Improving tournament formats
What does HLTV.org's Tomi "lurppis" Kovanen think about tournament formats in modern Counter-Strike? Check out a new proposed group stage format and changes in other stages of the tournament below.
The idea for writing this article came to me after a friend told me of a tournament format for group stage that is apparently used in some StarCraft2 events. I have always felt like group stages often provide very random results for various reasons; whether it's because top teams aren't fully motivated to give it their all knowing their tournament isn't on the line as they can still make up for a loss, or because of threeway ties that have ended tournaments of countless teams over the years due to terrible tie breakers that rely on rounds.
Regardless, in my opinion group stages were never a good or necessarily even a very fair way to decide the advancing teams, but unfortunately, they were, and still are, a necessity. While on the subject, I also went over my thoughts for other stages of a tournament and touched up on map selection and picking sides.
First of all, the score in a match shouldn't matter – you should be motivated to win, and as long as you do, I do not think the score should make any difference at all. It's a cool statistic for people to look at after all, like any other statistic, but I think that's all it should be.
Second of all, I think group stages need to be simplified. Often fans are confused regarding round differences and various different tiebreakers when trying to figure out if their favorite team will advance with a 16-8 win or if the other match tying would mean their team no longer has a chance to advance.
What? We needed two more rounds to advance?
I was a vocal member in the conversation and believe I kickstarted the process of ESL changing their rules from having to play out all 30 rounds in group stages (which always ended up wasting time because the winning team was no longer trying), and although this is a bigger change, it also makes sense from all perspectives – I would argue it's better not only for the player, but also for the spectators and the event organizers.
I think the best way to setup groups is still with different seeding pools (high-medium-low-unranked), as this limits the possibility of manipulating groups by tournament organizers, and also helps later on with the re-seeding process of playoffs I will explain. It also takes some pressure off event organizers, as it is significantly easier to place all teams in a correct seeding pool than seeding them one by one, especially towards the bottom.
Below is an example of how the new group stage system works:
Notably no match can end in a tie – overtimes will be played out until a winner is found. This way each team has a chance of advancing with two wins, or can be knocked out by two losses. The system is very simple and clear to follow and there is no confusion. You can no longer win your last match, and still go out; you have full control of your destiny and each match will always be exciting, although one could argue Match #1 will be quickly over in most groups with weaker teams, every other match should be very interesting.
From an event organizer's point of view this should be a no-brainer assuming players agree on it – it removes the possibility of threeway ties, all problems regarding tiebreaker rules, the need (if there ever was any) to play out all 30 rounds and with that some time, and one match, also saving time. At IEM 6 World Championship, Carmac told me that each match in the same round had to start on the same time so teams could not lose on purpose or know just how many rounds they needed to win to advance or mess up some other team's chances by manipulating the rounds. This format will get rid of all those problems. The only negative is that it would be easy to set yourself up to finish second due to a lack of threeway ties à la the thrown match by mousesports against Lions at DreamHack Winter 2011, but if you set the bracket play up correctly, that won't be a problem either.
Speaking from a player’s point of view – this would be a great upgrade for the current situation where you never really know where you truly stand due to threeway ties, round differences and the possibility of ties. I think the fact that your last match will determine your fate in the tournament is the perfect way to put it, since it will always mean that one is the most meaningful. Also, you will never be able to drop out after two wins anymore – there can no more be 2-1-0 ties among three teams. Not having to play meaningless matches after being eliminated or advancing that likely can influence the other team's chances in the tournament is also a very welcome change, as sometimes a team can get a free win over a good team simply because they are already out and have no motivation to try to win. The winner of the group will also get another reward as they get to rest slightly longer and only have to play two matches and are able to save their strategies and setups for more later, potentially more difficult, matchups.
An entire group stage filled with potential threeway ties coming down to round difference
The final way to look at this would be from a spectator's point of view. I think the current group system can be very confusing especially with tournaments choosing to use different tiebreaker rules, and this system obviously would fix that issue entirely. Spectators and players alike should not have to keep a mental tally in their head of every match going on in order to figure out who will be advancing from each group. With the systems we have now, people are constantly questioning how the tie breaker system will work. Due to the fact this group stage system can't have ties, it would actually make groups seem like playoff matches as well, adding more intensity and importance to them, which obviously bodes well for the viewer. A spectator can at any given point of time look at the standings, and know exactly what is at stake in every single match. Every round will count and every match will matter.
