I'm talking about.

It just works better than nothing. But your posts never addressed the concerns I raised and don't prove that the current formula is better than what I suggested.

I wouldn't want it to be moved to another thread because what I'm talking about is not an off-topic.

Upd:

It makes sense to use rating instead of any sort of pre-qualifying if you want to use this championship to promote ranked racing. You want to get into higher splits because they have significantly more points, and to get into higher splits you want both your rating and reputation as high as possible.

I'm sure that's what's happening, because it's called "Ranked Championship", not "Raceroom Championship" or something.

 

Id say is good the formula beacuse at higher rates, the lower rated + high reps can race with slightly higher rated as opportunity.

While at mid rate the gain/possibility is higher, promoting a good behaviour at lower rates than at higher.

here the 3 graph cuts:


as you can see in next image the reputation parameter bad and good outlings... so the cuts is more about how much punish / promote in lower / upper grouped categories same colored, here :


another case... :


this last case is the most intresting.... 2 pilots, same ratings... one has 95+ the second has around <65 reputation.
How bad or good behave a 95+ rep compared to a <65 having same ranked rating ?
(trying to respond this, came into mind the 2nd question... how many races they have ? and how much the rep goes up and down [rep consistency] )


P.S: could be also added the "how manies races one did" and "the reputation consistency" into the formula to better tune it bit more (just came in mind, but didn't thought howto)

 

here the graph with initial rate/rep cross + finer grids colored with official formula (cuz looks good )

 

Yes, the difference between formulas is in how much your rep affects what split you get.
Another alternative is this: rating * (reputation + C)/(100 + C), where C is like 100, 200, etc. Same effect (reputation has less effect on what split you get), but with a slightly different curve.

Opportunity of what, exactly? Driving in the bottom part of the grid and being lapped by fast but dirty drivers doesn't sound quite exciting to me. Although you get rewarded with higher points.

 

low rep is not always about being dirty, is also about being fast but consistent (mistakenly self out tracking).
Yep the opportunity is pretty much about Ratings.

A low Rated driver whom is climbing and improving has an opportunity in racing in same race with a messy high rate one, because being in front of higher rated pilot grant an higher rating gain score.

Also, the high rated/ low rep driver who race with same rated pilot with higher rep, is risk factor for drivers with better reputation which chasing even higher rate/rep zones.

BTW: high rating is not always being fast, but having best position among the drivers one find in a race.
So is not an absolute indication of fast driving, is mostly a relative positioning indicator among drivers one find in races in a given period.

For instance:
a group of driver having an average 1850 rating in a given period 1, move scores relative to each others positioning.

a 2nd group of drivers in a period 2, having same average 1850 ratings, also move scores relative to each others positioning.

But the group of period 1 compunded can be way faster than the group of period 2.
So, ratings movements are relative to drivers one find rather than being absolute fast.

 

Yes, but people race against a lot of different drivers. They are never stuck racing against the same small group of drivers. Which is why a faster driver will have a higher rating (on average).

 

True.
Picking the afterrate and afterrep dots cluster density of given period... seems active drivers rounds given scores areas:


so variabilty is probably higher in denser areas.

It also displays/let guess (remembering the cutting bands) that the original formula grant more groups in the denser areas

 

Huh? The number of groups solely depends on how many drivers sign up for a race and drivers will be divided into equally-sized groups, no matter how dense their area is.

 

True.
It does affect only the choosing order as below.
(if one does not participate, the next along the line path).



N.B: are graph made on fly, to be consider diff color lines united among first of a color to the last of previous (right to left)