Lessons Learned from First and Second Edition Tournament Formats
If you stare into the abyss for too long, it might just stare back
Introduction
Just before AMG announced the 2.5 transition, Tim Horsburgh and I were working on a document for tournament organizers to use. During the FFG/AMG handover (or lack thereof), there were effectively no resources for anyone trying to run events. To make matters worse, the more time passed, the more organizers that had this knowledge drifted away into other hobbies. Just as we were approaching being ready to release all of this, AMG announced massive changes to the game, effectively making all of this much less pressing and relevant, though I hope it’s still interesting to some.
The following is the original writeup, with some minor edition modifications to hopefully make things more clear in the current 2.5 framework.
Graduated Cuts
The Progression Cuts defined in FFG’s event document are structured by players participating in Swiss rounds until a predetermined round count is met, ending in a cut to a specific number of players (Top 8/16). Over time, large official events adopted a modification of this format. Instead of taking the top predetermined count of players from standings, all players that meet a specific threshold proceed instead. For example, in a 7 round event, this might be that all players with 5 points (game wins) from Swiss proceed to single elimination. This will usually not lead to a clean single elimination bracket, so to alleviate this byes are assigned to each player starting from the top until the bracket for the first round is assigned.
For example, in a tournament where 6 players proceed to single elimination, the top 2 players would receive byes in the top 8.
In tournaments with multiple days of swiss qualifiers to a single progression cut, players are seeded alternatingly from best performance to worst, starting from larger qualifier. For example, if two qualifiers lead to 7 players from Qualifier A, and 5 players from Qualifier B, the bracket would be seeded starting with Seed 1 from A as the highest seed, then Seed 1 from B as second highest, Seed 2 from A as the third, proceeding until Seed 5 from Qualifier B, with Seeds 6 and 7 from Qualifier A being the final and lowest seeds, with byes filling the final space.
Tournament Structure
5-8 Players: 3 Swiss Rounds, No Cut
9-16 Players: 4 Swiss Rounds, No Cut
17-20 Players: 5 Swiss Rounds, 4-1 Progresses, Max 2 Cut Rounds
21-44 Players: 5 Swiss Rounds, 4-1 Progresses, Max 3 Cut Rounds
45-84 Players: 5 Swiss Rounds, 4-1 Progresses, Max 4 Cut Rounds
For each Round 1 bye for an event, one player should be added to the Number of Players when using this guide.
Editorial Note: Some of these numbers are off by 1 or so in extreme cases of pair down/pair up wins. The last remaining task to do on this document was get these exactly right.
For events with multiple days of Swiss Qualifiers, the Size of Cut should be adjusted accordingly. For example, for an event with two days of Qualifiers using this table, if there are 80 players in each day of Qualifiers, instead of a Top 16, this will need to be doubled to a top 32.
Tournament Tiebreakers
Traditionally, Margin of Victory has been used as a tiebreaker for X-Wing events. The goal of using Graduated Cuts is to eliminate Tiebreakers entirely, as there are not enough rounds played for them to be meaningful. The Tournament Organizer still needs a method to order players who qualify at the same score for bracket seeding. Because of this, a modified Strength of Schedule is used for seeding. To prevent Strength of Schedule from becoming warped by drops, each player’s lowest scoring opponent is removed for the calculation.
Round Time
All rounds (Swiss, Cut, Final Game) are played with a 75 minute time limit. AMG has announced that the X-Wing rules will be updated to include a round timer for 9, 12, or 15 rounds, but it is unclear how this will be implemented. If you would like to use a round count to prepare for this, it is recommended to increase the time limit to 90 minutes to support 12 game rounds.
Editorial Note: Now we know the time limit is about 75 minutes, with a 12 round limit that effectively no one ever reaches. When this was written, it was in a world of no scenarios!
Game Tiebreaker
In the event that a game ends with both players having destroyed the same number of points, the winner is randomly selected by a Judge or the Tournament Organizer via die roll, coin flip, or other random agreed upon method.
Events Larger than 84 Players
Once an event has reached 85 players, there are trade-offs to consider based on the resources available to the Tournament Organizer. Time and space permitting, the Elongated format can be used to allow for more games to reach more accurate tournament results. In most scenarios, the Standard format should be used. Generally, the Elongated format adds an additional Swiss and Single Elimination round, and results in a cut to roughly the top 30% of players instead of the top 20%. In situations where players would prefer to play more rounds, this may be preferable, but take note that 6 rounds of Swiss generally takes the course of an entire day at these scales, so it is common for things like restaurants to be closed by the time this event completes each day.
