In Search of the Best Tournament Format
With the Hearthstone scene currently buzzing from excitement over the Blizzcon qualifiers held in the past weeks and other high-profile competitions, I thought it’s high time to take a deep and analytic look at the different tournament formats. What would the ideal one be for the game? And – dare I ask – how presentable each of them are for the Twitch audience?
If you’re reading this article, I assume you also agree with me that Hearthstone is, in fact, a game of skill – and while it involves an element of randomness, it is not a coinflip or a crapshoot. If you don’t agree with this sentiment, let me quickly ask you something: how come the same players regularly get to Legend rank? It should not be possible if the game is entirely based on luck. (Also, why are you not among them?)
The right amount of randomness
When it comes to competitive tournaments, I’m sure you will all agree with me that the main goal is to minimize the effects of randomness so that the skill shines through in the end. Of course, Hearthstone is a card game so you cannot – and probably should not – fully eliminate this. On the other hand, if your game is skill-based (which is possible even with a random element), you want to make sure that mere prayers to RNGsus will not win you tournaments.
The way to do this? Minimize variance – increase sample size. I’m not going to give you a lecture on statistics here, but here’s a quick example: if you’ve got an 80% chance to win, you’ll lose once every five times. If you can make sure that you have that chance ten game in a row, the odds of you losing five of them in a row is 0,032% - less than once every thousand times. The more skillful player will therefore benefit from having a longer string of games so that his or her abilities can shine through over the random elements. Essentially, being good at a game like Hearthstone (or poker or MTG, etc.) means that you can maximize your odds of winning. That, of course, does not mean that you will win every single game – in fact, maintaining a steady 70+% winrate in Legend would most likely get you to the top 1 spot.
This concept of increased sample size doesn’t only apply to the tournaments but also to the individual matches. A “best of seven” format favors the more skillful player more than the “best of three” format. Needless to say, the stronger player also benefits if the tournament organizers ditch the single elimination format.
Single elimination – Dreamhack, Tavern Takeover
Ah, good old single elimination. Knock-out tournaments are probably the most exciting to watch, as the stakes are constantly high. On the other hand, it increases the odds that the stronger player just has a bad series of draws or a few bad matchups (more on decks later). The increased element of randomness makes it less likely overall that the best player will win your tournament.
In this sense, Dreamhack was dreadful. 128 players in a single elimination bracket with best of 3 matches until the final, where you get a best of 5? It’s a joke. Even Tavern Takeover had the courtesy to have best of 5 games and a best of 7 in the final. Thing is, single elimination is only one of the problems: only being able to use a deck once also limits your abilities as a player. This and the ban system turns the matches into what is essentially a chain of best of 1 games as you will always try to pick hard counters to your opponents’ decks. On the other hand, this is the simplest system to manage as an organizer.
Swiss versus round-robin
Round-robin is where all players play all other entrants, sometimes even more than once. This is essentially the fairest system available, as the elements of randomness and the impact of a few freak results are minimized. On the other hand, well, it takes forever.
Consider this: if you have 128 players, like Dreamhack did, it would take 128x127/2=8128 games. Over eight thousand matches! That is obviously not feasible for a tournament like this, no matter how fair it is. Imagine if the players were to play each other twice as well! Impossible to organize, of course.
The system, however, does a great job in different scenarios. For instance, football leagues are organized in this manner (a double round-robin format), where each teams play each other twice – once home, once away. You can make a clear comparison between the respect for the winner of the leagues and the winners of the cups – the former ones are generally considered to be the stronger sides.
You could, for instance, take the 48 strongest Hearthstone players and create a league with weekly rounds, each participants playing each other once in a best of seven match. This is probably as close as it would get to a fair but still broadcastable tournament.
