Which tournament format best emphasizes winning (as opposed to avoiding last place)?

Question

I've played enough Settlers that I think I could do well in a tournament. However, a cursory glance at upcoming tournaments suggests that most Settlers tournaments use a Swiss scoring system that distorts the incentives of the game significantly. To take one specific example:

http://www.boardgamers.org/yearbkex/setpge.htm

Using this format, it is theoretically possible in three rounds to have:

• Player A wins 1 game 10-6-6-6 for 5+4=9 tournament points, and 2 second place finishes for 3 each, or 15 total.
• Player B wins 2 games 10-9-7-7 for 5+1=6 tournament points each, and last place in the other game for 1 point, or 13 total.
• Player A gets more points than B because of one strong victory and 2 second place finishes.

I have always played Settlers to win, not to avoid coming in last place. Playing with this format, I would emphasize avoiding last place over winning. I might also attempt to cooperatively engineer a certain number of victory points per player on the third round if I had 1 win and 1 second place finish coming into the game and some of the other players were in similar situations.

I would personally rather play a tournament where the emphasis is on winning as opposed to avoiding last place finishes and possibly brokering third game deals. Do such tournament formats exist? If so, what is the name of the tournament format(s) that best emphasize(s) winning (as opposed to avoiding last place)?

Answer 1

The tournament format that most emphasizes winning is single-elimination. I mean, literally: if you don't win, you're out.

The problem with single-elimination is that it the order of elimination doesn't necessary tell you something about a player's overall achievement. Assuming the tournament winner is legitimately the "best" player, we really can't say who is "second best" based on tournament standing alone. If the winner's first-round opponent was actually the most difficult match, maybe he or she is really the "second best" player, despite getting out immediately. So you'll know who's first but everyone else's place in the standings is kind of a crapshoot.

The single-elimination format is also rather frustrating for high-variance games (I would say Settlers is one), and being knocked out early can be very dissatisfying for tournament participants.

---

Most Swiss systems do emphasize winning, in the sense that they're literally counting wins as the main scoring system. One pitfall in Straight-Swiss is that it's possible for a player to functionally win the tournament before getting to the last game -- this is similar to what happens when one player scores 3-0 in a "best of five" match. A popular way to avert this is a "Top N," where top-scoring players from the Swiss rounds come together for a single-elimination bracket.

Top N has the benefit of making sure the top players are actually pitted against each other, and the single-elimination part of the format does promote a heated final where everyone is playing to win. However, this actually encourages what you've called "brokering third game deals" in the last rounds of the Swiss -- if I can get into the Top 16 with a third-place finish in my last Swiss game, for instance, there is no reason to play for the top.

As best as I can tell, the Swiss system will always produce some irrelevant matches and predetermined outcomes. The best you can do is to tweak where those occur in the tournament structure. Top N is a compromise that keeps those features out of the finals, at the cost of filling the final round before finals with a series of tricky forfeits and draws.

It's my opinion that it's better to accept "brokered" draws (with a prohibition on bribery and collusion to keep the game fair) than to try to force people to play irrelevant games. What you can do to discourage intentional draws and forfeits, however, is to associate an incentive with your overall ranking going into the Top N. In current Magic: the Gathering tournament rules, for example, Top 8 games give the higher-ranked player the ability to choose his or her position in the turn order (instead of deciding randomly, as they do in the preceding Swiss rounds).

---

I think the 5-3-2-1 structure you've cited is, in general, a good example of a Swiss-style format that emphasizes winning over second-place finishes and promotes a reasonable level of competition in the final Swiss round before Top N.

For starters, first is disproportionately more valuable than the other results: a first-place finish and a last-place finish are worth as much as two second-place finishes; and finishing first, second, and third is worth more than finishing second three games in a row.

Note also the role of wins in tie-breakers:

1) Top overall score [of three rounds of Swiss] (15 being max)

2) Most 1st place wins

3) Top winning margin (Ex: a game ends 10-7-7-6, thus the winner has a +3 margin; All margins are summed)

4) Win in game 1

5) Win in game 2

6) Win in game 3

7) Highest dice roll -- 2D6

Here the margin of victory is used only as a tiebreaker, as an attempt to factor in the decisiveness of wins if players are tied for quantity of wins.

Finally, this bit is important:

Playoff rounds will be seated by rank (see below) but the game format will be the same

...

Tournament Playoff Round:

Once the top 16 players are set, seating will be done according to the following chart:

Table 1 - #1,5,9,13

Table 2 - #2,6,10,14

Table 3 - #3,7,11,15

Table 4 - #4,8,12,16

So, there is some incentive to finishing the Swiss rounds with the highest rank possible: you can get a leg up in the turn order (I'm assuming the tournament organizers consider this an advantage in Settlers).

That said, because I don't consider going first to be always preferable in Settlers, I'd do something like this instead: "After the board is laid out, players at the final tables get to choose their seats in rank order." A little bit of extra privilege (and strategic depth) for placing first!

Answer 2

This is a pretty universal tournament problem. I've seen the same situations in Warhammer 40K tournaments as well. Basically the problem occurs when you have less than the mathematically required number of rounds for the amount of players, and reward players according to degree of victory.

