Looking for the perfect competition

This article is translated from HLTV

Let’s take an in-depth analysis of the three group stage formats in use today and work together to find the perfect format – one that can determine the best teams in the group stage to place them in. Group seeds in the inner ring as fair and accurate as possible.

For at least a decade, we’ve been searching for the perfect group stage format. The format can identify the strongest teams and place them in the correct ranking order in the first group stage of the tournament. The seeded teams will then be placed in their respective brackets. These things don’t happen often, and that’s why let’s work together to find ways to improve.

There are three basic competition formats in use today: round robin (round robin), GSL and Swiss. At the highest tier of Counter-Strike, round robin – is the oldest of the three, and is the format used in both traditional sports and other esports. But round robin has been pushed back, we still see them sometimes in the recent major tournaments, like at IEM before stopping completely at Katowice this year.

The GSL format (named after the Global StarCraft League) appeared in the final year of Counter-Strike 1.6, 2012, after Tomi “lurppis” Kovanen introduced GSL and explained its benefits. It became the standard in CS:GO Majors, with many other organizations following suit. DreamHack is probably the side that uses it the most, from the open round to the bigger Masters. ESL also added changes to their tournaments in 2018.

Swiss, the format that comes from chess tournaments, is the newest format in CS:GO. Teams pass this round by going against teams with the same win-lose record, until they reach the required number of wins to advance to the next round, or lose the allowed number of games and be eliminated. Swiss was first used at ESL One Cologne 2016 Main Qualifer, and in early 2017, the format was adopted at the Major.

It is understandable that the organizers use different competition formats. The tournament is organized by them, so they have the right to choose the system they consider most suitable, whatever the reason, such as the schedule of matches, the number of participating teams, the choice of track of viewers.

However, at the Major, due to Valve’s participation in the organization, it needs to be revisited. Valve is the ‘parent’ of CS: GO, which is also the event with the highest scale. Therefore, they should use the best format to select the 8 strongest teams to enter the next round, as accurately as possible. And that can be seen as the perfect form.

Let’s take a look at the three modes and explain the holes before picking our own.

Round Robin

Obviously round robin has the biggest vulnerability. The lack of tiebreakers, especially in cases where more than two teams have the same score, is the most concerning. Usually these cases are decided by the difference of rounds between the same teams, but that is difficult to see as a solution, as teams should not be too worried about they may lose to their opponents on difference. the rounds; Winning the game is the most important thing.

The best way is for the same teams to play against each other, but that is often unlikely when the schedule in the group stage is quite tight. That’s why, ahead of IEM Katowice 2016, ESL introduced the mr3 tiebreaker: teams will play against each other on the same map but in overtime, after completing the veto. Luckily it only happened once, at Katowice 2017, and we see that ESL has abandoned this idea and the round-robin format.

Second, the results of the round-robin group stage could have been determined before the final matches took place, suggesting that these matches are meaningless. For example, it happened in the group stage of ESL Pro League Season 6, where Luminosity played against Liquid and North faced NiP at the end of the group stage with nothing to lose, other than a few minor changes.

While this problem rarely happens nowadays because of the high risk, round robin allows teams to intentionally lose to avoid facing a difficult team in the next round. This can be limited by scheduling matches, allowing both groups to take place at the same time, but this cannot be prevented completely. In the end, round robin only determines the strongest teams in that group stage instead of the strongest of all teams, so it’s still pretty much dependent on who’s seeded.

GSL

The GSL format is far superior to round robin because it doesn’t need tiebreakers and meaningless matches, because GSL is almost a double-elimination.

However, it’s not completely perfect. The previous shortcomings still exist, sometimes even more than round robin. A team only has to face three opponents, or two if the last game is against the same opponent in the previous round. Overall, they just need to outperform one opponent to get to the next round.

Recent improvements by ESL, when they used the GSL format for eight teams per group at IEM Katowice 2018, and some of their future events, have somewhat overcome the above difficulties. However, with the tournament using this method to find the three finalists (first place into the semi-finals, second and third into the quarterfinals), if four teams had to be selected, the final matches of each draw would just be team placement. Whichever bracket they won, even though they made it to the quarterfinals. Therefore, you still only have to beat two teams to get here.

Swiss

With the Swiss format, teams face three to five opponents and must beat three to advance to the playoffs. In addition, they can meet anyone in the group stage, depending on their performance.

At StarSeries, BO3 was introduced into the system for the first time. This minimizes the possibility of “surprises” (upsets), helping teams with large map pools, rather than teams that can only play three to four maps and escape with a BO1. But Swiss is not very convinced, because the matches are still randomly arranged.

As it turns out, this semi-automated system relies solely on teams’ scores to determine the next match, meaning the chances that the two best – or weakest – teams could meet sooner than planned, for example. such example. That causes the system to lose its effectiveness because a team can fall into a group where they shouldn’t be, thereby making subsequent matches increasingly misleading.

