As more tournaments continue to pop up in Counter-Strike, it’s worth taking a look at what the most optimal format might be — even looking toward other games for guidance.
From my perspective, there are five qualities a format must fulfill to varying degrees: Competitive integrity (How good the format is at getting the best team to win the tournament), time, number of games, narrative structure and entertainment.
In general, competitive integrity for CS:GO tournaments can be fixed by using a proper seeding system, whether that’s from a panel of experts or using HLTV’s rankings to seed the teams. The biggest trespasser on this instance are the Majors, as the seeding works in such a way that it ignores all other events that happen outside the Major despite selling itself as the culmination of all CS:GO and the peak of competition.
Narrative structure is very similar to competitive integrity, as the whole point of a tournament is to find out who is the best team at the event. The most important point for this part is to have an easy structure to follow so that it takes less than 30 seconds to understand and explain. MLG was the worst trespasser of making easy to understand formats.
Time and amount of games are similar to each other. Time is how long you have to run your event. This will give you a basis for what formats are possible. The number of games is how many games are necessary in order to fill whatever quota of games needed to get sponsors.
The last quality to think about is entertainment. This is a vague term, as tournament formats are limited in how much they can affect how a tournament is viewed. But if you use the correct tournament format, you can turn an ordinary tournament into something different.
So, what’s the best format for a major tournament? I’ve explained and graded eight that I’ve seen run in watching esports tournaments. I gave every format a rating from 1 to 10 in each of the five categories I mentioned above, making 50 the maximum score. They’re not listed in order, but you can see them ranked 1-8 at the end.
Global StarCraft League Format
This is essentially the standard format. It happens when you split teams into groups, have them run a double elimination and then get seeded into the bracket. Although it isn’t necessarily exciting, you know exactly what you get and won’t have to deal with a mess. It works for all skill levels of tournaments, meaning it doesn’t matter if the teams attending are Tier 1 or 3.
Competitive integrity: 7. This can go up or down based on seeding, so I averaged it out at 7.
Narrative structure: 7. There is no re-seed once you get into the playoffs so you can get some wonky brackets where the true “finals” match is in the quarters or semis.
Time: 10. This system is used so often because there are no extra tiebreakers to be played.
Number of games: 8. It’s the standard so I assume this is what most tournaments want.
Entertainment: 7. Not bad, not good. Entertainment depends on the teams.
League of Legends worlds format
This is a double round robin group stage that leads into a playoff bracket. This is good if you have a lot of time, and the double round robin is interesting because of how it was scheduled. Basically you have a week where you prepare for specific opponents each day and then the next week you fight against everyone in your group all on the same day. It tests a team’s overall form on a day and how well it prepares. I’d use this only if you had a large global event with a lot of time.
Competitive integrity: 8. This rating is dependent on good seeding. Based on the way Riot seeds worlds, it would be a 5 or 6.
Narrative structure: 8. You get a better group stage, as there are more games to be played and thus more lines of data to draw from to make a narrative.
Time: 6. The entire system is based around having a month long tournament with only one stream. You could bump it up a point if there are double streams for group stages. Still has problem of tiebreaks.
Number of Games: 8. Gives just as much game time or more than GSL format.
Entertainment: 7. There are some things better and worse. I’d need to see it played out with good seeding, but as only Riot uses this format, that doesn’t happen.
The International format
Again, this can only be used for super large global events that go on for at least two weeks. It is the best format for finding out who is the best team at a tournament. Sixteen teams are split into two groups where they play round robins to decide seeding. They are then placed into a double elimination bracket based on the results. It’s great because teams have the most chances to prove themselves in this format.
Competitive Integrity: 10. The best format I’ve seen for players. There are no questions as to who the best team was.
Narrative Structure: 10. This has even more data points of the group stages than GSL/LoL worlds, but it also has a double elimination. The more matches there are the better you can convey a story.
Time: 6. This can be run faster than Riot’s worlds, as group stages are for seeding, so a double or even triple stream is expected. The problem is running the double elimination bracket, which takes time.
Number of Games: 10. Same reasoning as the previous two.
Entertainment: 10. As the system should only ever be used for tournaments with the best players/teams in the world, every match is interesting.
Double elimination format
This is used in EVO. The perks of this tournament compared to the previous three is that it skips over the boring group stage and goes almost immediately into elimination mode, which is every fan’s favorite moment to watch. The problem with this is that bad seeding can ruin the tournament, but at least every team gets two chances before they are out.
Competitive Integrity: 8. Depending on how big the format is, sometimes the best player or team is knocked out because of a bad draw that pitted the team/player against a singular bad matchup, whereas against anyone else they would win.