Editorial Note: I’ve removed these numbers and suggestions, as generally we wanted to discourage 6+ Swiss Rounds in a single day. Rather than risk someone misuse them thinking it is the better format, I’ve opted to remove them.
Further Detail and Explanation
These decisions and formats come from lessons learned from different tiebreakers and formats that have been tried over the history of X-Wing. Below, the high level concepts and justifications are listed, though every detail is not included. A recurring theme to these concepts is that players tend to not have strong opinions about tournament formats unless it feels as though they have been unfairly impacted. Because of this, it is difficult to obtain objective feedback on tournament formats.
A Brief History
Initially, the majority of X-Wing tournaments used Strength of Schedule (SOS) as a tiebreaker, to a predetermined cut size, such as a Top 8 after 5 Swiss rounds. This led to a lot of players making or missing the progression cut based entirely on their SOS. It is understandable that SOS would be the default for events, as it is the most fair tiebreaker for formats in which players do not play all other players, seeing widespread use in competitive settings using Swiss pairings. While Strength of Schedule is appealing for events with few rounds, a player’s score could be heavily influenced by players who have dropped from the event. The most common complaint when using SOS is that players have no control over their Strength of Schedule, which feels unfair to them.
This led to the creation of Margin of Victory (MOV) as a tiebreaker. Instead of SOS measuring the difficulty of each opponent a player faced, the more a player won a game by, the more they were rewarded in tiebreakers. Additionally, instead of a simple win/loss system, a more complex scoring system was created. Players received 5 points for a full win, 3 points for a modified win (a win by less than a threshold used to determine a close game), 1 point for a tie, and 0 points for a loss. There were multiple issues created by this pair of changes, the details for problems with MOV will be detailed later. The 5/3/1 points for wins and draws lead to issues relatively quickly as players realized the number of points awarded per game was not constant, and a full win resulted in more net points being rewarded for the game. For example, a relatively common occurrence for two players both on the cusp of the progression cut playing the last round of Swiss was players realizing that one participant could no longer win the game and therefore could not make the progression cut, but they could make the game into a modified win for the opponent instead of a full win, which would also prevent their opponent from making the cut. This situation led to upsetting situations and left some players feeling wronged. Additionally, draws were possible in this format. For whatever reason, portions of the X-Wing community have a strong aversion to intentional draws, which was compounded with unfortunate recommendations for Swiss rounds based on player counts that led to occurrences of enough players taking intentional draws such that the progression cut was determined before the last round of Swiss was played.
In response to issues around draws, scoring moved to a binary win and loss, 1 and 0 point scoring system, in which there are no draws. Instead, if the score is tied when time is called, the Final Salvo rules in the current Tournament Regulations are used, in which players roll a number of dice equal to their combined ships attack values. While this removed the possibility of draws, it had the unintended consequence of letting players maximize the game tiebreaker in the list building step, which has created a problem of players not engaging. In potentially the most egregious example, at one of the World Championships, two players didn’t even set up ships as both players realized they were uncomfortable leaving their fortress, as the final salvo was tied.
Most recently, in response to these issues, the sections on Fortressing and Stalling have been added. Fortressing has been explicitly defined as having all of your ships not move in any way for 3 turns, which unfortunately is relatively easy to sidestep without meaningfully changing anything by having ships move in circles, etc. Additionally, there is language in the Stalling section that attempting to exploit a stalemate is considered unsporting, but it is quite vague, and there is widespread disagreement about how that should be enforced, or if it should be enforced at all.
In addition, when scoring was changed to 1 point for a win and 0 for a loss, the largest official events gradually migrated to Graduated Cuts. Initially, this involved dropping all players who did not have at least 2 wins after round 3, with just the remainder playing the rest of the tournament, but over time it became apparent that there were sufficient players happy to play games for the fun of them, even if they were eliminated from making the progression cut. As a result of this, the largest tournaments were run as described in the Graduated Cut section as listed above, but with Margin of Victory as a tiebreaker instead of Strength of Schedule.
