Counter-Strike 2

Counter-Strike 2

63 ratings
How the ranking system works
By RaimaNd and 3 collaborators
Many people speak about valves official csgo matchmaking systems and there are a lot of theories out there. This guide is based on collecting data through roughly 1.000 MM games over 3 years and will show the results. It doesn't explain in detail how it works, but it gives a very good direction.
2
6
2
   
Award
Favorite
Favorited
Unfavorite
General explanation of Glicko2.0
Valves official matchmaking system is using a heavily modified Glicko2 system. Behind the 18 ranks are hidden (Glicko2-)points. We don't know how many points exactly are between the ranks tho. When you rank up you are at the bottom of the new rank pointwise and you'll derank quite fast and need to work through the whole point range to rank up again. This also counts the other way: When you ranked up you'll derank very fast, usually it takes two losses in a row to derank. That shouldn't make you sad since - again - you're close to a rank up and usually it takes 2 wins to rank up again. Usually the pointranges are also not fix, that's why you don't derank directly after one loss when you ranked up. So they vary a bit so the ranks don't shift that much around.
Valve Statement

Faceit uses an elo system which works quite similar and to visualize it we can use it as an example:

As you can see each rank has a range of points and when you "hit" a specific point range your level will go up.

This are two example, look how fast I ranked up again:

Keep in mind that this also can go the other way, shown here:
Comparison between both team Glicko2 ratings
Elo- and Glicko2 rating system also compare both teams and count the Elo/Glicko2.0-Points together and based on the points it calculates how the match should end. When one team now performans better than it was expected they will earn more points and vise versa.

Example:
5 globals play against 5 globals. It's expected to be a close game. When now one team wins by a lot, they get even more points.

To make it as easy to understand as possible here an extreme example:
If 5 silver 1's play against 5 global, they will barely lose any Glicko2 points because they're expected to lose. They will never gain "positive" points in a lose, but a 3:16 will be huge already. And when they win or tie they get a big bonus.
Each round matters
Winning is good. Tieing is good. Losing is bad. But that isn't all. Unlike Faceit system where you always lose/win the same points no matter how many rounds you won in a loss and lost in a win it's different in Valves official matchmaking. A 16:12 win will give your team less points than a 16:4 win. That also means that losing 12:16 isn't as bad as losing 4:16. This should always be in your mind. When your team is winning by far you shouldn't joke around but finish the game to get as much points out of the game and at the same time don't throw a game just because your team is losing 0:14 already. Always try your best and win as many rounds as possible without losing as many rounds as possible.
Ranking Deviation
The less you play the more each game counts. This can be used in a positive but also in a negative way, in this case it took me two wins after a rank up to rank up again - played over half a month:

When you play more it will also take longer to rank up as shown here:
Ranking decay
When you don't play at least one match in 28 days your rank will decay. This means that you're shown as unranked. It requires you to either tie or win a game to get a visible rank back again. You'll lose some hidden Glicko2.0-Points and on average you lose one rank. This means when you were LE for example and your rank decayed, you're usually DMG afterwards. If you're still LE afterwards it means you were already very high in the point range of LE and now are very likely in the lower end on the LE point range.
Myth 1: Personal Performance is very important
To answer it short: Personal performance matters, but not as much as people might think. It's completely ignoreable.

Explanation and evidence:
I mainly played with the same ~10 people throughout these 1.000 matches and the personal scores are very different, yet the ranks we've are the same. When personal performance would matter much, then I'd be global and they'd have a lower rank. This can be researched easily by browsing through these 3 links, compare my stats with the player and the ranks we have:
https://csgostats.gg/player/76561197974580372#/players
https://csgostats.gg/player/76561197978008075#/players
https://csgostats.gg/player/76561198875387116#/players

To make it even more obvious a friend and me created a new acc each and we played til 10 MM wins, all together. I tried to play as good as possible and he played a bit worse so that our stat differences even are way bigger to make it as different as possible. Based on these stats, if they would count "a lot" I'd get a much higher rank. The results:

Links to the matchhistory:
https://csgostats.gg/player/76561199128434668?date_start=1614466800&date_end=1615244399#/

https://csgostats.gg/player/76561199145053459?date_start=1614466800&date_end=1615244399

As you can see my total stats were a lot better. I received MGE as my first rank while my friend got MG2. This means even when all stats were far better I just got one rank higher than him. If the stats would matter much then I'd have gotten a way higher rank than him. This proves that personal performance does matter - but not as much as most people might think.
Myth 2: Are ties positive, negative or neutral towards the rank?
To answer it short: It's positive.

