How long should tournaments be?

Tournaments and Events
1

TL;DR: How should the number of rounds in a tournament be chosen?

In the Excalibur tournament software thread, 5N00P1 suggested that Excalibur should include some guidance for tournament length:

My initial thought was that tournaments should be long enough to determine a winner conclusively and short enough to fit in the available time, and that within those two limits, more rounds is probably better, because more Netrunner is good. So long as there’s a lunch break, anyways.

So if I were to write guidance for how many rounds tournaments should have, it would probably take into account both the player count and the available time. And if it were intended to accompany Excalibur, it would probably recommend either single game or two game rounds based on that.

For instance, if you have four hours and four players, a round robin of two-game rounds makes sense; if you have the same time and six players, a round robin of one-game rounds might be better than three two-game rounds of Swiss; with twenty players, five one-game rounds of Swiss is likely the best choice. But if you have eight hours and six players, then the best option would probably be a round robin of two-game rounds.

I calculated the time needed for the tournament structures from the tournament regulations, since I figured that tournament length advice should take into account the official recommendations. I assumed that all rounds take five minutes longer than the allotted time. That’s probably too short for a big tournament, but in small tournaments rounds sometimes don’t even go to time, so I figure five minutes is probably a good middle ground.

Official tournament structures lengths

###Basic structure
Players Swiss rounds Cut size Swiss time Cut time Total time
4-8 3 3:30 3:30
9-24 4 4 4:40 3:20 8:00
25-32 4 8 4:40 4:50 9:30
33-64 5 8 5:50 4:50 10:40
65-96 6 8 7:00 4:50 11:50
96-128 6 16 7:00 6:20 13:20
129+ 7 16 8:10 6:20 14:30

###Advanced structure
Players Swiss rounds Cut size Swiss time Cut time Total time
9-20 4 4 4:40 3:20 8:00
21-32 4 8 4:40 4:50 9:30
33-56 5 8 5:50 4:50 10:40
56-80 6 8 7:00 4:50 11:50
81-128 7 8 8:10 4:50 13:00
129-176 7 16 8:10 6:20 14:30
177-272 8 16 9:20 6:20 15:40
273+ 9 16 10:30 6:20 16:50

(The spreadsheet is here if you want to check my math or make a copy and try different assumptions)

Looking at that data, I think many tournaments should probably have smaller cuts than the official basic structure. With 25-32 players, the cut takes longer than the Swiss. That may be fine for more competetive tournaments, but to me that seems counter to the goal of being “very accessible, especially for newer participants.” And it makes the overall tournament very long - a twenty-five player tournament under the basic structure is nine and a half hours, not including any breaks, which seems awfully long for a one day less-competitive tournament. But having a two-day tournament for a twenty-five player store championship seems silly.

2 Likes
2

For reference, I pulled up what FFG uses for their single-game games like AGOT.

Basic Structure: (For more casual events and Store Champs)
Number of Registered Players - Number of Swiss Rounds - Size of Cut
4-8 3 No Cut
9-16 4 No Cut
17-24 4 Top 4
25-40 5 Top 4
41-44 5 Top 8
45-76 6 Top 8
77-148 6 Top 16
149 and Above 7 Top 16

Advanced Structure: (For Regionals+, and for longer/challenging events)
Number of Registered Players Number of Swiss Rounds Size of Cut
9-12 4 Top 4
13-24 4 Top 8
25-40 5 Top 8
41-76 6 Top 8
77-148 6 Top 16
149-288 6 Top 32
289-512 7 Top 32
513 and Above 8 Top 32

3

Comparing to other games is a great idea!