Explanations of decisions
Graduated Cuts over Standard Cuts
The high level goals that lead to these player count and round counts are again the result of lessons learned over the years of tournament organization in X-Wing. Due to the length of time required for individual games, players dislike playing 7 rounds of Swiss in a single day during multiple day events. It was very common for events to start early in the morning, and finish up just before midnight, and while a short lunch break was generally provided for players, generally by the time all rounds of Swiss had concluded, it was near midnight, and the surrounding area around the venue had already closed. Additionally, the Progression Cuts provided by the given ranges result in roughly 20% of players from Swiss making the Progression Cut. There is a large range of skill of players at X-Wing events, making Swiss best thought of as a Qualifier for the most competitive players. For larger events, especially using the Elongated format, more than 20% of players from Swiss will make the Progression Cut, which is by design. Generally, this format is expected to be used in cases for very large events where there are more high end or competitive players in general at an event.
The main advantage of Graduated Cuts over the more standard Top X cuts used previously is to remove the need to use tiebreakers to determine which players make a Progression Cut. Explanations of the flaws of currently used tiebreakers will be listed below, but the benefits of Graduated Cuts are difficult to overstate. From a Tournament Organizer’s perspective, all that a Graduated Cut does to scheduling and round counts is move one round of Swiss to one round of Single Elimination. In most cases fewer games are required. For example, under the previous Advanced Structure in the FFG Tournament regulations, an 80 player event would be, assuming no round 1 byes:
“Advanced Structure” Standard Cut
6 Rounds of Swiss
Cut to Top 16
2 players at 6-0
7 players at 5-1
18 players at 4-2 - half of which make the cut, based on tiebreaker
4 Rounds of Single Elimination required for Top 16 Cut
10 Total Rounds Played
Graduated Cut
5 Rounds of Swiss
All players with at least 4 wins progress to single elimination
3 players at 5-0
12 players at 4-1
Number 1 Swiss seed receives a bye to fill out Top 16 Cut
4 Rounds of Single Elimination required for Top 16 Cut
9 Total Rounds Played
Notably, this player count happened to nearly fill a square bracket for the Graduated Cut. With slightly more players, for example, 88 total, we see Graduated Cuts end up with an extra Progression Cut round. Despite this, the total net number of rounds played remains the same in the Standard Format, as one fewer round of Swiss was played.
Standard Graduated Cut
5 Rounds of Swiss
All players with at least 4 wins progress to single elimination
3 players at 5-0
14 players at 4-1
Seeds 1-15 receive byes through to the top 16. Seeds 16 and 17 play the only game in the Top 32.
5 Rounds of Single Elimination required for Top 32 Cut
10 Total Rounds Played
Elongated Graduated Cut
6 Rounds of Swiss
All players with at least 4 wins progress to single elimination
2 players at 6-0
8 players at 5-1
20 players at 4-2
Seeds 1 and 2 receive byes to the top 16, while all other players play their games in the Top 32.
11 Total Rounds Played
The Elongated Progression Cut results in roughly the same or one more round than was played in the previous Advanced format, and is mostly intended for extremely large events with hundreds of players, where previous 7 rounds of Swiss play would be required. However, we have provided it at lower player counts to allow for events that have the space and time required to use if they prefer.
Tiebreakers, Benefits and Drawbacks
Unfortunately, few Swiss rounds can be played in X-Wing in a day due to the required time for each round. This leads to tiebreakers that do not provide enough information to determine Progression Cuts with. This is the original problem that Graduated Cuts solve, but even in a Graduated Cut, some method for ordering players with the same score is required. A quick overview of tiebreakers used in the past, along with their drawbacks and benefits is provided.
Strength of Schedule
In other spaces, this is generally referred to as Opponent’s Win Percentage, or the Buchholz system. This is the gold standard tiebreaker for nearly every competitive activity using Swiss. This awards tiebreakers to players who had the most successful opponents. Unfortunately, with how few rounds of Swiss can be played for X-Wing events, combined with low scoring players dropping, many players have been frustrated missing progression cuts based on this tiebreaker. As stated earlier, players also feel frustrated that they have no ability to control their Strength of Schedule. This led to a tiebreaker that players could exert some control over, in the form of Margin of Victory.
Margin of Victory
Margin of Victory also tends to go by many names in different spaces. This tends to be common in formats where there is no variance in players, for example in a sports league, or round robin event, in which each player plays every other player. While this is not the case in X-Wing, Margin of Victory makes sense in these cases, as each participant is playing each other. The margin by which each participant won each game can be directly compared. In short, instead of using information gathered by the tournament to determine tiebreakers in the form of Strength of Schedule, this attempts to use information gathered from individual matches in the form of how close each game was to use as a tiebreaker. The immediate upside is that players feel this is more fair, however the downsides are numerous and have noticeable effects on how the game is played. These fall into two main categories, where the effects can be noted in both the tournament and in individual games.
