The Swiss format has made waves in the Counter-Strike scene since its introduction at the ESL One Cologne qualifiers last year. Since then, it has been included in multiple tournaments in CS:GO, including the ELEAGUE Major. It was then inducted into the Dota 2 scene for the Kiev Major last month. After watching it play out so many times, I have a much better grasp of its positives and negatives and how it affects the CS:GO and Dota 2 scenes differently.
First, let’s note the big difference between the two games when it comes to implementing the Swiss system in tournaments. Counter-Strike has a much larger LAN circuit compared to Dota 2, which allows for more international competition and makes it easier to make accurate rankings at any given time — especially when Dota 2 has many more game-altering patches. That is critical when we consider the implications of the Swiss system.
The point of any tournament format is reward the best team in the tournament. In order to do that, you try to avoid scenarios where the second or third best team meets the best team in early rounds. Otherwise you will have situations where the final standings are influenced more by fortunate matchups or competitive “dead zones” than gameplay. Therefore, good seeding is imperative to maintaining the competitive integrity of a tournament.
The two games differ greatly at that point.
In Counter-Strike, there is almost nothing resembling a true offseason. Top teams regularly face a slew of competition, ranging from their equals in ESL Pro League and ECS to weaker, less-known competitors in online qualifiers. The relatively robust LAN circuit in Europe and North America ensures teams regularly play each other from month to month. That makes it completely plausible to seed every team at an event. There’s enough data and methods to filter it to justify every step of the ranking process. In reality, all of that is ignored at the Majors. Valve counts only the results of the previous Major and the Major qualifier and ignores everything else. In contrast, Dota 2 tournaments do some kind of internal seeding, but making it an accurate representation is more difficult because of the smaller number of LANs.
The two games diverge again when it comes to the consequences of running through the Swiss system. In Counter-Strike, only half of the contending teams advance to the next stage after running through the format. The Dota tournaments that have utilized Swiss system so far create seeding based off results; all the teams advance to playoffs. For the sake of convenience and broadcast time, both games depend on a single-elimination playoff bracket. After the first round, draws are randomized without reseeding. For instance, teams that went 1-0 in the first round won’t be reseeded, so there’s nothing to ensure the best seed plays the worst seed of all the teams with a 1-0 record. Further draws are random from there.
Because no teams are eliminated in the group stage for Dota 2, its system is superior for reflecting the actual form of the teams. Since the only purpose of the group stage is seeding, it gives more chances for the better team to advance. A good example of this being useful in CS:GO was OpTic failing to reach the playoffs of the ELEAGUE Major. Regardless of OpTic’s actual gameplay, the team was put in a precarious position from the start. OpTic’s victory at ELEAGUE Season 2 and second place finish at ECS Season 2 were ignored, and as a result, the team drew Virtus.pro in the first round. After losing to VP, OpTic drew Astralis — the best team at the tournament and eventual winner — because Astralis was upset in the first round and also 0-1.
Two other problems constantly emerge in the CS:GO system. First, all matches are best-of-one compared to the standard best-of-threes in Dota 2. Part of the criteria for distinguishing stronger teams from weaker teams is the concept of a “deeper map pool,” which is best shown in a multi-game series with a meaningful pick/ban phase. OpTic’s disastrous road in the Major was thanks in no small part to some unfortunate map results. OpTic had to face VP on Cobblestone, one of the former’s best maps at the time. Part of the reason OpTic had to play Astralis next was because Astralis played Train in its loss to Godsent — perhaps the only map possible for Godsent to upset the Danish team.
The second problem is that the last map is randomized because the pick/ban stage only accounts for six maps. If not for random chance, OpTic could’ve faced VP on Cache, a far more alluring proposition. The Astralis/Godsent matchup was decided by a pick between Train and Overpass; if the latter was selected, Astralis almost assuredly would have won and advanced to the 1-0 pool. I don’t know why tournaments insist on randomizing the last map because all maps are being played right now.
Despite my reservations, the current Swiss system is still adequate on a practical level. In CS:GO, Valve still refuses to seed based on any results outside of the Major. The Swiss system may be optimal purely for its ability to correct any initial poor seedlings that occur. In Dota 2, it feels as if Valve doesn’t want to use the The International’s format for the Majors, so the current choices are either GSL system or Swiss system. In that sense, Swiss system is better because the lack of LANs means it is harder to get correct seedings for GSL groups.
So long as Valve refuses to seed CS:GO Majors to account for performance over time or refrains from using TI format in Dota 2 Majors, Swiss format seems to be the best one possible. If we had accurate seedings, I believe GSL groups would be better for CS:GO tournaments because of guaranteed Bo3 series. I’m still undecided on Dota 2 in that aspect.