Is there a perfect CS:GO Major format?
Join me as I delve into the CS:GO Major formats over the years and propose an entirely new format of my own creation.
Preamble
On the 13th of November 2022, Outsiders lifted a Major trophy. It was the last in a string of unfortunate events that occurred at that Major, including Cloud9 and NaVi losing to underdog teams, FaZe and NiP going 0-3 in the Legends Stage, and G2 not qualifying at all.
The latter two got a few people thinking, including me - is it the fault of the formats of the RMR and Major not giving teams enough of a chance? Leaving the reductive logic of “they lost so they must not be good enough” aside, it seems that a simple change removing the Bo1s in favour of Bo3s would make most people happy.
But aside from the most popular complaints, there have been a few interesting things going on with the recent Majors. Teams like Copenhagen Flames and Bad News Eagles qualifying to the Major, to the Legends stage and even to the playoffs with “easier” routes, the seeding that resulted in multiple top teams facing each other early on in the Legends stage of IEM Rio - some of these issues are not as simple as removing Bo1s.
Therefore, I thought it was only right to look back on the Major formats used historically, discuss some notable points such as seeding, qualification methods, the number of teams, and so on. A massive shout out to everyone at Liquipedia; without their site, finding information for the older CS:GO Majors would have been next to impossible, and as I am newer to the scene, I have nothing but respect for anyone doing the hard yards and cataloguing all of this.
If you want to keep up to date with what I’m working on, or just what I’m thinking, my Twitter is still the best place to go!
The GSL Era
Long ago, when Swiss was just a type of cheese, CS:GO Majors used the GSL format. This format is still used in many CS:GO tournaments, such as for the group stage of IEM Cologne 2022,* such as ESL Challenger Rotterdam 2022, and is a double elimination format that works best for splitting groups into the top two and bottom two.
*Correction: GSL groups originally referred to groups of four teams - the double elimination groups of eight teams used in IEM Cologne 2022, as well as any other group sizes, are not the same as the original GSL group format. Thanks to Collin @CarbonDogma on Twitter for pointing this out!
Every Major from DreamHack Winter 2013 until ESL One Cologne 2016 used this format for the main stage of the Major (the top 16 - the quirks of the Offline Qualifier and all seeding issues will be discussed later). Teams would be split into four groups of four teams based on the seeding of choice, with the top two of each group qualifying to the playoffs.
This format is one of the easiest ways to decide the top 8 teams in a 16 team format. Teams know who is in their groups ahead of time, giving them more room to prepare more specific strategies, and compared to the Swiss system that succeeded it, it is easier to construct narratives from a selected group.
The largest detriment towards this system is the so-called “group of death”. Having a relatively small group size of four meant that on occasion, one group would be significantly more difficult than another. For example, in Cologne 2016, Group D consisted of SK, G2, fnatic, and FaZe - the top three ranked teams in the world according to HLTV at the time, and FaZe at #17.
While bad groups are not the sole fault of the GSL format, using it in smaller groups makes it easier for them to be imbalanced compared to each other as the variance will naturally be higher in these scenarios. Perhaps the deadly group at Cologne 2016 was the final nail in the coffin for this format being used at Majors.
The Swiss Era
A format many have grown to hate over the years, the Swiss format is best used for events with a large number of participants such as an open chess event. Cologne 2016 trialed the Swiss system in the Offline Qualifier, and from ELEAGUE Atlanta 2017 onwards, it has been the format used in the Major’s top 24.
Since the format is used to find the top 8 out of a group of 16, teams are qualified after three wins or eliminated after three losses. Much like the GSL system, this gives teams a clear goal. It is a tournament format that does not rely on seeding as much as other formats, as it matches up teams based on their score (1-0 vs 1-0 and so on) and therefore, in theory, should naturally aggregate the best and worst teams.
Part of being a top CS:GO team is being able to prepare effectively for your opponents - as matchups are decided after an entire round is completed, this gives teams one night to prepare for specific matchups. While a GSL group would also only reveal your opponent one night in advance, the small group size means it is feasible for teams to prepare for every possible opponent in a more impactful way.
The most infuriating part of this format, in my opinion, is the lack of control teams have over their possible matchups. Anyone who has spent time fiddling with that handy matchup calculator (shout out) knows that changing one result in a round can result in a complete reshuffle of the next round’s matchups. This “randomness” also means some teams inevitably get an “easy” path (as mentioned in the intro) - and it feels less fair than if the same matchup were to occur in a smaller group, where you know it is a possibility from the start.
The Challengers Stage: An Awkward Format?
“Did you know that dupreeh is the only player to have attended every single Major?”
The above fact has been repeated ad infinitum ever since it became true - but before that, it was a bit more confusing. What exactly is the Major? If you search in the event archive on HLTV and sort by “Major”, the Challengers stage does not appear. At IEM Rio, dupreeh did make it to the top 16, but if he doesn’t next time, will the fact turn into a factoid? What is the truth?
