It just doesn’t work wonderfully though.
small rant incoming
Similar to the penalty system, it is implemented with good intentions to improve the racing quality of online lobbies but falls short in effectiveness as you push the boundaries because of how simplistic its algorithm is.
The algorithm is as follows, assuming you are A and other drivers are B...
- You lose 80 + ((A.DR - B.DR) / 500) for every driver you finish behind.
- You gain 80 - ((A.DR - B.DR) / 500) for every driver you finish ahead of.
The calculation takes nothing into account except for finishing position. There is no bonus for fastest lap, no bonus for a CRB, no bonus for overtaking, and no penalty for finishing the race with a red SR (beyond the hard resets).
Where the failures of the system become most glaringly obvious is with the door numbers. Everyone knows that the door number you get in a race roughly corresponds to your DR rank within the lobby. Lower numbers means you have higher DR compared to the rest of the drivers (Ignoring the D/C ranks because of PD’s infinite wisdom to use String math).
So if the DR system “properly” represented your driver skill that would mean finishing around your door number would result in around a net zero change in your DR, right? Get into a lobby where you have the 10th highest DR, aka “skill”, and you should theoretically finish in 10th place.
This obviously is not how it works. Instead we have a system where regardless of your “theoretical” placement within the lobby you are almost always guarenteed to gain DR with a top 10 finish and lose DR with a bottom 10 finish. Scaling with how far away from center you are.
So no, the system doesn’t work properly. It “works” in the same way that the penalty system “works”.
No, you finished in pretty much one position below halfway (so 11th in a 20-car race, 9th in a 16-car race, 7th in a 12-car race).
Broadly speaking, the front half of the field will gain points, because they will finish ahead of more drivers than they finish behind, so lose points to fewer drivers than they gain points from. The opposite applies to the back half. High DR differences across the grid will change that somewhat, but that's the rule of thumb.
Right... in FIA races, the spread from 1-20 DR rank is not usually too broad, so it's pretty simple: finishing 10th or 11th means you will be neutral to DR change, at the end, usually. Your 'door #' will determine whether that break point is 10th or 11th, but it doesn't vary much in most circumstances otherwise (a completely inverted finish, say 20th DR guy winning and 1st DR guy last, etc., may be an extreme that tips it somehow, if your result was completely out of step with that). The 'door #' may come more into play with wider range matched lobbies in Dailies or something, but for the FIA events it's usually pretty easy to sort out.
In case people haven't seen it before, there is a nice estimator available here on an older post, but it's hardly worth pulling out because it's typically as simple as the above.
https://www.gtplanet.net/forum/thre...ion-to-maintain-gain-dr.379934/#post-12398462
To that end, my thoughts on the DR system, having read Grenadier's remarks:
1) There is some theory or point of view that any measuring system that FORCES change (in this case, DR moving up or down) instead of trending toward stabilization, are more effective in some cases. I think there's a little of that at play, in the interest of variety and not matching racers with the same people all the time. Also to force changes in behaviors, so that people who want to get good, rise up if they are successful, and people who don't invest in it are relegated over time, rather than allowed to hold station. In that sense, I think it works well.
A race is a race, and I'm not racing the guy ranked ahead of me, I'm racing the grid... the DR system should have that perspective as well. The weighting of the 'ante' based on the DR differences is a nice accounting of the variation in the grids, which is to be expected in an open environment with any range of matching potentials.
2) If I'm matched with a grid of racers based on 'some' criteria, I do like that I'm more or less grouped as 'equal' at that point. The lobbies are binned in order, as far as I know, so lobby to lobby, there should be distinguished sets, but within the lobby, once we are there: we are equal and have more or less equal chance, albeit with weighted movement depending on who we beat. So again, I don't think that's a flaw, compared to just beating my DR rank/door number. That would be kind of boring and we'd all sort out with the same guy in front or behind us as well, which would cause people to complain!
3) I totally agree the mechanism between the SR and DR is flawed, based mostly on the flawed SR system, though. So if they'd just address that, I think the DR system itself, as it stands, is pretty effective.
FYI, post referenced above:
TLDR Fill in the Yellow boxes on the spreadsheet to find your target finishing position.
1. Formula for FIA Points calculation:
Points for race winner = (Average DR x Number of Drivers)/479.2
(the 479.2 was rounded to 1 decimal place, but still calculates within 1 point)
2. Target Finishing Position (FIA races only)
As we can see the points available in FIA races, the average DR can be calculated.
Since we also know the equation for the DR calculation, it's possible to quickly find the target finishing position.
I've made a spreadsheet in which you fill in the yellow boxes and it calculates the position required to maintain DR and the DR gains/losses for every finishing position.
(The spreadsheet only work with 20 driver lobbies at the moment but I plan to fix this at some point)