Changed player incentive in the tournament
Over time, players have realized that it is equally or more important to maximize the Margin of Victory scored over a tournament than it is to win individual games. The first and most obvious problem is that players are rewarded for beating opponents by large margins. Margin of Victory values a veteran player getting a perfect win against a novice player in their first competitive game. Giving players control over the tiebreaker in the form of Margin of Victory also leads to issues in general during Swiss, as players strategize over the tournament itself. When two friends are paired and the winner of the game would be on the cusp of making the progression cut, there is a large incentive on the players to have one immediately concede to the other, nearly guaranteeing making the cut, or lie about the final score that occurred. Unfortunately, players misreporting scores to benefit tiebreakers is common.
Changed player incentive in individual matches
In addition to the tournament level issues caused by Margin of Victory, there are warping effects on the manner in which players play individual games. There is nothing inherently wrong with that, and in fact a good tiebreaker generally incentivizes players to behave in a certain manner. Unfortunately, these effects do not seem to be intended. The largest effect on games is that the goal of each individual game has changed. Players are not attempting to win or lose the game, but instead try to win or lose by large or small margins respectively. For example, a player would generally prefer to win 60% of their games, but score significantly better Margin of Victory in those games, than they would win 65% of their games but with a poor Margin of Victory. It does not seem to be that players are consciously making these decisions, but instead as lists that score Margin of Victory well are more likely to proceed to Single Elimination, they tend to do better overall, and players copy those lists. Also, as when a player concedes, all of their ships are destroyed, Margin of Victory tends to force players who are in games that cannot be won to spend the rest of the game attempting to retain as much Margin of Victory as possible, which is a very unpleasant process for both players. This tends to lead to accusations of slow play, or unsporting conduct. While it was more common in First Edition than currently, there are also board states that are possible where neither player can damage the other, and the outcome of the game is determined, but the losing player cannot concede and instead both players are forced to play a game with a known outcome.
Why Strength of Schedule over Margin of Victory in Graduated Cuts?
Despite all the listed downsides of Margin of Victory, there tend to be less complaints of using Margin of Victory as a tiebreaker for the previous Standard Cuts to Top X players. This is likely because players felt the agency they gained was worth the drawbacks which are more subtle. Thankfully, in a Graduated Cut where tiebreakers are no longer used to determine which players proceed to Single Elimination, none of the downsides of Strength of Schedule apply any more. The only thing this tiebreaker is being used for is to order players, for which the result might be that some players at a given score have a bye in the first round of Single Elimination. Strength of Schedule is the most fair option available for this, as the players in a given score bracket with the highest Strength of Schedule have already demonstrated that they have played and won against the other players in the Graduated Cut, where players with lower Strength of Schedule will have tended to play weaker opponents by definition. Additionally, while it sounds intuitive, the results of Margin of Victory tend to reward the opposite players with higher standing than the tournament ought to prefer. Using data from all recorded Second Edition tournaments, which is roughly 2,000 final standings for 5 round events with players with 3 wins, and 3,000 for 6 round events with 4 wins, the relationship between Margin of Victory and Strength of Schedule tends to be a negative one. As players’ Strength of Schedule increases, their Margin of Victory decreases. This is not claiming a causal relationship, that if a player has a high Margin of Victory, they must have a low Strength of Schedule. This is just an observation that if we were to judge how well Margin of Victory is selecting players to have higher standing, it appears to be having the opposite effect as intended. Generally, Margin of Victory tends to favor those players with easier Swiss pairings.
Round Time
Originally, the finals of Single Elimination tournaments used to be untimed. When board states in which neither player could destroy the other became more common in First Edition, this was changed to a two hour time limit. However, the time limit is a large determination in what players play, general strategy, and what makes certain components good and bad. Forcing the players in the finals to play what it effectively a completely different game at a 120 minute time limit instead of 75 minutes does not make sense, as a large portion of games played are going to time, and players are not trying to destroy all of their opponents’ ships, they are attempting to destroy more points than they have destroyed when time is called.
