The Quizbowl Resource Center

Mods- I’m not sure whether this should be here or in Theory, but since there’s only one tournament affected, I put it here. Please move if necessary.

A quick summary of the problem: We currently have a format in which there are 12 preliminary pools of 8 teams to start. These sort into four playoff tiers. It is possible to get sorted into an exceptionally strong prelim pool, go 6-1, lose a half-packet tiebreaker, and end up being able to finish no higher than 25th.

I think we can all agree that losing at least 1.5 games in a seeded prelim pool means that a team isn’t a legitimate championship contender and can have no complaints about not being able to play for the title. But, I think the difference between the 24th-finishing team and the 25th is minimal. If I had to guess, I’d actually expect that if the teams finishing 23-26 played another mini-pool post-tournament, the teams finishing 25-26 would generally end up with better records than teams 23-24. This is a proposal to allow for more movement in the standings after the prelim pools, while not significantly increasing staffing or packet requirements.

I got the idea from European soccer leagues which have “relegation playoffs” between some of the worst teams in the upper division and some of the best teams in the lower division. Let’s say that we sort teams into 24-team groups that link together one 6-team playoff bracket from each tier- so for instance, Algeria, Egypt, Iran, and Mongolia from this year. There are two possible ways of running this. The second method allows for far more movement than the first. For convenience, let’s say that teams who end up in the top 24 are “promoted” and teams who end up in second tier are “relegated”.

Method #1: Only teams that finish 5-0 in their second-tier playoff bracket- therefore, clearly better than the other teams they played that afternoon- get to play for promotion. They get to challenge the team(s) with the worst record in the top tier.

Here’s how method #1 would work for movement between the top-tier Algeria and the second-tier Egypt, assuming that a team finishes 5-0 in Egypt. There are three possibilities: A single 0-5 team in Algeria, a pair of 1-4 teams in Algeria, or three 1-4 or 2-3 teams in Algeria.

With 1 Algerian team, there’s a straight 1-game playoff. Winner goes to 17th-24th superplayoff, loser goes to 25th-28th superplayoff.
With 2 Algerian teams, the Algerians play each other on a half-packet. Winner is safe. Loser plays the Egyptian team. Winner goes to 17th-24th superplayoff, loser goes to 25th-28th.
With 3 Algerian teams, they get placed into a 4-team single-elimination bracket in which pairs of teams play a half-packet, then the winners and the losers play another half-packet. The teams from the top tier do not have any advantage over the challenger. The team that lost twice gets relegated; the team that won twice goes into the 9th-16th superplayoff. If the double winner happens to be an overachieving Egyptian team, they go into the superplayoff counted as having lost once to the 3rd placed team in Algeria.

Also implementing this process for Egypt/Iran and Iran/Mongolia would allow for movement between the other tiers as well, but would require more moderators.

Method #2: All of the teams with the worst record in Algeria are at risk of relegation- this can be one team at 0-5, or up to three teams tied at 1-4 or 2-3. The teams with the best record in Egypt- anything from one team at 5-0 to three teams at 4-1 or 3-2- get to play for promotion. The problem with this method is that it will require an additional packet and potentially two playoff rounds instead of one on Saturday evening.

The number of slots open for promotion is equal to the number of “worst” teams in the upper bracket- so it’s entirely possible that we could have up to 3 Algerians playing 3 Egyptians. The way I see this working is that if there are even numbers of Algerians and Egyptians, they pair off for a full playoff game. If there is an imbalance, then if it’s 1 and 2 or 2 and 1, they play the typical half-packet ladder we’re used to. If there’s 2 and 3 or 3 and 2, we have one pair and one ladder. If it’s 3 and 1 or 1 and 3, we do single-elimination on half-packets to identify the winner or the loser.

After the promoted and relegated teams are decided, we would potentially need one more ladder packet to determine which of three promoted teams go to the 9th-16th superplayoff and where each of three relegated teams end up for Sunday.

This method would allow for potentially a lot of movement across tiers. In theory, literally half of Algeria could fall into the 25th-36th range and half of Egypt move up into the 13th-24th range. Done over all of Algeria/Egypt/Iran/Mongolia, it’s possible that every team in Egypt gets either promoted or relegated and none end up finishing 25th-48th!