Explanation and evidence:
Throughout all games collected it only happened once that a player deranked after a tie and the reason for that derank was that he left the game early when his team was losing by far, match:
https://csgostats.gg/match/12266753

In any other tie the only rank changes visible were upranks, many times upranks of multiple player - even for both sides. In roughly half of all ties at least one player ranked up. (Note: in many of these games the bottom fragger ranked up, showing that personal performance + a tie is even positive):
https://csgostats.gg/match/6236429
https://csgostats.gg/match/5436565
https://csgostats.gg/match/4005360
https://csgostats.gg/match/3901789
https://csgostats.gg/match/29707500
https://csgostats.gg/match/28695502
https://csgostats.gg/match/7083709
https://csgostats.gg/match/3782971
https://csgostats.gg/match/38118276
https://csgostats.gg/match/36620935
https://csgostats.gg/match/14481922
https://csgostats.gg/match/11583811
https://csgostats.gg/match/10256801
https://csgostats.gg/match/10199001
https://csgostats.gg/match/8640469
https://csgostats.gg/match/8602276
https://csgostats.gg/match/7696731
https://csgostats.gg/match/7453817
https://csgostats.gg/match/5132690
https://csgostats.gg/match/4521423

These matches are especially interesting since many people ranked up (mainly again "bottom fragger"):
https://csgostats.gg/match/4348835
https://csgostats.gg/match/6172203
https://csgostats.gg/match/23648165
30 Comments
RaimaNd  [author] 12 Jan, 2023 @ 8:17am 
Thank you! :lunar2020ratinablanket:
Rakaa3 12 Jan, 2023 @ 12:40am 
Pretty fine explanation
RaimaNd  [author] 25 Nov, 2022 @ 6:40am 
Thanks bro, your page is awesome! :steamthumbsup:
rchh 25 Nov, 2022 @ 5:59am 
great use of csgostats.gg! <3
RaimaNd  [author] 10 Nov, 2022 @ 1:54pm 
In your specific case there aren't many matches uploaded so we can't guess what is going on in your games, but based on these 8 games you lost 6 games, mainly by a large margin. You also play solo and since you play in EU you play against very equally ranked player, meaning, you play against S1-S3's most of the time. The two main parts to rank up is to win games / rounds and play against higher ranked player. You can affect the chance of winning rounds by performing well tho so keep that in mind. If you struggle with randoms, which happens quite a lot even in the highest ranks and very likely in most Silver games (so I feel your issue) you should look for friends/teammates to play with. This not only makes more fun, but also increases the chance of winning by a lot.
RaimaNd  [author] 10 Nov, 2022 @ 1:54pm 
Hello again,
the age of guides don't matter as long there wasn't any changes affecting the guide. The ranking system was changed completely in 2016/2017 thus guides before that time aren't up2date anymore.

Guides made since then tho are up til now. We had a little rank shift in august but it didn't affect how the ranking system works. It was more a bug fix because it wasn't working in some regions where the amount of player was too low which affected how the automatic rank system calculates things.

In EU nothing changed for example because the playerbase is still very big. You can say the guide is the role model how the ranking system works and when it doesn't work as written in the guide it's not the fault of the guide, but because the guide has dependencies (player count).
Yetoia 10 Nov, 2022 @ 12:49pm 
In a thread someone think your guide is too old. I don't think because Glicko2 still counts for ranking. Can you say something about it?

https://steamhost.cn/steamcommunity_com/app/730/discussions/0/5239403117925045257/?tscn=1668113095
RaimaNd  [author] 10 Nov, 2022 @ 10:38am 
You're welcome Yetoia!
Yetoia 10 Nov, 2022 @ 9:34am 
Great guide thanks for your research in this. :steamthumbsup:
El Capitán 20 Aug, 2022 @ 7:46am 
thanks for the info!