Narrative Structure: 9. This gives the classic revenge arc, where a player/team gets dropped by someone early and has to make it all the way back to the top. Also, elimination is inherently fun.
Time: 9. Pretty easy to do and organize, though things can run over depending on if you’re running a best-of-one or best-of-three.
Number of Games: 6. This won’t give you as many games as the group stage formats and sponsorships needs those hours. It is optimal in FGC games, though.
Entertainment: 8. Elimination from the start is more fun than group stages.
IEM Oakland format
This features two round robin groups, and then the teams are seeded into the bracket depending on how well they did. This format requires a double stream at least. The good part is it’s a bit better than the GSL format at measuring the strength of the teams in a group, but much worse in that it can (and often does) create tiebreak scenarios.
With tiebreakers, there is no legitimately good way to say who should go on. Head-to-head ignores all other context of the group, as the whole point is to see how well teams do in the group rather than against a single opponent. Round differential and time scores are both dumb (in Counter-Strike). When possible, the tied teams should play a game to see who makes it out of a group. At the very least, they need to play the tiebreaker for teams on the verge of elimination. Maybe have teams play overtime against each other for tiebreaker purposes (again, in Counter-Strike). It’s not perfect, but it is better than using head-to-head or round differential.
Competitive integrity: 8. Better than GSL, not as good as TI. Also tiebreaks not being decided by more games knocks this down a bit.
Narrative structure: 8
Time: 8. Knocked down because of the potential of tiebreaks.
Number of Games: 8.
Entertainment: 8. Goes through the group format faster.
The ESL One Cologne Major qualifier used an altered system where teams were eliminated after losing a specific number of matches or made it through based on how many they won. This is an OK system, but the original Swiss format is a non-elimination style where you play a certain number of rounds. The problem is meaningless games can be played, but they can be skipped (assuming there are no sponsor obligations to stream them all).
Competitive integrity: 8. Sample size is small because I rarely see this format used
Narrative structure: 8. Easy to follow
Number of Games: 7. Fewer games than round robin
Entertainment: 7. It’s a better group system but not quite as exciting as single elimination.
This format is used in the playoffs of League of Legends Champions Korea and in the StarCraft Proleague, which I called pagoda. It is best used for qualifier tournaments, which seed the winners into another tournament. A good example of this was the recent IBP Masters. In that tournament they ran a typical GSL style but had only the top seeds advance to play a final. The problem was that the prize money allocation was spread evenly so from an outside perspective, it didn’t seem that interesting for a spectator.
The pagoda format is in contrast to that. You run a group (either round robin or Swiss) of about eight teams. In a round robin that is about 28 games in group stage or in Swiss that is about 12. In this case, because the top four were seeded to go to Oakland, you let the top three move on. Then you seed the remaining five in a pagoda where the eighth seed plays seventh, winner plays the sixth and so on until the final elimination match for the last berth. From an entertainment standpoint, that is much more dramatic than a standard double GSL where the top two qualify.
Competitive Integrity: 6. It’s terrible at measuring a team’s ability in best-of-threes.
Narrative Structure: 7. Easy to follow, but fails a bit if it’s used as a full tournament rather than a qualifier.
Time: 9. Should be able to be done fairly easily
Number of Games: 7. You don’t get as many playoff games.
Entertainment: 8. This is averaged out. As a qualifier, it is a 10 but as a regular tournament that just decides a winner, it’s a 6.
Red Bull/ESL Barcelona
This is a format where each team gets three lives and after the initial starting match, the winner gets to choose which two teams play next on what map. In theory, it sounds great, but in reality the strong teams first eliminate the weaker teams to get the money. A good way to fix this is instead of having the teams choose, have a host choose the matchups so they can specifically get rivalry matches or create interesting matches that they want to see. You could even have players try to convince him to play a different team or give them a specific map with their arguments. The ideal person to choose is Thorin.
Competitive Integrity: 3. Does this even count as a tournament?
Narrative Structure: 10. It’s basically the WWE. What else do you want?
Time: 10. Easy to do.
Number of Games:: 7. Fewer than you’d probably hope for.
Entertainment: 10. If there is no host that forces good matches or drama then it falls to a 3.
Total: 40 (or 33)
Ranked in order:
1. The International (46 out of 50)
T2. Double elimination (40 out of 50)
T2. Double elimination (40 out of 50)
T2. Red Bull/Barcelona (40 out of 50)
5. Global StarCraft League (39 out of 50)
6. Swiss (38 out of 50)
T7 League of Legends World Championship (37 out of 50).
T7. Pagoda (37 out of 50)