In a 1-on-1 winner take all tournament you need X rounds where 2^x = number of players. With games that have more than 2 competing players, you'll need even more rounds to determine the true winner. With games that can take 2-3 hours to play, it is not feasible to set aside the 15+ hours required to handle just a 64 man tournament. Most tournaments can only feasibly support 3 or 4 rounds.

The theoretical goal of a tournament is to have only one player at the end who won all of their games; the "best" player. In normal tournament situations you won't be able to find this theoretical best player for more than 8 to 16 people. So most tournaments try a few different compromises to make a tournament work in these constraints.

Swiss format is the most common form. In these types of situations you apply a score to the results of a game. At the end of the day, the player with the higher score wins. There are drawbacks to this approach. One, when you reward degree of victory, you reward players that play easier opponents, and penalize players that have close games. So even if you won all your games you could still lose the tournament because you weren't brutal enough in your victories. I've even seen people lose tournaments to people they beat.

In the 40K tournament community we call this the "clubbing baby seals" effect. It can sometimes allow a mediocre player who has been lucky to be paired with far weaker opponents to outscore a better player who was unlucky to be paired with his peers and only manages marginal victories. You can read this blog post I made illustrating the effect in better detail.

You probably want to know some alternatives to get a better result out of your tournaments. Here are some solutions that the 40K community has tried:

• Limit the number of players to the number of rounds you can hold. This means that you will only have 1 undefeated player at the end of the tournament. Only count wins and losses and don't bother with marginal victory scoring.
• Allow players with equal records to share the victory. If at the end of the tournament you have 3 3-0 players, they all share an equal prize. This can be a bit unsatisfying because you don't have 1 winner, but you don't have anyone lose out due to arbitrary scoring.
• Accelerated or Opposite Seed pairing. These approach use seeding to vary opponents to try to overcome the inherit disadvantages of the Swiss Format. They can theoretically reduce the amount of rounds you need to get a clear winner. Accelerated can be hard to pull of if you don't have preexisting player rankings. Opposite Seed pairing is a slight alteration where you just pair high degree of victory winners against low degree of victory players for the first few rounds. The idea here is to test for the seal clubbing effect.

Answer 3

In researching a scoring system for our own tournaments I modified the latest F1 scoring system (2015).

https://en.wikipedia.org/wiki/List_of_Formula_One_World_Championship_points_scoring_systems#Points_scoring_systems

For 4 player games take the 25, 18, 15, 12... and subtract 12 from each so you get:

13, 6, 3, 0

For the OP's example it would be:

A: 13+6+6= 25

B: 13+13+0= 26

B wins.

This scoring system also works well for other 5 and 6 player games (just subtract so last place always gets 0 points).

Answer 4

As someone who has played in Settlers tournaments hopefully I can shed some light on how the tournaments are run but can not comment on Swiss vs other tournament system. The first round of a sanctioned Settlers tournament is Swiss as you described with Wins ranking first, then total points, then a percentage of strength of win used to break any ties. After the 3 games are played, all players are then ranked and the top 16 go on to a single elimination semi-final/final setup. At PAXEast, they run two qualifying rounds and you can play in both of them to try to generate the highest possible score over the weekend. i.e. If I get zero wins and 27 points which is not likely to get me in the finals I can play on Saturday and get 2 wins which will most likely guarantee a spot on Sunday.

Ranking in the top 16 does not determine seating position in the semi-finals. Players are seated to a board based on ranking but then have to roll for position. Even at a very large event like PAXEast, Mayfair has a REALLY hard time finding enough people to play. Usually a single win will get you in the semi-finals. But once you are in the semis, everything game from there to world championships is win or die.

Sanctioned qualifiers this year could send you to the World Catan Championships so your tournament host should be very strict and check if a player is eligible for your specific qualifier. PAXEast was a sanctioned qualifier which meant Canadians were not eligible to play (They have their own qualifiers north of the border). Some players had dual citizenship US/Canada and thought it was ok they could play in the qualifier throwing all the numbers off. One had reached the semi-finals only to be told they had to leave because they shouldn't have played in the first place. Many of us found this very annoying because people who finished in top places may not have done so if the mix was correct.

One thing I will say is that tournaments are very fun and I would highly encourage anyone who plays Settlers to give it a try. You will meet great people and have a great time.

Answer 5

Origins, run by Mayfair, prior to 2007 used to do 3 games and it was 1) wins 2) total points 3) "Quality of Points," or your points as a fraction of the total points. After 2007, you played 4 games, randomly against opponents.

The top 16 would play in a 1-game playoff with the winner of each playing in the final.

The 4 games qualifying match was really grueling. I think most people qualified for the playoffs with 3 wins. There were occasionally 2 wins 37 points or 2 wins 36 points, but that was usually right on the edge.

This does cause some gaming of the system (for instance, some people would just deliberately take longest road and the 2 points to raise their score, knowing they would lose; some people would drop out midday after losing their first two or three matches; two friends, if one had been knocked out of contention, might be tempted to throw the game the guy still in contention).

Let me throw a wrench into the works. The top seed usually has a bull's eye on his/her forehead. Psychologically, people think that the person to beat is the guy with 4 wins or 3 wins/39 points or whatever. Say a person won the first three matches; should he or she still strive to win the fourth match or maybe stop around 5-6 points? That's part of the metagaming :)