Finally, it takes out some of the importance over which side to start a map on because round difference doesn't matter anymore, so winning 16-14 with a nice comeback in the second half is not only just as good as a quick 16-1 overrun, but it can obviously benefit the team due to boosted morale. With this system, higher seeds recieve the benefit of playing their weakest competition first. As it stands now, there is no system of organizing who will be playing who in the group and inevitably you end up with matches that will mean nothing to the advancing teams. This new system would remove that waste of time.
Tournaments also have to stop giving teams preferential treatment regarding schedules and actually enforce the normal way of placing teams in groups – the number one seed has to be in group A, number two in group B and so on. Currently it's either completely random or organizers are able to give teams preferential treatment.
While the current way to pick maps is generally fine and I don't really feel the need to necessarily improve it, I do think there is a better alternative out there, and it was in fact used by ESL's first Extreme Masters seasons. Here is how it worked:
Lower seed removes the first map, higher second, and they alternate until there is just one map left for a best-of-one, such as a group stage match. For best-of-threes, they alternate on removing maps until there are three maps left, at which point each team picks their own map. The thing about this system is, that the last map remaining, is actually played first, and the loser's map always follows after.
Map selection waiting to happen at e-Stars Seoul 2011
This adds elements of strategy for teams who have a very strong map and don't want to risk a series going to 3 maps. Currently map picks are often not very interesting because most teams have figured out what maps their opponents remove or play most of the time, so this would definitely add more strategy to it.
I believe a knife round is the best way to pick sides as it still does come down to having a skill, even if not used during the regular match, which is better than the pure luck of a coin toss. I think we could figure out certain standard rules for knife rounds such as are you allowed to use heights to your advantage or not (which in my opinion you shouldn't, it should be on even ground to keep it fair and speed up the process). Regardless, knife rounds are a far superior method of choosing sides over either team getting to pick, which benefits teams whose strong maps are heavily one sided.
One major problem certain tournaments have caused over the years, such as DreamHack Winter 2011, is announcing how the bracket will play after group stage, which can lead to losing matches on purpose a´la mousesports vs Lions. That should never under any circumstances be announced before.
Now, there are a couple of different ways to go at it, you can either do it by seeding based on the original seeds and pre-determined positions in the bracket, but if that is the case they should never be announced before hand.
I also think randomly drawing brackets is a completely acceptable way of doing it, as it completely eliminates the chance of teams trying to manipulat the brackets, but it can also lead to mTw-Na`Vi type scenarios where one team simply doesn't catch a break and always ends up playing the team that they had had difficulty with early on in single elimination.
So to avoid Na`Vi we should actually lose in group stage?
Lastly, I think re-seeding teams before the bracket stage would be a good idea. This would have to be based on finishing first or second in the group (as in, if there were four groups all winners would be seeded 1-4, second places 5-8), but you would not look at the round difference to determine it. This is where the traditional seeding with rankings would take place, and you would go back to recent history to determine the seedings. With less teams now, it's even easier now to seed correctly.
As far as double elimination goes; it's a CLEARLY superior format, however, unless there is time to do it in best-of-three, I don't think it should be used. I think playoffs in any Counter-Strike tournament should always be played in a best-of-three format, as best-of-one can simply be way too random.
Something interesting the NHL does in its Stanley Cup playoffs is re-seeding after every round. Normally this makes a bigger difference in brackets that include more teams, but it would help even with just eight teams.
The main advantage of re-seeding is that a low seed can not steal a top seeded team's path in the bracket by a single upset win, and it would therefore award teams with long standing history of good placings and solid play. As it stands now you can take one glance at a bracket and unless there are major upsets, know how it will play out in the end based on the upcoming matchups. Re-seeding adds a little variance into play, and makes it more interesting.
After teams are assigned seedings in the beginning of bracket play, you will adjust the matchups after every round of play based on the seeds left in play. For example:
As you can see above, the highest seed will always face the lowest seed available, therefore maintaining their advantage they've earned by obtaining the highest seed.
I hope players fans and tournament admins alike agree with me that this format provides the most viewing entertainment for fans, the best playing experience for players, and the least amount of annoyance for the admins. It allows for the fairest and most straightforward format of tournament play.
I have contacted the admins I have kept in touch with including ESL and urged them to change formats, but I can't do it alone, so if you agree this format works out best for all involved, talk to your local event organizers, point them to this article and ask them to consider changing.
Everyone knew DTS was done the second this bracket was published
What do you think of the changes proposed in this article? Do you agree with the group stage, map picks, side picks and the bracket play? Which ones would you add in your tournament? Leave a comment below.