If you caught the Dreamhack France finals this last weekend, then you saw some of the changes that are in TU8. Specifically the changes made to Trials Fusion’s Spectator mode. The most requested changes are going to be gone over again a bit, as these had increased priority based on feedback from you.
Ranked MP is making its console debut for Trials Fusion, and as much as I would like to go into all of the details… I can’t. Not because I have been sworn to secrecy, but because when math starts to have more letters than numbers, my brain says “I’m out”. For those who are so inclined, here is a link to a PDF by Professor Mark E. Glickman that goes over the “basic” math behind the type of system we now have in place for Fusion.
User tracks in Ranked MP is something that is now possible, but we have not yet added customs in. Be prepared for those when they start landing, but remember that tracks (especially single lane) need to meet some minimum playability standards, so don’t be offended if your favorite tracks aren’t included.
Xbox One brightness and Ghosts for all platforms have been gone over again to make playing in full games a bit less confusing, but again, this is all stuff that you know.
What you may NOT have known is that this update is now live on Xbox One & Playstation 4.
You can check out a full list of changes in the Update 8 Patch Notes on our forums.
Hello,
I am going to answer in the official thread to keep everything about the "how the new ranking system works" question. I'll redirect all the different threads to this one.
The goal of the ranking is to give a way for players to be able to compete with each other in a real time context. (compared to leaderboards which are supporting asynchronous competition)
The ranking system we are using is linked by Shifty in the initial post. It's really hard to understand for people (like me) who are not totally comfortable with advanced mathematics. However, the system ensures that in a context like we have for Trials MP, people get a ranking that matched their skill. It takes into account the performances of the different heats of a race and not only the final result of the race. It also takes into account the ranking of the opponents. It shouldn't be possible to have negative ratings if you win all the races, don't do any fault. However, several players are reporting it, so we're looking at it.
Several things that can be happening:
- Mistake in the implementation of the method. (but we already verified several times, so it's unlikely)
- Having a context that is a little bit borderline for the method. (maybe when we have a very wide range of rankings in the same match)
- Something else.
We're going to investigate and see how we can tweak the algorithm to avoid the unfair situations. I'll let you know when we have some conclusions.
There is another thing to take into account that is now not really well explained: We need a certain number of matches for the ranking to stabilize and for the system to work properly. I don't know if this has a link with the negative ranking change that we see sometimes when a player wins everything but it's not totally impossible. We'll see if we can hide the ranking until it's stable (maybe 20 matches) in a next update.
About the deviation, an helpful resource to understand would be wikipedia. http://en.wikipedia.org/wiki/Standard_deviation
I would also like to precise what that means exactly in our context: Lower is the deviation, more consistent is the player.
We're going to investigate what we can tweak and how to improve this and I'll get back to you on this thread later.
Cheers,
May be instead of displaying the ranking points at the end of the 4 heats, it would be better to display it at the end of each heat.Originally Posted by En0- Go to original post
Now I understand why I losed points for my last game. I won 2 heats on 4 but still be the 1st in the final result.
Ok. Let me begin by saying that I am not a mathematical genius. I think everyone already knows this, but it's important to say. The actually workings of the Glicko-2 system fly so far over my head that it will likely hit another planet.
That being said, here is my understanding from reading the NOTES made about the system, as well as a spiffy little graph I made up in Photoshop (It wasn't photoshop). Think of this more as a question to En0- that I just happened to post on the forums, as it kinda makes sense in how I currently understand it working.
The minty fresh green line is the rating of a highly skilled player with an average score of 1800 (top 1%)
The grey line is the is the rating of a not so skilled player (top 95%)
A marks the average scoring for that player across however many matches they have played.
B marks the match in question, where the highly skilled player still wins the overall match, but ends up losing "points".
C marks the difference between the highly skilled player and the lower skilled player.
From my limited understanding, the closer that two players from vastly different skill groups are performing, the more change that those players will see at the end results, especially if a limited number of matches have been played by each player. This would mean that a player who was previously performing at a very high level may encounter "score loss" (actually a change in ranking) due to barely winning against players of a much lower level. This would result in a larger gain by the lesser skilled players and a loss by the higher skilled player to bring the two closer in line to their actual match performance.
Much like if you match a pro football star (eggball or "soccer") against a youth league player. If the football star wins by a single point, then there will be questions about how well that star is actually playing.
I tried to make this as easily understandable as possible while leaving the math out of it. Does that make sense, En0, or am I completely wrong with how I understood it working? I had a brief conversation with someone on Twitter last night where I explained roughly the same thing, so if I am completely wrong, please let me know so I can correct it.
Hello,Originally Posted by oO-shaolin-Oo Go to original post
I believe it's something that is not possible because we need the result of all the races to get the value but I will verify.
Regards,
We'll look at that.Originally Posted by NSPIV Go to original post
Cheers,