Every major since ELEAGUE Boston 2018 has had a Challengers stage - before that, there was a LAN Qualifier/Offline Qualifier for every Major since 2015 in some capacity (except for Cologne 2015, which had regional LAN qualifiers). From Katowice 2014 onwards, Valve began releasing sticker capsules for each Major, and until the LAN Qualifier was rebranded as the Challengers stage, only the top 16 teams were available.
While Valve would like you to believe that the Challengers stage is as valuable as the Legends stage is (hence their stickers), many people still believe that the rebranding of the qualifier does not make it any more legitimate. As a fan of Oceanic and Asian CS:GO, while I would love to pretend that it is as prestigious as the main stage, there is a reason why tales like Bad News Eagles and Imperial making it to the top 16 are notable - because the Legends stage is still seen as where the “real” Major starts.
The conversation about whether it is part of the Major or not belies the reality - that it is simply an awkward format that doesn’t feel like the Major. Assuming that the best teams qualify as Legends, the Challengers stage is filled with teams that scraped through the difficult European qualifiers and the already subpar teams from all other regions. It is difficult to give it the gravitas of the CS:GO Major if there are almost no matches which could catch the interest of people not invested in the scene to begin with.
I believe that Valve simply wanted to sell more stickers and rebranded the stages for this result - but the second problem is that a 24 team format which decides a top 8 is not as easy to construct as one may think.
Keep reading until the end to see how I would create the perfect* Major format!
*according to me
Seeding: What Works Best?
Even the perfect format would be failed by improper seeding. Aside from the very first Major, which I investigated for a while before simply assuming that it was an arbitrary ranking based on how the teams qualified to the event, Majors have had a few different ways of seeding teams.
The most common method of seeding was based on their performance at the most recent Major (if they are Legends) or based on their performance at the qualifying tournament (whether it was the LAN Qualifier, regional qualifiers, or Minor qualifiers). Some recent Majors have implemented the Buchholz scoring system after the initial Swiss rounds, which seeds teams based on the ongoing performance of their past opponents, to create better Swiss matchups.
The most notable forms of seeding aside from this are the formats used in the post-COVID Majors and the 2019 Majors. From PGL Stockholm 2021 onwards, every team has to qualify through the RMR system, meaning all seeding is dependent on the results of the RMR and Challengers stage. The latter led to some interesting results, like Cloud9 and Vitality being the worst seeds at IEM Rio’s Legends stage due to them barely scraping through the Challengers stage, causing some terrible matchups.
The seeding used for Katowice 2019 was unique - it started with a “self-seeding” process where teams ranked other teams, and outliers were discarded to create a ranking. This ranking was used as a base for a “Live ELO” algorithm which would increase or decrease the “ELO” of teams based on who they defeated or lost to. This meant teams that were underrated would increase in ranking, but not by an unreasonable amount. Berlin 2019 used the same system, but opted to use HLTV’s ranking instead of self-seeding - both would use the final ELO standings to seed playoff spots.
While I think Valve will no longer allow their Majors to rely on HLTV’s ranking or a self-seeding process which involves some bias, the ELO concept has me intrigued. Recently, Valve has started working on a “Valve Ranking”, and have shown interest in releasing and accepting feedback on their algorithm. If Valve is interested in creating a simple ELO formula in tandem with their ranking, this could be the way forward instead of relying on shifty Swiss seeding.
Valve’s Directions, Harsh Truths
A title like this would lead you to believe I am about to give a dreary perspective on how Valve does not care and thus we will never get anything we want. This is not quite how I view things, however.
Valve clearly cares about some things when it comes to Majors, and looking at their decisions makes it easier to understand.
They care about having 24 teams in what they call the Major, due to various reasons (sticker money).
They care about Legends spots continuing to exist to some extent, considering that every change in the Major format has given these teams some advantage, from getting a free invite to the Major to skipping an entire stage, and the spots have never been removed.
They care about the Majors being independent and open circuit, considering the way they structure their Major circuit revolves around giving every team an equal chance.
They care about viewership numbers.
This last point is the most notable because it means everything else that is wrong with the Major can be fixed by appealing to this point.
Things like Bo3s instead of Bo1s, new group formats, even a complete restructure of the Challengers and Legends system can be proposed through this lens.
This brings me to the harsh truth - that the bottleneck for hosting a Major is not Valve being stubborn, it’s the TOs cutting costs and using Valve as a scapegoat.
Consider the changes that you could make immediately to the Major that would fix most of the issues people are complaining about - best of threes is the most popular request. Is there any reason that Valve would prefer a Bo1? It’s clear to me that while Bo1s might benefit viewership in the short term of a tournament, the more consistent format will lead to more viewership later on as the established teams make it through more frequently.
Valve have already realised this. We don’t know the extent to which they intend to implement changes, but we know that they now have an official ranking of their own, and that they plan to implement a closed qualifier based on this ranking that allows teams to skip the painful online qualifiers. This means that they are aware of the seeding issues which plagued the last Major, and that established teams having to play in open qualifiers is not necessary - both of which they would not care about if it did not harm viewership or sales.
The conclusion I have reached is that a format needs to be proposed which a tournament organiser is fine with covering the costs of, rather than one which specifically caters to Valve. The passive approach Valve takes to Majors should not be viewed as an obstacle, but as an opportunity to innovate without too many restrictions in place.