The issues with the round-robin format are partially fixed by the Swiss system. It takes the same amount of rounds to complete a tournament in this format as if it were a knock-out tournament, however, it is more fair. It is also a lot more complicated to organize. A short excerpt from scichess.org should give you a nice idea of it:
“In the Swiss system, after the first round, players are placed in groups according to their score (winners in the 1 group, those who drew go in the 1/2 group, and losers go in the 0 group). So each round, you play someone with the same score as you. Since the number of perfect scores is cut in half each round, it doesn't take long until there is only one player remaining with a perfect score. The actual number of rounds needed to handle the number of players in the section is 2n, where n = the number rounds. […] In actual practice, there are usually many draws, so more players can be handled (a 5 round event can usually determine a clear winner for a section of at least 40 players, possibly more).”
The early winners’ matches still maintain the same excitement as those of a knock-out game but mistakes or bad luck at the beginning of the tournament will not make it impossible for you to make a comeback and win. If you’d like to read more about these formats and the differences between them, here’s a link to the cited article. It’s also worth noting that the competitive MTG tournaments also favor this format for the first part of their competitions.
Healthy mixtures – Prismata Cup, VGVN
My personal favorite of the bigger tournaments of the scene has to be the Prismata Cup. For one, the first stage is not a knockout scenario but a Swiss system. Five rounds, all games are best of five. Also, which is my favorite element, you can pick any deck for each game a maximum of two times – which makes the selection before each game a bit more interesting than just automatically going for the hard counter; it also allows you to have a second try with the same deck in the same match. It is probably the most involved and the most skillful yet broadcastable setup I’ve seen to date. The second day followed with a single elimination stage with best of seven games where you could use each deck a maximum of three times – that, as you can probably tell, was a bit less to my liking.
VGVN went with what I like to describe as the “FIFA World Cup” format: a group stage followed by a knockout stage. A short round-robin stage is followed by a knockout part. It is, again, a decent “middle of the road” solution: less random than the fully knockout formats but also more broadcastable than a full round-robin tournament.
Double elimination – The Road to BlizzCon
What’s the road to BlizzCon like, you may ask? I've got something unexpected to show you.
One thing you may not be aware of is that the qualifiers broadcasted on Twitch are merely the second phase of a much larger event. If you read the full tournament rules, you’ll be in for a surprise.
“Phase 1 of the tournament will consist of “The Constructed Arena Phase.”
Players will be matched against each other in a Swiss format based on their initial seeding.
If a player loses three matches (not games) in Phase 1, that player will be eliminated from the tournament.
If a player ends a round of the Swiss format tournament with a +5 match record, that player will advance to Phase 2. (Match records are determined as Wins minus Losses. 5-0, 6-1, or 7-2 records are all considered +5 match records.)”
Yeah, this is essentially the same format as what we’ve seen in the Prismata Cup, coupled with the fact that the second phase also has double elimination. As annoyed as some of the viewers may have been by Reynad’s handlock not drawing any giants in a crucial game, for example, should be aware of the fact that it is still probably the most fair system that we’ve seen in a tournament to date – it’s just a shame that we’ve only gotten to see its less fair part.
As I stated above, I’m not a fan of knock-out formats, be they single or double elimination. Granted, the latter holds some elements of the Swiss format but losing a mere two matches in a field of 16 should not mean that you are guaranteed to lose out on a top 4 spot.
We now know that the EU and US qualifiers are using the same format. I’d like to give you more information on the other ones, but… yeah.
There’s definitely room for improvement when it comes to tournament formats. The scene currently seems to favor the flashier knock-out methods, which are detrimental to the professional scene. However, recent competitions seem to have moved away from it slightly, though not as much as I would have personally liked.
There are two games you can and should compare Hearthstone’s situation to: chess and (no limit hold’em) poker. Chess tournaments usually implement the double round-robin format with a small amount of participants: poker tournaments feature so many individual hands (where you can decide whether to wager your tournament life or not) that the skillful players are bound to shine through. I think that the former would fit the game better and would also be easier to implement. Here’s hoping the scene will continue its shift towards more skill-oriented formats! It’s the quality of the games, not the fake drama of the organization is what we’re watching the tournaments for.