I think allowing at least the limited movement of the first method would improve the “fairness” of the system by letting a dominant second-tier team challenge up without being overly difficult to implement. Thoughts?

What seems more fair would be to use advantaged-final-style playoffs for these cross-tier matches. Teams in the top tier should still be afforded some advantage by virtue of doing better in the preliminary rounds.

This seems like an interesting proposal, and--while I want to be clear, I am not promising anything nor am I speaking on behalf of PACE as a whole--I think there could be something here worth using.

Ndg wrote:What seems more fair would be to use advantaged-final-style playoffs for these cross-tier matches. Teams in the top tier should still be afforded some advantage by virtue of doing better in the preliminary rounds.

I think that teams in the top tier were already afforded an advantage- they had the advantages of still being alive for the title as well as five afternoon games to prove their worth and avoid getting into such a situation.

The top tier teams may have only done better in the preliminary rounds because they (unluckily/unfairly) had a much easier draw, so allowing for relegation playoffs is just a way to even up the inherent luck involved in prelim bracketing. Barrington ended up 3rd in Hydra, but I am quite confident that in ten games, they would have beaten TJ B (who were in first tier) at least six times. I don't know the other teams in the 17th-24th range nearly as well, but it's hard for me to see Barrington doing worse than .500 against them.

Fair enough. The other part of this that makes me uneasy, though, is that in both proposals, teams on one side of the divide don't have full control over whether they get a chance to change tiers. In the first one, Algeria could have an extremely "weak" (relatively) team that loses all its games by 450 points but doesn't have to win a relegation match simply because of how Egypt shook out. What makes this even more problematic is the not-technically-totally-arbitrary-but-pretty-darn-close way in which the groups need to be paired off between tiers one and two. We might very well encounter a situation where there are two 0-5 tier-one teams who obviously "deserve" to be in tier 2, and two 5-0 tier-two teams who obviously "deserve" to be in tier one, but no one gets to play relegation matches because those teams' groups don't line up. You bet there'll still be complaints from those tier-2 teams.

Vice versa for the second format.

Ultimately, there are a lot of short ends of sticks to go around in a tournament of this number of teams and rounds, and I don't see this doing much more than distributing them in a more complicated manner.

If I were planning to use this proposal and had full power to do whatever, here's how I would do it:

(thinking out loud mode on)
There are exactly four promotion/relegation matches, each between the bottom team from a Tier I group and the top team from a Tier II group. There is no need to earn the right to playoff, and no case where a tie at the top or bottom means multiple teams are involved.

(method 1)
These matches would occur during the Tiebreaker B period, due to packet constraints. The only time the bottom team from a Tier I group is playing in TB B is when there's a three-or-more way tie for the bottom spot. In the case of a 3-way tie at 1-4, you'd have the teams seeded D, E, F on PPG as usual, with E plays F, winner plays D to get into the top 16. Now, you'd have the loser of E/F playing the pro/rel game during that second half. Four-way ties are rare, and would require more work.

(method 2)
There is one additional packet written, specifically for pro/rel, after Tiebreaker B. Tiebreaker B now serves to split Tier I groups into 1-2 / 3-4 / 5 / 6, and also to crown a winner of each Tier II group when teams are tied.

In terms of matching the teams up, it'd have to be either pre-ordained or calculated really quickly on PPB basis. And while the latter is obviously preferred, that gets into the heart of the issue: entering stats is not quite real-time, especially at a tournament as large as NSC. Saturday already runs a long time. It might not run quite as long chronologically as HSNCT, but without the byes, teams are playing 12 games with just the lunch break, and possibly the tiebreakers. Teams might not want to sit around for 30-45 minutes after Round 12 waiting for another game. Plus that does mean renting a handful of rooms for another hour.
(thinking out loud mode off)

One of the things that PACE, and our tournament director and logistics crew, will have to balance is whether the utility of having these pro/rel games will offset the additional moving parts of getting them organized (and potentially the expense and work of an additional packet were one to be written and I want to emphasize that neither of those things is trivial!). Suggestions like this (and people should feel free to make more!) will certainly be taken into consideration, but "small" changes can be deceptively complicated.

ETA the note in red