The latest Clarity blog post regarding the post season break mentioned format development as one of the areas that we’d look to spend time on ahead of the next season. To that end, I’d like to speak broadly on the various aspects that need consideration in crafting the format, and how any given option or decision impacts the player experience. My main reason for doing so is my feeling that most players, when analyzing particularly the drawbacks of a certain format, do so in a vacuum without considering what the alternatives are — or, more specifically, what other issues those alternatives may bring. At various points throughout this post I’ll also touch on more general rulings when and as they relate to the nature of a certain format option. The goal here is to foster a better general understanding of these various considerations, and to ultimately land at a format that minimizes the impact that its cons have on the player experience. I emphasize the idea that the goal is to minimize the negatives — there is no perfect solution that everyone is consistently happy with, and instead it’s a matter of getting as close to that as possible.
The Playday Question
I’ll preface this section by stating that Clarity will, for the foreseeable future, default to a single playday format. Exceptions will be made only in scenarios similar to season 6, where dual playday allows for a season to be played in full without lengthy breaks due to holidays or the like.
That being said, I’d still like to go over my thoughts on either option.
The main argument for single playday is also the one that ends debate on the topic — a greater number of players can participate in an event played once per week. No matter which second day you pick, dual playday will inevitably disallow some from participating, and the inverse is very rarely true for single playday. The only scenario where someone who would play a dual playday event can’t instead play a single playday event is when they know that they cannot commit to a longer season due to their schedule. Ultimately, this is much rarer than someone being unable or unwilling to commit a second day to amateur Dota.
On the other hand, dual playday has a number of advantages — none of which matter more than maximizing player count, but are nevertheless notable. For starters, many formats necessitate breaks or bye days, and these tend to be more harmful for teams when it means they won’t get to play an official for two (sometimes more) weeks, compared to (for example) not playing on Tuesday but then playing on Thursday. This idea of teams having momentum that gets disrupted by lengthy breaks will come up multiple times throughout this post.
It should be noted at this point that many of these considerations are ultimately completely subjective. I will do my best to focus on describing the more objective aspects, but I am, at the end of the day, also a player in all of these events, so my own biases will creep in here and there. In this case, many players will have individual preferences for either single or dual playday setups, simply because they enjoy one over another for their own reasons.
One other boon of dual playday events is that, put simply, you can run more of them. Many players become soured on the idea of participating in leagues simply because they have a negative experience with a team, and they’ll feel stuck in that situation. We have provisions for allowing disbands, but this is something people (understandably) rarely pursue, be it because not everyone is on the same page, they simply want to play it out and get it over with, or whichever other reason. Shorter seasons mean less time spent enduring a negative experience, while the opposite sentiment — ie, players being upset that the season is shorter because they’re enjoying their team — is far less common. In that respect, dual playday can be better for retention of your playerbase — you’re just working with a smaller one to start with.
Division Size
Once again, it’s worth stating at the start that the size of a Clarity division will remain at the usual 8 teams, with the smaller division size based on player count being 6 teams.
The obvious note here is that divisions with an odd number of teams are completely off limits — giving one team a bye every playday is incredibly detrimental, and there are no advantages that even come close to making up for this.
At that point we can look towards even numbered options. 4 is too small; 6 is suboptimal but does a good enough job of emulating the 8 team division, particularly with the season 6 system of eliminating the bottom 2 in an 8 team div (with 6 team divs allowing the choice between moving all teams into the playoff or opting for a 4 team playoff, eliminating the bottom 2).
Larger divisions, on the other hand, can start to defeat the purpose of a division based format. Without considerable expansion (in the sense of attaining a much larger — and much more consistent — playerbase), larger divs mean bigger MMR spreads within teams, and there generally doesn’t seem to be much incentive to try a 10, 12 or 16 team division to make up for that. The other issue is that the group stage into playoff bracket concept becomes a bit unwieldy between 8 and 16 teams: with 10 teams, you can only have one big group (as 2x5 leads to byes), which by default necessitates an extremely long group stage for round robin. Beyond that point, multiple groups do become viable — 12 teams can become 2x6 with a straightforward elimination of bottom 2 in each group to allow for an easy 8 team playoff bracket, and 16 is just a doubled 8 team div. In any case, none of these seem to have much value.
The 8 team div hits the middle ground of having decently tight MMR spreads, and giving staff a choice on what the season ought to look like. If we want short groups, two groups of 4 give us a 3 playday group, if we want longer groups, we have a 7 playday group. This is another incredibly important part of the format, one that necessitates a more in depth look.
Group Setups
As 8 teams has been established as the default division size, all further sections will work off of that assumption, with potential comments regarding the alternative 6 team div setup.
The only real options then are running one group of 8 teams, or two of 4. This is an example of a choice where both options have significant advantages and disadvantages, and as such often come down to preference. I’ll try my best to present arguments for both. We’ve used both in Clarity, so many players have had the chance to compare their experiences with either.
There are very many things that this decision impacts: the length of the group stage, the importance of group stage performance, how many teams someone will be guaranteed to play against, the series type during the group stage, the minimum guaranteed duration of the season for a given player.
We started with a dual group setup, running two groups of 4 teams. In Clarity, we ran one set of round robins, with each team playing the other 3 in their group once. A notable option with this setup is trying two sets, ie all teams playing each other twice, resulting in 6 playdays for the group stage, rather than 3. At that point though, you’d rather simply run a 7 playday round robin with one group, so with that in mind, the only real option with this choice is a 3 playday group stage.