Game Tiebreaker
Much like Margin of Victory affecting player incentive and behavior during both the tournament and individual games, Final Salvo causes similar problems during games. While the root problem of removing draws as a possibility seems to have been an improvement for player engagement and happiness, Final Salvo has caused serious problems as well. Generally, as players have understood the tiebreaker better, they have found better ways to build lists around it to win games in ways that are presumably unintended and are very uninteractive. For example, this seems to have started when players started building lists with a high number of attack dice in a list that was extremely efficient at trading shots, and “fortress” in the corner such that none of their ships moved without risk of any opposing list being able to outmaneuver them. Their opponent was either forced to go to the enemies board edge, and lose by not being able to trade shots as effectively, or accept the final salvo tiebreaker. As the final salvo odds are deceptively harsh, for example a 12 to 8 final salvo roll favors the 12 dice roller 85% of the time, this became gradually more common. Additionally, this also allowed lists which were extremely good at maneuvering to refuse to engage in games where they happened to have more salvo dice than their opponent by just running away the entire game. To combat these, players and organizers have suggested and attempted many tiebreakers to alleviate the issue. A brief overview of the most common is provided.
Draws
Definitely the most simple, just allowing draws, and likely intentional draws as well removes the root problem of needing a tiebreaker for individual games. Given the extreme reaction the player base had to intentional draws, it is probably best to avoid returning to this. Unfortunately, the intentional draws were exacerbated by tournaments with round and cut sizes that led to it being somewhat common for the last round of Swiss to be mostly irrelevant, and while this problem could be solved, it is unlikely players will be willing to try it again.
Final Salvo
As described above, it tends to heavily warp how the game is played as it changes the goal for players by allowing for winning the game without ever engaging the opponent. Despite this, players seem to prefer the uncertainty in who will win the game. As the victory for the list with more dice is not guaranteed, they seem more willing to accept that tiebreaker loss than if it were guaranteed.
Tiebreakers Tied to Bidding
The most common is the suggestion that instead of points that are left unspent in a list being unscored until an opponent’s full list is destroyed, they are considered destroyed immediately by an opponent. While this sounds like it may alleviate the problem, this is a slightly modified Final Salvo, and much like Final Salvo incentivizes players to spend their points on the best jousting list they can, and sit in the corner and refuse to engage. At worst, it is a coinflip to win the game, and if their opponent has any bid at all, they automatically win the game. Some other variations on this are that the player with the smaller bid wins the tiebreaker, which has the same effect. Other common suggestions include either the First or Second player winning by default, which again create player incentive problems, just in opposite directions. When First player wins the game by default, players are able to bring large bids without ever intending to engage or damage their opponent and take first player, for example, taking two phantoms which they leave cloaked for the duration of the game and attempt to take no damage with. When Second player wins the game by default, again players are able to bring large bids without intending to engage, but instead likely in the form of high initiative ships that simply run and dodge arcs until the game ends with no damage dealt.
Random Winner
Having a player randomly win the game removes the list building incentives around tiebreakers, but it still forces players to consider tiebreakers when setting up engagements. There are two timings for Random tiebreakers, the winner of the tiebreaker can be randomly determined at the start of the game so both players are aware who will win the game in the case of a draw, or after the draw has occurred, and each leads to different player behavior. When players know who will win the tiebreaker at the start of the game, players will definitely engage, as one player knows if they do not engage they will automatically lose the game. This gives an advantage to the random winner, which they can listbuild around. If a player brings a strong jousting list and automatically wins half their games when they win the tiebreaker by fortressing, the net result may be more fortressing in general. Players may also choose not to engage when the winner of the draw is randomly determined after the game, as both players may feel uncomfortable about their odds of winning an engagement should they commit to going to the other player's edge of the board.
Why Random Winner, Determined After a Draw Occurs?
The main goal of the tiebreaker is to minimize the effect of the tiebreaker on each individual game, in terms of what lists players play, in addition to what their behavior is in games. Unfortunately, Random Winner determined after a draw occurs, still has issues around player engagement incentives, but seems to be the least bad set of outcomes. We are not game designers, and the lack of incentive to engage is a game design problem. For example, even in this system, neither playing being willing to engage the other is still possible, consider Alice and Bob who have both brought a list and set up fortresses on their side of the board. Alice determines she has a 65% chance of winning the game if Bob approaches her fortress, and a 30% chance of winning the game if she approaches Bob’s fortress. Bob determines he has a 70% chance of winning the game if Alice approaches his fortress, and a 40% chance of winning the game if he approaches Alice’s fortress. Neither player agrees on the relative odds of winning or losing the game based on their decisions, but both have a better chance of winning the game by fortressing, than they do by engaging their opponent.
Credits
Thanks to:
Tim Horsburgh, for doing a lot of math and writing probably half of this
Alex Raubach, for keeping and maintaining ListFortress, which makes any kind of analysis on data possible
Zach Bart, for helping with data analysis and visualization
Lots of others for proofreading! I’ll let you keep your anonymity.