Christian Crab

x.com/im_Vega1/status/1835775338070056965

Vega, a VERY good Millia player, has announced a healthier alternative to rating update that is currently being worked on that ONLY factors your tournament runs! He plans to announce more details when the October patch drops, so look out for that! Here is the Twitter page of the website:

x.com/ICARUSELO

I was VERY curious about this so I asked him a ton of questions in DM's and he was very helpful and answered thoroughly, so here's the whole conversation we had! LONG post ahead, everything after this is the nitty gritty of this so if you just want the big picture you can stop here. Obviously, everything said here isn't set in stone, so don't be surprised if the final website is a bit different than what is described here.



Will only the end result of a tournament set matter? For example, will going 2-3 and 0-3 have the same effect?

**"As for results only the win or loss will matter. You don't have to worry about losing 1 game and losing a million points like rating update haha. All that matters is the set score. If you win the set you gain points if you lose the set you lose points. All scaled according to elo of course."**

(new question)



I saw you say it will be linked to start gg, will it somehow include how "stacked" a tournament is when doing calculations? For example, will TNS be more impactful than a backpack brawl or Jeekly?

(new question)

*"As for how tournaments will be weighted there will be a longer video posted in the coming days that will explain it more in-depth but, the TLDR is that for online tournaments, they will all be weighted the same but will have to meet an attendance/high elo requirement. This ensures that high elo players don't farm 10 man brackets while also making sure that people have a "safe place" to go to smaller tournaments to practice without stress. As for what that requirement is, its still being determined and will def get tuning after launch to see what people like."*

(new question)



And for the top 32 thing, I assume this will be an online event? How will this factor for people all around the world? I imagine Gobou will be doing well, will he just have to deal with the netcode when fighting anyone from the USA?

*"As for offline tournaments it will dynamically scale in weight based off attendance and high elo attendance. Stuff like evo will be worth way more points then say a regional for example. And currently on launch it will start with only NA. For right now, once other regions are added they will have to have their own elo independent elo leaderboard as mixing them like that all the time doesnt really work mathematically. But for majors and stuff where they are bound to meet there is a plan that will be put in more detail a bit later. But all of this is still heavily open to player and TO feedback."*

(new question)

Just from what you've given me, I am curious if you should factor the difference between going 0-3 in a set and 2-3 in a set. On one hand, it is probably reassuring to know that despite losing a game you can still end up positive, but on the other side, it would make going 2-3 BRUTAL. Though I think that is just the nature of competing, but it feels like it will be amplified with the winners take all nature of scoring.

*"This was actually an idea that was tested very early on in development and it ended up going in such a way that it made things work out really weird mathematically and also inflated elo very high very quickly so we decided to remove it. Thats not to say it won't be something added if a lot of people say they would like it but I personally would like to see how it plays out without it first before implementing it again due to the small sample size I tested it with"*

That makes sense, plus it would at least be different enough from the rating update way of measuring elo. Not sure why the math would be wonky there but maybe it can be visited later if needed. Just curious, how did you test this already? Just by using existing online tournaments? And if so, how would that be a low sample size?

*"Yep, just by using existing tournaments. And I was testing on a scale of 5-10 tournaments not a scale of a lot more of that which would be reality. And the big issue was just the inflation of elo being very quick. In a fgc tournament you are only losing 2 games and winning a lot more than that. Currently that inflation is relatively tame and managed by the equation, but when you start messing with how much people lose it gets to be a bit much from my initial testing off of just a few tournaments.
The issue with elo inflation that quickly being that players who dont enter often have an even harder time catching up while also penalizing those who play when they lose off a greater amount bc they are elo inflated. Although it would feel good in the short run to "gain more" by losing less it would feel bad when they lost way more for losing to someone they should be losing to anyways.
At least that was the initial interpretation but once again if things change id be happy to re implement it if thats what people want"*

oh my god you're right people win WAY more than losing, assuming they go far that sounds like a nightmare to figure out sound math for that good luck lol I got no advice for that.

*"Luckily it seems that its fine when the losses are equal to the gains but thats the great part of data is that itll be really easy to tell once it starts being populated haha. I look forward to seeing how it all plays out when it launches."*

(new question)

Just a personal concern, how many tournaments would one need to do to place high? Maybe Icarus could have a secondary use as a tool to find tournaments, to assure everyone has similar chances to enter brackets. I personally don't do more than like 2 online brackets a week and I don't see myself being able to do tournaments like every other day just cause of personal time reasons. This might just be more of a personal issue than a universal one but it's something to think about.

*"And I 100% understand that concern. Luckily, that is one of the really nice benefits of an elo system and the elo equation. Even if you don't enter that much, as long as you win, your elo should auto adjust pretty quickly to match your surrounding players if they are playing more for example. Also, since only certain tournaments will qualify based off attendance/elo there wont be a tournament every day or anything like that. It will only be 2-3 a week max is my prediction short of rare occasions. But again this is all theoretical right now and not really able to be tested until its actually live and well be happy to adjust everything once that happens."*

I see, that is reassuring to hear. A good feature to implement may be saying ahead of time, what tournaments count. But then there is an issue where, assuming that players elo influences what counts or not, you won't really know until the tournament starts cause people register so late lol. It sounded simple when I started typing but now idk how to make that clear lol.

*"Yea the late registration part is definitely the hard part in doing that 😅

A glass half full approach to it though that some TO's i talked to brought up is that people wont ever be stressed to enter a tournament bc of elo since it wont be known until afterward anyways."*

That's a nice way to look at it, but on the flip side, some of the more competitive players might want to know before hand if it will be "worth" the risk. But I don't think that level of information is possible just cause of late registration.

*"Yea for sure. I mean the plan is that as more top players start caring about it more register for online tournaments therefor increasing the number of tournaments counted. But thats really glass half full. At the end of the day there will always be at least TNS to rely on and most of the time people can tell what tournaments are going to be stacked even if there are the outliers that are fringe."*

Just off of the cuff, I think the three most "stacked" online brackets from most to least (relatively) are TNS, Jeekly, and Baccpack brawl. But this largely seems like a numbers issue that can't properly be tackled until you just have access to more data.

*"yea for sure its a numbers thing 100% that will just take tuning with time."*

(new question)

Imo the first few weeks should be test phase, then once you have (mostly) everything ironed out, you can (maybe???) do an elo reset and then let it run wild. That brings up another question actually, are you gonna consider elo resets at all?

*"The plan is that the only elo resets will be with major balance patches but the seasons will ideally by quite large."*

(end)

Thanks again to Vega for making this amazing website, and for answering all of my questions in depth! I am super excited to be able to grind this out!

And if you read all of this, thank you! Glad to know that people are equally invested in competitive Strive as I am lol.

11 months ago (edited) | [YT] | 19