Counter-Strike: Global Offensive

Counter-Strike: Global Offensive

50 ratings
Estimating CS:GO ELO / Rank Changes
By CERiNG
Using variables from a modified ELO / Glicko-2 system, the player can draw a rough estimate of their competitive skill level.
   
Award
Favorite
Favorited
Unfavorite
Preface
An update will be coming soon, which includes my personal data collection supporting the SG2 Model.

As a CS:GO player, I believe it is vital to know your true rank when participating in competitive play. Not just to see your rank itself, but a value of where you are in the scope of all competitive players. For me, it is important because seeing a numerical representation of my rank pushes me to perform better to cross the threshold of the next rank. For others, it may just be an indicator to be cautious when close to de-ranking. Regardless, I started working on this model a year ago, and I would like to share it with the public to further test this model.
Introduction
In an attempt to simplify Valve’s cryptic “ELO” system in CS:GO, I have created a multi-purpose equation to track how you place within your given rank. Using some basic algebra, you as a player can track your placement within your rank and estimate how close you are to ranking up or de-ranking.

Note: This equation is not 100% reliable and does not yield perfect results. By putting this guide out for public use, I hope to get feedback to make this formula more precise.
What affects your rank?
A popular rumor states that there are only two major factors that affect your rank: Number of MVPs, and number of rounds won. Every other in-game factor does not matter in the scope of rank unless it leads to an MVP. Because of the unreliability of these two factors alone, there is one other minor factor that is loosely based on a variable from the Glicko-2 Model. This is implemented to modify your earned “ELO” as to lengthen the process of ranking up or down.
Introducing the SG2 Model
The Simplified Glicko-2 Model is a formula accompanied by a set of rules that we will use to track your rank. I have developed this system off of the current modified “Glicko-2” ELO system that is currently used in CS:GO ranking calculation. Listed below, there are some variables that you must use to use the equation.


  • Δϵ – The change in your SG2 skill level.
  • Sᵧ – Your team’s score.
  • Sₒ – The opposing team’s score.
  • Eₒ – The opposing team’s total skill/ELO.
  • Eᵧ – Your team’s total skill/ELO.
  • m – The number of MVPs you earned.
  • σ – A “Sigma” modifier based on win/loss streak.
Your SG2 Skill Level
When you rank up or de-rank, you can start recording your SG2 skill level according to the SG2 base value of your new rank.

Examples include that...
  • Silver III has a starting value of 300.
  • Gold Nova 2 has a starting value of 800.
  • The Global Elite has a starting value of 1800.
Note: Silver I has a skill floor of 0, and The Global Elite has a skill ceiling of 1900.
Recording Team Skill/ELO
You may be wondering how you can figure out you and your opponent’s average skill level. Remember that at the end of a competitive match, every player’s rank is displayed for a few seconds. Take a screenshot of your screen when the ranks pop up.

Note: You will not be able to download your match and view the ranks from the demo.

Both the Eₒ and Eᵧ values are a sum of ranks on one team. Each given rank (including your own) uses a numerical identifier, rather than an SG2 skill value.

Examples include that...
  • Silver I has a point value of 1.
  • Gold Nova Master has a point value of 10.
  • Legendary Eagle has a point value of 15.
So, if your team is made up of 3 Nova Master (10) players and 2 Nova 3 (9) players, that team’s skill would be 48.
Scoring in Incidental Situations
If someone is unranked, simply average the players who have ranks, and assign a substitute rank for the unranked player.

For example, if your team is made up of 3 Silver Elite (5) players, 1 unranked player, and a Silver IV (4) player, the substitute rank would be 19 divided by 4, or 4.75. Round the number to the nearest integer so the substitute rank is 5.

If someone leaves or is kicked, consider how many rounds a specific bot is on your team. I usually record this information on a scrap sheet of paper, tallying how many rounds a bot is on one team. If the player returns before the match ends, subtract the number of rounds they were gone divided by 10. If they do not, use the same technique as for an unranked player, then subtract the number of rounds they were gone divided by 10.