Because I apparently have too much time on my hands, I put Netrunner, AGoT, and The Other Game in one spreadsheet for comparison, and then I made charts. It’s a LibreOffice spreadsheet, because I got fed up with fighting with the Google Sheets chart editors

Big charts

tournament_length_chart.png1319×833 9.46 KB
swiss_length_chart.png1323×405 6.24 KB
cut_length_chart.png1319×417 5.74 KB
tournament_cut_time_chart.png1316×806 13.3 KB

Interesting points:

  • Small Netrunner tournaments are in line with the other games for length, but large Netrunner tournaments are significantly longer.
  • All the games and tournament types spend at least 34% of their time in the cut, if they have a cut, which seems long to me but maybe that’s just because I often don’t make the cut :smiley:.
  • For large Netrunner tournaments, the basic tournament structure spends a larger percentage of its time in the cut than the advanced structure, because it has the same cut size but less Swiss.
4

For those charts, I should mention that Magic doesn’t have a time limit for finals, so I put in ninety minutes, which is the time limit for semi- and quarter-finals. I’ve never played Magic, but that seems like a rather generous time limit to me, given that Swiss rounds are fifty minutes. So I think it’s possible that the top cut in Magic tournaments is shorter in practice. If someone who has played Magic can provide feedback on that, I can add another line to each chart for typical Magic tournaments.

5

Related question: How should the cut be run?

Single-elimination cuts with two-game rounds and ninety minutes for the final would take less time than the current system. The catch, of course, would be splits.

Single-elimination with single-game rounds and the current time limits would take much less time - a sixteen player cut would take as long as a four player cut does with the current rules. It would be less fair, but in this old thread about the switch to double-elimination, several people claimed that any kind of top cut is less fair than straight Swiss, so maybe fairness isn’t the top priority for deciding how to run the cut.

The fundamental event document says, “Elimination rounds are intended to be a way to create exciting gameplay as part of the conclusion of an event.” Do they work?

The largest tournament I’ve played was a regional with around thirty players, so there was an eight player cut. I made the cut. I’m not sure I’d describe it as exciting. Tense, but not exciting. Basically everyone who didn’t make the cut, and who wasn’t carpooling with someone who made the cut, left after Swiss. So it doesn’t seem like it was that exciting for them either. Or at least, not exciting enough to be worth the time commitment. I feel like a shorter cut might have had a better chance at “creating exciting gameplay.”

But! There do seem to be a fair number of spectators for the final matches at Worlds, from what I’ve seen on the stream, so maybe the current system works ok for larger tournaments?

6

That’s because few people at Worlds can simply get into their car and drive home - most have to wait for their plane anyway, so they have nothing better to do.

7

Point.

What would people think of a finals only? Top two players, best of three games, sides chosen randomly for games one and three? That might be short and exciting enough to actually encourage spectating, and it’s equivalent to a very small double elimination cut, so it’s not really less fair.

8

For me less cut is always better. More people get to play games when Swiss lasts longer, smaller cut means there’s less incentive to ID, we get straight to finals. One thing I would do with such “Top 2” system would be to not award the spots to the top 2 players at the end of Swiss (as then, whenever possible, top 2 would ID final round into the final), but in let’s say 7 rounds event I would give one spot to leader after 6 rounds (and then drop them from the final round) and the other to the leader after final round.

3 Likes
9

Officially the finals are untimed. Realistically, they tend to be the second-fastest match in the day (Semi’s and Quarter’s are faster. In Finals, there’s more pressure so you see people taking their time more often.) I can’t recall the last time I saw a match go past an hour in length in the Finals of a MtG tournament.

(That’s for Best of 3, though. Larger tournaments like the Pro Tour will have Best of 5, which I don’t have experience with.)

Yes, the cut takes 33% of the total tournament time. For a significant fraction of competitors. The point of a cut, mostly, is to allow the majority of players to stop playing so they don’t have to use any further time when their chances of winning Overall are (very) slim. (‘Messy’ cuts where people with the same Win/Loss record may or may not make it in, based on other tiebreakers, are the exception to ‘slim chance of winning overall’.)