In practice, this leads to groups being a sort of preparatory warmup for the playoffs, which then become the the main part of the season. The final standings of such a short group stage will tend to be less representative of how the teams actually stack up against each other, and can often mean that teams simply don’t have time to mesh and get going —particularly with single elimination playoffs, but even with the partial double elimination that is usually paired with a shorter group stage. For context, when we did multiple groups in prior seasons, the bottom two teams moved into the lower bracket of the playoffs, rather than having everyone who makes the playoffs start in the upper bracket. This then meant teams who had a rough start could be eliminated only 4 playdays into the season.
With all that being said, this setup has its advantages. None of the games you play in a group stage this short are meaningless, as can be the case with the final series’ of a 7 playday group, with a team being guaranteed eliminated before playing their last series or two. It dawns on me here that I might slip up and use the term week in place of playday or vice versa — which is the same in single playday, but very different in double.
The single group setup gives you one option for the group stage — 7 playdays with each team facing each other once. By far the coolest thing this allows for is that a team will get a chance to face every other team, and there’s a lot of value to that — both in the sense of giving people a better picture of how everyone stacks up, but also in terms of community building, as you’re guaranteed to interact with the rest of your division, standins and forfeits notwithstanding. This also arguably makes a team’s group success (or conversely failure) more representative and thus potentially meaningful, largely because consistent good performances imply a team is actually overall strong in the division, and they haven’t simply stumbled into an easier group. That also brings up another point: when your playoff seeding is based on your performance against all the teams, rather than just your group with 3 other teams, there’s a lower probability of a scenario where a team has a worse final placement because their group is tougher. This means that 1x8 generally allows for the placement of each team to be a better show of their level relative to the rest of the division, whereas 2x4 might skew the accuracy of the overall placements — though neither of these have much bearing on the final winner of the season.
This decision is also tied quite closely to the topic of playdays — the issues with one setup are often compounded by the choice of playday setup. For example, the matter of the 6th or 7th series in 1x8 being potentially meaningless for some teams becomes significantly more problematic with single playday, because it essentially doubles the time they spend playing out a pointless season. On the other hand, dual playday means that a 2x4 group stage flies by incredibly fast, and thus inherently rewards teams working well out of the box without really allowing time for them to work on issues (though whether or not this is an actual negative for the league and player experience is debatable).
A number of other considerations hinge partially on the setup as well. We switched to BO2 groups when playing 1x8, compared to BO3s in 2x4 — largely because there was significant community demand for BO2 groups, which are hard to justify in 2x4, because 3 sets of BO2s lead to essentially guaranteed tiebreakers, while 7 still allow some leeway while reducing the time commitment required of players for the majority of the season. Naturally, this setup still ended up having ties most of the time, so there is strong incentive to play 1x8 in BO3s as well to reduce (but unfortunately not eliminate) the risk of having to play tiebreakers, since they’re effectively a break for most teams, and breaks of any kind are generally very detrimental.
On a less tangible level, this decision also sets the tone for what is the more “important" part of an event. Shorter groups act as more of a warmup for the playoffs, versus a lengthier group stage which serves as an important interlude to the final act of the event
Series Length
Single vs Double Elimination
What does the “perfect” format look like?
Having considered all of the aspects of a format — and the implications and impact of different options —we can begin to think about what might be closest to ideal. The emphasis here is on getting as close to perfect as possible, because an actually perfect format doesn’t exist. To do this, it’s most prudent to consider which major issues are most detrimental to player experience.
Byes and breaks are by far the most complained about occurrence. How do we eliminate them? Let’s review where they come from in Clarity.
Double elimination playoff brackets become almost instantly unviable. To illustrate, let’s consider what one looks like — this will be done through reviewing the bracket we’ve used previously, which isn’t a full double elimination bracket (because some teams here start in the lower bracket, rather than all teams starting in the upper bracket).
Assuming all 8 teams proceed to the playoff, 4 in UB (seeds 1-4), 4 in LB (seeds 5–8), all teams play day 1. Day 2, the 2 in UB can play their match up (or it can be scheduled to a later playday) and the remaining 4 play their LB matchups. Now, in day 3, the 2 winners in LB need to play to determine who faces the team who dropped from UB during day 2. Thus, break day for 2 of 4 remaining teams. Day 4 sees the two remaining LB teams duking it out for a path to the finals. 1/3 teams has a break, their second in a row. Assuming tiebreakers were played, one team has had 3 potential breaks in 5 playdays. We can cut this number down by turning LB games into BO1s, condensing 2 playdays into one. A break still happens for the UB team. This still happens for other numbers of teams proceeding to playoffs (ie, 4 or 6 proceeding into double elim).
Okay, so, single elim. Top 4 makes for a very simple setup — 2 series and you’ve got your champions. Not very exciting though. 6 means top 2 wait for opponents on day 1, which is again bad.
All 8 it is then. 3 playdays, everyone has their chance, all gas no breaks.
Except — whoops — a top placement in groups becomes meaningless. You lose the guaranteed top 4 placement for being a top 2 seed which exists in 8>6. How do we make it up to these teams? Should we do so in the first place? The only real option is to give the higher seed choice of pick/side all games in their first round series…but that’s a very significant advantage to give the 4 seed over the 5 seed, and not significant enough for a 1 or 2 seed.