For example, let’s say your team is made up of 2 MGE (13) players, 1 MG2 (12) player, 1 DMG (14) player, and a player who is kicked. The game lasts all 30 rounds and the kicked player was replaced by a bot during the middle of the 12th round. This means a bot was in-game for 18 rounds. The substitute rank before factoring in leave penalties is 13, but after subtracting 18/10 (1.8), the new substitute rank is 11.2. Round this number to the nearest integer so that the final substitute rank is 11.

Forfeit wins and losses modify the Sᵧ and Sₒ values rather than the Eₒ and Eᵧ values. The team that wins the forfeit game receives 32 points, and the team that forfeits the game receives the amount of points they have at the end of the game.
The Sigma Modifier
The σ value at the end of the equation is a special one, and takes on different properties when on a winning streak, losing streak, or draw. For winning streaks, simply let the value equal the number of games you have won in a row. For losing streaks, your first loss sets the value to 1/16, the second 1/8, then 1/4, 1/2, and then 1, 2, 3, and so on. The maximum value that this modifier can take on is 16, which is also the value that is set to in any tie game.
Deviation
Just to make sure the equation is as accurate as possible; the set deviation of this equation is 15 points above and below from the SG2 base values of a given rank. This means you could rank up from Silver Elite Master to Gold Nova 1 in the range of 685 to 715 points, but is most likely as you approach 700 points.
Example Calculation
Let’s say your team wins a game:
  • The final score is 16 to 9.
  • Your team is made up of all Nova 2 players.
  • The enemy team is made up of all Nova 2 players.
  • One of the enemy players left during the 13th round and came back during the 20th round.
  • You earned 2 MVPs.
  • This is the third game you have won consecutively.
  • Your current SG2 skill level is 839.
Your values will look like this:
  • Sᵧ = 16
  • Sₒ = 9
  • Eₒ = 4(8) + 1(8) - (8/10) = 39.2 = 39
  • Eᵧ = 5*8 = 40
  • m = 2
  • σ = 3
Your initial calculation will look something like this:

This comes out to the final value of 7.64 (round up to 8). Adding this to your initial SG2 skill level puts your new skill level at 847.
Closing Remarks
For your calculations, I recommend Symbolab[www.symbolab.com]. It’s a great website for calculating almost any equation, it’s free and it may help you with school work too.

If you made it this far, thank you for reading this guide, I really appreciate it! I hope to further fine-tune this guide to make it as accurate as possible.

You can donate to me by clicking the button below if you want more guides like this.

https://steamcommunity.com/tradeoffer/new/?partner=66191637&token=wvBw3Wek

Thanks again and happy gaming,
Tekryon
< >
17 Comments
SNE1KE | 60Hz May 26 @ 7:13am 
if I tie the game, in the sigma do I put 0.5?
mork Apr 23 @ 12:53pm 
2 years later and I’m wondering if this is accurate or is it only a theory
Tig3r Apr 22 @ 9:56am 
Nice post ! Does this mean that with enough MVPs you could gain SG2 points even if you lost the game ? I understand this an approximative formula, but popular belief is : you win = gain points, you lose = loss of points, no matter what
Jglake Jul 2, 2019 @ 10:56am 
Is this accurate?
Jglake Jan 6, 2019 @ 5:08pm 
Is this accurate or just a way to guess your rank?
Mob Aug 31, 2018 @ 9:24pm 
I cant eat mafs
Grape. ε>|I have stage 3 cancar Aug 31, 2018 @ 2:27pm 
are u ainstain or something or im just a retard?
thestoreybook Aug 13, 2018 @ 1:47pm 
I am a little confused on the use of the sigma modifier for streak. My confusion lies in why a winning streak number ie (1, 2, 3 etc) causes the end rank adjustment to be less each time, and why a losing streak number (0.5, 0.25 etc) causes the end adjustment to be higher? Wouldnt this be the opposide? A winning streak should increase your rank as the streak continues.
N00dles Jul 13, 2018 @ 10:12am 
this is more interesting then my math class