10

Ok. I’ve added an “MtG typical” line to the charts using 45 minutes for semi- and quarter-finals and 50 minutes for finals. It sounds like that may still be a bit on the long side, and it doesn’t account for tournaments with best of five, but hopefully it’s a bit closer to typical for comparison purposes. I didn’t redo the Swiss chart, since it’s unaffected by the finals lengths.

Charts!

cut_length_chart.png1319×417 6.03 KB
tournament_length_chart.png1319×833 15.3 KB
tournament_cut_time_chart.png1316×806 14.2 KB
LibreOffice spreadsheet is still here

With these numbers, Magic’s top cut length and total tournament length are a lot shorter than ours. That may be in part because the lines for Netrunner are still assuming that all finals rounds go to time, but I don’t think adjusting them to typical times would cut nearly as much off our estimated tournament lengths as it did off Magic’s.

Aside from discouraging IDs, are there other advantages to this? Also, how common would it be for the top two players to be able to ID in the last round? They’d need to be further ahead of players 3 and 4 than with larger cuts.

Also, how would you choose the number of rounds of Swiss in this system? For instance, let’s assume you have 32 players. If you do five rounds of Swiss, then you should have no more than one player who has won every round. At four rounds of Swiss, you might have two. (They would face each other in round five.) Would you do five rounds, and award the first finals spot at the end of round four, when it’s possible that a player who has swept every round would lose it via tiebreakers to another player who has done the same? Or would you have six rounds to avoid this, even though five would otherwise be enough even without a final?

Double elimination cuts don’t seem like a great way of doing that to me. If that’s the goal, it would make more sense to cut to more Swiss, or to encourage a culture where no one minds if people drop from a tournament when they’re out of the running.

11

Double Elimination cuts mean you can only drop one game.
Single means you can’t lose any games.

With Best of 1 matches, Double Elimination is sufficient to limit the effect of variance on the goal of ‘finding the best player’. Single Elimination wouldn’t reduce variance enough.

I’ve never actually thought of ‘Cut to Swiss’, but afaik, that’s the same as a Single Elimination. For example, if you cut to 8, you’d play 3 Swiss rounds. (One person is 3-0. He wins.) That’s the same as Single Elim…

12

If it’s two-game Swiss, then it’s the same as single elimination if no one splits, and more complicated if they do. But you could run four rounds of Swiss in less time than a top eight, which makes it a bit more complicated again.

On the other side of things: Why not run the entire tournament as double elimination? People who have no chance of winning would get to quit as early as possible, everyone would be able to drop one game, and it would still be shorter than the official tournament structures.

13

I’ll be honest, didn’t read the whole topic yet, but from my experience in local metas:

  1. People who show up at tournaments (not the ones that usually ID to get into the cut) like to play as many games of Netrunner as their time permits. You’d take this from the ones who usually end up at the bottom.
  2. Kind of builds on 1., but when new players ask me how to get better I tell them to come to any tournaments they can as then they’ll play a variety of opponents who try to play at their best. You’d also take this learning opportunity from people who usually end up at the bottom.
  3. I have the feeling that people of 1. will have less incentive to pay the entry fee and go to an event if they have a good chance of losing 2 games and being done in an hour.
3 Likes
14

I tend to agree with you. I asked more because I’m interested in where precisely the limits of removing people from the tournament to save them time are, if that’s the purpose of the cut. I’m not really convinced at this point that it is.

15

Thank @Absotively to answer my question, I think we lost a bit track to the tread start but anyway :wink:

There is a video from team covenant where they are talking about why people go to tournaments and that it usually take so long until you play against the players on your level, I think this is where Excalibur shines as it supports more players you play against. Also I would prefer more rounds of swiss and a smaller cut, but this is my personal preference.

On top of what @hsiale sayed top cut is to long to raise interest, this is different when the top cut is really small and you have a kind of finals. We had a tournament with a top 3 (3rd against 2nd, winner against 1st) where we had more then normal audience. So I can really recommend this. On top I like the idea that the first plays against the second and this is done by removing him from in the second last round and the last first one plays against him, no ID anymore. When doing another community tournament I will consider this, thanks for he suggestion.

