The unacceptable flaws of the round robin format

Round robin has been the dominant format since Counter-Strike moved online earlier this year, but it has a number of flaws that, among other issues, present dangers to competitive integrity.

By early June, round robin will have been used in group stages at the biggest tournaments for nearly three months in a row, with ESL Pro League Season 11, ESL One: Road to Rio, DreamHack Masters Spring, and BLAST Spring Showdown all featuring the format after the coronavirus pandemic struck and the CS:GO scene moved online. But there was a good reason why it began giving way to the GSL format eight years ago, after Tomi "⁠lurppis⁠" Kovanen had called for changes in mid-2012, and why it ended up eradicated altogether from big LANs in 2020, when BLAST finally gave up on it after two years. So why are we going back in time?

Round robin is an appealing format for tournament organizers for several reasons. Perhaps the biggest one of all is that the large number of matches played typically translates into more broadcast hours than other systems such as the GSL or Swiss, which means it takes up more of the day and the calendar and/or gives fans more variety in case of concurrent streams, allowing the tournament to attract a bigger audience.

With ESL Pro League producing record-breaking numbers during the pandemic, almost reaching a whopping 500,000 concurrent viewers mark, it's difficult to fault organizers for hogging as much time as possible. That approach makes perfect sense from a business standpoint, in particular at times like these when many of us are stuck at home, glued to the screen with nothing better to do than watch some Counter-Strike, and when a new audience is more accessible than ever.

To a degree, there's an argument to be made that it is also the most effective way to find the best teams in the group stage, especially when the participants are split into two groups as opposed to four or more. With everyone playing each other, you get a larger sample size than with other formats and a more accurate representation of where the teams stand among the rest.

But when it comes to non-ideal scenarios in which tiebreakers have to be used, the first problems appear. Even today, when time constraints are not nearly as big an issue as during LAN events and when schedules should allow some leeway, potential rematches are far on the bottom of tiebreaking scenarios and tournament organizers prioritize factors like map difference or worse, round difference.

You can see what these scenarios look like in an example from the ongoing ESL One: Road to Rio's rulebook:

1. Points amassed between the tied participants (direct match win > direct match loss)
2. Map difference between the tied participants (3:2 maps > 3:3 maps)
3. Number of map wins between the tied participants (3:3 maps > 2:2 maps)
4. Overall map difference
5. Overall number of map wins
6. Round score difference between the tied participants (23:21 > 23:22)
7. Number of round wins between the tied participants (24:22 > 23:21)
8. Overall round score difference (39:31 > 40:33)
9. Overall number of round wins (40:32 > 39:31)

Only after nine possible tiebreakers — out of which only the direct head-to-head results should matter — would there be rematches, and even in the virtually impossible case where we get that far down in the order they would be run in a shortened, overtime-like setting:

10. Sort by win-loss difference in the tiebreaker overtimes.
11. Sort by round score difference in the tiebreaker Ots.
12. Sort by number of round wins in the tiebreaker Ots.
13. If there are still unsolved ties, replay the tiebreaker Ots.

The exact same ruleset was used in ESL Pro League Season 11 and ended up eliminating Natus Vincere in fourth place of the second group stage in Europe due to map difference in a tie between them, MOUZ, and Astralis. Just before that, G2 had exited the tournament early because of the same scenario with FaZe and OG in the first group stage. Road to Rio is heading for a similar situation in both groups in Europe and MIBR already suffered the consequences in North America after a tie with Liquid and Envy. One map loss that held no significance within the series victory against the latter team cost MIBR a spot in the playoffs and could theoretically end up causing the Brazilians to miss the Major.

Teams simply don't operate under the assumption that they can't afford to lose a certain number of maps or rounds. They don't plan vetoes to have the best chance of winning 2-0, they structure them to have the best chance of winning the series, and they aren't thinking about how many more rounds they can afford to lose during a map beyond stopping the other team from reaching 16. It is ridiculous to try to objectively decide whether a 2-1 series in which the map scores went 16-3, 14-16, 16-5 was a better result than a 2-0 series where a team won both maps in overtimes, so don't. It shouldn't matter. The win is all that matters.

Often times round robin also produces a different kind of situations in which a team's fate relies on a result completely out of their hands. SK found themselves in that undesirable position during their run to the title at BLAST Pro Series Copenhagen 2017, where they watched on anxiously as Astralis and FaZe were locked in a nail-biting battle that would have snatched a grand final spot away from the Brazilians had Nikola "⁠NiKo⁠" Kovač's team won the match.

This inherent dependence on results other than your own and the lack of proper tiebreakers make the round robin a shoddy format at best, and those are just two of its many problems, as well as perhaps the least worrisome.

One of its more dreadful characteristics is that more often than not at least one team's fate is determined before the end of the group stage. For example, look at Complexity and Dignitas in Road to Rio; they are about to play a match that, if it weren't for the 200 points' difference in the Regional Major Ranking in this case, would be competitively meaningless apart from a $1,000 increase in prizemoney, because both teams are already out of the running for the playoffs.

This essentially creates a showmatch-like environment in which neither side cares about the result, leading to bad Counter-Strike being played at best and at worst producing a sketchy habitat that is more likely to be fixed for betting earnings. If one side does have something to play for while the other doesn't, there's even more potential for the result to be manipulated, as there is a competitive advantage at stake. And it gets even worse when a team has the power to decide whom they will meet in the next stage of the tournament based on the outcome of a match they're about to play.

To see how this could work in practice, consider the round robin group below:

Astralis (4W - 0L)
Cloud9 (3W - 1L)
Natus Vincere (3W - 2L)
MIBR (2W - 1T - 2L)
FaZe (0W - 2T - 3L)
Ninjas in Pyjamas (0W - 1T - 4L)

If this seems familiar, it's because this is a real situation that occurred at BLAST Pro Series Lisbon in late 2018. At 4-0, Astralis had already guaranteed themselves a spot in the grand final, and they played Cloud9 in the last round in a match that would determine their opponent in the title decider. The two possible outcomes were very straightforward: Lose or tie and face Timothy "⁠autimatic⁠" Ta's squad, at the time ranked 26th and playing with Ismail "⁠refrezh⁠" Ali as a stand-in, again, or win and take on the third-best team in the world, Natus Vincere.

Although it was close, the Danes ended up winning the match 16-14 and beating Danylo "⁠Zeus⁠" Teslenko's team in the final, avoiding any scrutiny coming their way, but imagine yourself in their shoes. It would be seriously tempting to consider losing the match — or not trying your hardest at the very least —, wouldn't it?

If so, you can see what dangers the round robin format presents to competitive integrity on top of all the other issues it has when it comes to unclear placings that come down to rubbish tiebreakers and depend on results the team in question has no power over. Does it really need to be used when with a little creativity more competitively viable and less dangerous formats could serve the same purpose?