Take for example Cloud9 in Kiev, when they are ranked third in the world. After winning the first match, Cloud9 faced FaZe (#1 in the world at the time) in group 1-0 and lost, then lost to SK (#2) in group 1-1. Like fueling the fire, when they beat Virtus.pro in the fourth round, C9 had their hardest match in the fifth round, mousesports, and was eliminated. Cloud9 could have made it to the next round with a little luck.

Perfect competition

From the above formulas lead us to the new solution. The most obvious idea is to predetermine seeding teams. Teams will be kept like that throughout the group stage, but it’s not as simple as that. It will require teams to be seeded correctly and limit the chances of matches being randomly matched.

In addition, the perfect scenario is that without the “explosive” phase, the seeding team will lead to the 9 strongest teams entering the next round instead of the 8 best teams. Partly due to the main regulation in the Swiss format; rematches cannot take place, unless this cannot be avoided.

To solve this problem, either this rule must be dropped – this will take away the strength of the Swiss format before GSL – or have to re-seed the teams in the next round.

There is one way to use tournament results to re-rank teams based on the opponents they face: the Buchholz system, commonly used in tiebreakers. This system determines how the team’s run in the tournament has gone by summing up the opponent’s points, so it can be used as a seeding tool between rounds to prevent teams were placed in the harder (or easier) match they could have been placed in.

Since they are based on tournament results, this ranking is very objective. But with pre-determining the seeding teams before the tournament, it is not objective, because the results are based on a longer time period, so the system can skip the period of improvement (as well as loss of form). ) closest of teams. However, we still need to determine the seeding teams in the first round, as Buchholz can only use it from the third round onwards, when each team has met two opponents, giving more information for the draw. Seeds.

To demonstrate how this actually works, let’s take the next StarSeries as an example. In it, in the first two rounds, everything will be as it was at first. Then, from the third round, Buchholz applied and started re-seeding the teams since then:

Buchholz scoring method: previous opponent’s current total score (WL: Win-Loss)

As you can see, Buchholz not only adds your opponent’s points directly after your team has played them, the system continues to collect data on previous opponents to determine the ranks of the teams you will face. in the tournament. From there, the team knows some information about the next opponent.

You’ll find that, in rounds 3 and 4, this is where the seed pools are determined rather than the exact seeding of each team. Because at this point, the amount of information about the previous opponent is quite small, so there will be a random arrangement in the next round. In round 5, things became clearer, at this point the vast majority of teams had completed the group stage, so the amount of information from Buchholz also increased.

According to the Swiss formula, if you continue to win, you will meet a more difficult and difficult opponent, while losing, you will meet an easier opponent. The Buchholz system takes advantage of this, as it prevents the strongest (and weakest) teams in the same group from meeting each other. If you meet a difficult opponent (group of teams wins more than loses), the Buchholz system will place you against an opponent you have not met before.

The system can be further improved. The problem lies in the “randomness” that occurs in the second round. If unlucky, could create a worst case scenario or something like that:

This is what we want to avoid if we use the Bunchholz system from now on. That’s why we should use the seeding confirmation before the tournament even into the second round – or at least the seeds from each group of four. After that, Buchholz will start from round 3 and calculate based on the results played in the tournament.

When you get to the final round to determine the final ranking, with eight teams making it to the next round, other problems arise. At this time, the following Swiss draw draws will be semi-randomly drawn:

1. Teams 3-0 are drawn against teams 3-2
2. The remaining 3-2 team is drawn to meet a 3-1 team
3. The remaining 3-1 teams meet

This has caused some controversy in the past, most notably at the PGL Major Krarkow, when Astralis (3-1) faced SK (3-1) in the quarterfinals, while North (3-1) had easier match, Virtus.pro (3-2).

It could be better improved. And guess what – the purpose of the Buchholz system is to avoid arranging tie matches. By calculating each team’s Buchholz score, you can seed them 1-9, instead of three groups (1-2, 3-5, and 6-8). Since then, the above problem has been solved.

Here is the inner round of PGL Major Krakow if applying the Buchholz system to each team to find out exactly who is seeded, after the Swiss group stage is over:

Astralis will not face SK, but instead they will play North – a team that has had an easier journey compared to other teams with the same 3-1 record. Meanwhile, SK – the team that has faced the toughest opponents in the previous three rounds – will face Virtus.pro in the other bracket.

In conclusion, not only did Buchholz ensure that teams were evenly matched in the group stage – based on results played during the tournament rather than pre-event predictions – Buchholz helped seed the teams for the next round fair play, accomplishing what we want towards a perfect format. The only weakness is that the metrics behind the Buchholz system are quite complicated to follow, but the basic rules of the Swiss format remain the same.

What remains, we can hope, is that it will not only be discussed in theory, but will be applied in practice. It’s your turn, tournament organizers!

Source: HLTV