16

As other have said, I think the ID-factor should not be overlooked when determining the number of rounds. It’s been really disappointing how common it is since they’ve changed the rules. It’s was probably better to have a few cheater get away with it before, than having half the cut ID the last 1-2 rounds into the cut. And now, with the drop off in attendance and the cut-heavy structure, it’s even more ridiculous, where a player that has won a bye, can sweep one round and ID their way into the Top 8.

It’s probably not a good idea to re-hash all the arguments about whether ID’s are good or bad, but the structure should definitely do everything it can to discourage it. A decent 32-person Regional should be 6 rounds and cut to Top 3 or 4. And there should probably be more incentive to play for the higher seed; maybe getting choice of side throughout the cut if there is an equal differential (instead of just choosing the first round).

17

Here are the regionals that are on Always Be Running, have valid match/points data, and had at least 25 players: 1 2 3

That’s a pretty small data set, so here are the store championships that meet the same criteria: 1 2 3

Looking at the final standings and last round pairings, it looks to me like there are few or no players who could have safely ID’d into a four-player cut in any of these, and no players who could have safely ID’d into a two-player cut. This seems to be true both for the tournaments as they were actually run and for if there had been an additional round. I haven’t gone through them very thoroughly, though, so maybe I missed one.

I think it’s possible that a smaller cut would discourage IDs a lot even without any other changes.

1 Like
18

IDs are good and healthy for the game. Those who dislike them are not interested in a tournament finding the best players. (I concede that some people are instead interested in playing in a tournament to play in a tournament. IDs are not good for those people.) There is a whole slew of discussion on this, and it boils down to being a mathematical fact as long as matches are two games, and tournament standings are directly tied to how many games you’ve won. Change to a one-game match, or a best-of-three match, and you’ll still see IDs, just less of them.

I can say a lot more, but really don’t want to bring up old arguments into this thread. You cannot make IDs illegal. If for some moral reason you’re opposed to IDs (there’s no logical reason to be opposed to them.) then the way to prevent them is to change the tournament structure. And even then, they won’t go away entirely, because they’re mathematically sound at most Swiss tournament brackets. (Single Elim, obviously, they don’t work with.)

(That’s the end of the actual post. Following is digression about IDs and Swiss math.)
So, the reason we have a Cut instead of straight Swiss is that it’s possible to drop a game but still win the tournament. A straight Swiss continues until you have an undisputed best player, which is usually when you either have only one X-0, or multiple X-1’s who have all played each other so that you have a direct ranking for them.

Double Elimination cut means you can drop two games and win the tournament.

Basically a lot of this gets screwed up because you’re supposed to only play against people who have an identical record as you going into the round. But, in Netrunner, each game counts for tournament points as if it were a full match in other games, and you play two games against an opponent before pairings are re-posted. This means that after one game, you have half the field as 1-0 and the other half as 0-1, in a sense of tournament points. BUT, we don’t repair! This means we have the entire tournament of 1-0’s playing 0-1’s, which breaks Swiss pairing rules. This is why it gets really weird for Netrunner, where after the first round, instead of having a known quantity of X-0 and X-1, the number is unknown. (0-4 X-0’s possible, and zero to eight X-1’s possible.)

So, the reason we have IDs is because they sort of restore Swiss logic to the pairings. If in the final round of Swiss, I’m playing against someone with the same record as me, and in the standings, there are X or less players within two game-wins away from my points going into the round, where X is the Cut number, I am guaranteed a spot. If there are X or less players within one game-win from my points going into the round, I’m highly favored for a spot; it depends on tiebreakers for those who are within a game-win from me. (If I ID and they sweep, we end at same game-wins.) The reason this restores Swiss logic is that you’re playing against someone in the same position as you, and at the end, you both have ‘played a game’ against someone with an identical record without changing the positions of the Swiss rankings.