This brings me to the most exciting part of this article - my own Major format.
A New Major Format: BPB Groups
Okay, so I stole the naming convention from GSL groups, because they are similar but pretty different. Have a look at this diagram for starters
There’s a lot to unpack here, so let’s get started with the thought process.
I wanted to get rid of the different stages of the Major, making it a 24 team event. Splitting it into two stages meant that using Swiss would end up recreating the existing format in a weird way (trust me, if you try selecting a top 8 from 24 teams using Swiss, you’ll end up with the same conclusion), so I created a modified GSL format with four groups of six teams. Having larger groups means it is harder to create a “group of death”, while allowing for a more predictable and straightforward group stage format.
There are eight Legends spots, with two per group - the Contenders and Challengers spots can be used for seeding purposes as they are now. You may find that the example group at the top looks familiar, because it is! A similar format is used for the playoffs of Cologne and Katowice nowadays, with the difference being in those events the teams that get to skip a match are the upper bracket winners of the previous group stage.
In my format, the advantage given to the Legends teams is that they need to win one less Bo3 to progress to the playoffs - given that the previous advantage was skipping an entire stage, or skipping at minimum one Bo3 and two Bo1s, this seems entirely fair.
The losers of the second round play each other for the chance to play the loser of the upper bracket final and secure their spot in the 3-1 pool, while the upper bracket final winner is guaranteed playoffs.
Winning three Bo3s gives you a chance at playoffs, and losing two Bo3s definitely eliminates you. Currently, a team can get eliminated via two Bo1s and a Bo3 - since one Bo3 is less “random” than two Bo1s despite them being the same number of maps, this also seems fair.
Everything so far seems clear, but there’s a few big questions left - what about the winner of the 0-1 bracket? What seeding will be used?
The seeding will start with Valve’s official ranking - while imperfect, it is much more reliable than using RMR results. However, I do not recommend using the points directly - I intend to use this ranking to rank teams from 1 to 24, which will be the initial seeding. This will then become the base for the “Live ELO” system that was used in 2019’s Majors to seed any matches that are not predetermined by the groups, such as the playoffs.
I spent a long time trying to figure out how to incorporate these teams meaningfully until I was inspired by Cologne 2015’s “reseeding in groups” method to create the Special Bracket. This bracket is a single elimination bracket where the 1-1 teams who lost their first game are placed (seeded according to their ELO) and a “winner” who goes 3-1 is found.
If you’re keeping count (or if you read ahead on the diagram), you have realised that there are five 3-1 teams and four 3-0 teams. Naturally, the 3-0 teams are already in playoffs, but what about the 3-1 teams?
This is where the Live ELO comes into play - the two teams with the lowest ELO in the set of 3-1 teams play for the final playoffs spot in the Last Chance Qualifier (not a very unique name). This may initially seem unfair until you realise that the format does not eliminate any teams based on ELO - it simply forces the two “worst” teams to play an additional match instead of giving them a free pass.
This is, of course, subject to some variance as teams will face more or less difficult teams and will have their ELO adjusted accordingly throughout the tournament - but as you still need to lose that last chance game to be eliminated, I consider this fair.
The playoffs are seeded based on the ELO scores of the eight qualified teams, with 3-0 teams put against 3-1 teams. I don’t mind whether it is a double elimination bracket or not, but since I have my doubts about whether a TO is willing to pay for an arena for as long as a DE bracket takes to run, I am assuming the same format as it is now.
As for the length of the games - 8 Bo3s are played per group, with the Special Bracket + LCQ adding four more games for a total of 36 Bo3s in the group stage. The Challengers stage takes 20 Bo1s and 13 Bo3s - doubling this to account for the same format in the Legends stage and subtracting the Bo3s means that there are 40 Bo1s in the existing format for an equivalent 10 Bo3s, making my format shorter!
In fact, if two concurrent streams run 3 Bo3s each per day, my format will take 6 days - realistically, it should take around eight due to scheduling the Special Bracket and LCQ appropriately, which makes it the same length as the current format from a broadcast perspective.
There are a lot of minor details to be ironed out in this format, such as more specific scheduling, how the Special Bracket, LCQ, and Live ELO would be explained, and whether anything I’ve come up with is misaligned with Valve’s mysterious mindset. Overall, it was the best I could come up with and I believe it has a lot of great aspects to it (especially the 9th place spot - a team can definitively be the “closest to playoffs” in this format rather than the 9-11th we see now).
Conclusion
If you’ve made it this far, thank you for reading! Yet again, a shout out to Liquipedia for providing me with information about the older Majors that I would never have been able to find otherwise.
Sorry that this article got delayed so much - my own life aside, there was a lot to dig into with the Major formats, and I almost lost interest before eventually deciding to complete it before the year ended. I intend to focus on more contemporary topics, like the “winners and losers” concept that HLTV used to do, as well as more opinion pieces as I feel more capable as a writer and CS:GO analyst.
Until then, a belated Merry Christmas and a Happy New Year to everyone!