I was lazy and only looked at Absotively’s third link. You can see from Swiss that 5 people have 18 points. We can’t see results or pairings, but we can guess that four of those people were paired for the final round of Swiss at 15 points. (I think Netrunner’s currently using 3 points for game win, anyway…) This means that those four could ID the last round of an 8-person cut. It also means that two of them could safely ID the last round of a 4-person cut. No one could ID the last round of a 2-person cut, but importantly: Only those four people could win the tournament. No one else can get the points and SoS required to make it into a 2-person cut. Heinzel specifically couldn’t have made a 2-person cut, yet he took Second in the tournament. This is why we don’t do a 2-person cut.

Heinzel and Mirilu swapped positions, and this is why we don’t do a 4-person cut.

A 16-person cut for a 27 player tournament doesn’t make sense. The only remaining one we can do is an 8-person cut. We cut out some 15-point players, ‘randomly’ (based on SoS, but SoS isn’t as strong an indicator as game-wins.), to give them a chance to win. Even though, in this case, we see that it didn’t make a difference, except #2 and #5 swapped places.

We use Swiss and then Double Elim cut to give the largest illusion of possibly winning the tournament for as long as possible to the players. Swiss also has the side benefit of allowing people who cannot possibly win the tournament to continue to play until the Cut, at least. IDs are a natural feature of this tournament structure because the Cut always comes before the number of Swiss rounds we would need to get an indisputable winner. (Normally it takes too many Swiss rounds to find this out, but for Netrunner it’s doubly true because of how a match is structured. There’s no guarantee we ever get a mathematically indisputable winner with Swiss pairings until it becomes Round Robin pairings.)

19

I guess some rehashing of old arguments about ID’s is fine since it directly contributes to the tournament length topic. @Absotively let us know if we’re derailing and I’ll refrain from posting more on the ID topic. I’d also appreciate if no one assumed why people dislike ID’s, without actually asking them.

I don’t think going to a 4 round Regional (~25 players) with a bye card and sweeping your second round opponent and ID’ing your last 2 rounds to get to the Top 8 cut is really that healthy for the game. (And, I’m not the only one, there was a lot of discussion from proponents for ID’s that thought it was ridiculous when it was first experienced in last year’s regional season.) As good as your math was in the example, I don’t think you could convince many people that players that played just 2 games and won them in the Swiss rounds has earned a spot in the top cut. Any of those 5 players at 18 points could be there with that strategy. I don’t think that’s a great method for finding the best players.

But, you’re right and I think for the most part we’re in agreement: it’s just a side-effect of the mathematics of the tournament structure. So, easiest way I can think of (and as you pointed out) is altering the tournament structure so that ID’s don’t make mathematical sense as often. As you can see from the post you responded to, that’s what I was recommending: an increase in number of Swiss rounds and a decrease in cut size.

20

They played two games, and won a tournament.

Don’t discount byes. (Well, maybe the random ones that happen because odd number of players… But not the byes from winning a Championship. :wink: )

For ANR and its odd match setup, you’d need X+1 Swiss to make it any better than it currently is. Uh, maybe I should define that. (Comes from mtg tournament organizer background…) Basically X Swiss means the minimum required to get a single X-0 at the exact player count. So for 8 players, X Swiss is 3 rounds. 9-16 players is 4 rounds. X+1 Swiss is sometimes used in lieu of a Cut for MtG. (So 8 players would play 4 rounds of Swiss, with potentially a playoff if the winner wasn’t indisputable… More than one person with the same record and hadn’t played each other.)

The main problem with X+1 Swiss is the time required. That and some people aren’t convinced it’s necessary (for mtg, anyway. Netrunner has a different match structure so it may fit better.)

We’re currently doing either X or X-1 Swiss with Double Elim, compared to MtG X Swiss + Single Elim. This fits Netrunner’s format better. I do think that X+1 or X Swiss with Double Elim would probably be better, from a ‘get the best player in the tournament’ perspective, but it would also add an hour, minimum, to tournament times, which are already kinda long.