Greater differentiation at tournaments / problem of brackets

[Standard-winged Nightjar]

Just want to run with something that came up a bit in Fred's recruitment thread as well as the Centennial tournament thread. I think that there's clearly a problem with tournaments where too many matches take place that are likely to involve perennial powerhouses going up against newbie teams. This is particularly so, since even if rebracketing takes place in the afternoon, many newbie teams don't want to stick around for them. For them, 5 rounds (or even fewer) is enough, and having gone through getting clobbered at least once, usually 2-4 times in the morning, many kids and teams are dispirited and just want to get home. I know that standard practices promote bracketing and lots of games, but I think this is a turnoff for many teams.
Greg Bossick (pro-bracket) and me (anti-bracket) have debated this point a number of times, and here's my take. Basically, while bracketing is certainly fair (though, like any tournament set up, it can often be hard to estimate team ability for unknown schools), it's also regressive. If you have 6 teams in a bracket, and imagine their strengths are evenly spaced 1-6, and you do a full round robin, then the best team plays on average teams with ability level 4. The worst team plays on average teams with difficulty level 3. This seems minor, but it's certainly not insignificant, since the net result is one more clobbering (which has dubious value for anyone).
The way I see it, there are 3 ways around this. The first is a card system, which works great at HSNCT, but is complex and requires a lot of know-how with a thin margin of error. The second is what we use at NHBB, which is to say, I make the schedules for all teams, and take known ability into account. Ideally, over the five prelim matches in an NHBB tournament, the best teams play a slightly harder schedule than average, the worst teams play a slightly easier schedule (this is the reverse of bracketing). Invariably, the top teams make the playoffs anyway - and very rarely is the seeding mixed up (we've had literally no complaints to this approach over almost 40 tournaments this year and last). But at the same time, the top teams get better matches, and the worst teams aren't as discouraged. If we have crossover games with the JV division, then typically B teams of unknown or weak teams play those matches. At the same time, I don't segregate completely - there needs to be some mixing of good teams and bad, since otherwise, the records would be the same and this would screw up the playoffs (though we could theoretically account for this by doing away with playoff advancement being based on record, and go over to an all point total playoff model. I don't really want to do that though- I want teams to play for the win in their prelim matches and have wins count.)
That system however only makes sense if you have a schedule setter who is aware of how to calibrate the rank of teams as best as possible. It's not always foolproof, but I think I do a good job with it. The other way, though, that I think needs to get used much, much more is to have different divisions at tournaments. Even a relatively small tournament (say, 16-20 teams) might do much better with having 2 divisions, and large tournaments such as Centennial, Half Hollow Hills, and many in the south, would do well to have as many as 3. I know that Chatham's tournament does this in NJ with self-segregating elite, standard, and freshman divisions. In NHBB, we have a JV division, which often needs to get lumped in with the weaker varsity teams, since we often don't get 6 JV teams (this will eventually change, though). For us, JV teams can't have anyone in 11th or 12th grade - I could see a case being made though for having 1 11th grader.
Honestly, I am surprised that other Nationals don't have a JV division, and that more tournaments don't do the self-segregating mechanism. That way, you give teams the chance to compete for the top title right from the start, but if they're more interested in having more matches against more even competition they can opt for that instead.
Anyway, I am a firm believer that a huge reason why the DC (and other?) circuits have a problem with team retention is due to the bimodal distribution of team ability. As more top teams get even better (again, the arms race has by no means even come close to its ultimate destination - the average top 100 national team 10 years from now would blow the average top 100 national team today out of the water, guaranteed) this is a problem that needs serious addressing. Otherwise it will be pyramidal quizbowl = elitism and lack of fun for many new and less talented teams. And we're all the worse off. Having more tournaments get away from strict bracketing conventions and especially adopting more divisions (which is not hard to do) can go a long way to revitalizing quizbowl circuits and jumpstarting developing ones.
Very interested to hear everyone's thoughts on this.

[Ben Dillon]

The first is a card system, which works great at HSNCT, but is complex and requires a lot of know-how with a thin margin of error.

Chess tournaments for years have used Swiss system pairing programs. Seems to me there could be a downloadable program a la SQBS to pair on the fly.

The second is what we use at NHBB, which is to say, I make the schedules for all teams, and take known ability into account.

How are you doing this? Numerically? Holistically? Historically?

The way I see it, there are 3 ways around this.

Perhaps I'm an idiot, but I couldn't discern what was your third option.

[nadph]

The way I see it, there are 3 ways around this.

Perhaps I'm an idiot, but I couldn't discern what was your third option.

I think it was buried in the middle of that second-last paragraph:

The other way, though, that I think needs to get used much, much more is to have different divisions at tournaments.

[Charles Martel]

Just seeding the top teams can be difficult as it is. Illinois has had more tournaments this year than anywhere else, but it's still not clear who should be seeded in the top 8 or the top 6.

[Mr. Scogan]

As a player, I think the following should be kept in mind when talking about power-matching.

Card systems when used have great intentions behind them, but they aren't applied right when you don't have powers of 2 as a field size. My personal experience of a tournament using "stupid" matching went like this; 36 teams on 12 packets (or maybe it was eleven?) having x amount of teams that play nine games, y amount of teams that play eight games, and z amount of teams (usually the pretty bad teams/teams that don't play up to the field) get seven games. At this tournament, we had the highest PPG of any team, yet we finished 5/36 because we lost our first game on one tossup. After that first round loss, we won the next seven games, but were ruled out of playoff contention (the "playoff" was a one game final between whoever held the #1 and #2 card at the end of the day). What's even funnier is that there was a team that finished with FOUR losses that finished ahead of us. I'm not someone who gets stupid weird about this, but its annoying to have to get three byes, hear less questions and play down to worser competition for the interest of keeping the game "enjoyable" for the inexperienced teams or teams that have a snowball's chance in hell of getting the championship, not to mention it certainly doesn't make the tournament enjoyable for the teams that are on the crest of "the championship".

What some tournaments do (and I like this idea) is arbitrarily have one "pre-preliminary" round with cards and whoever gets the A card goes to the A pod with x amount teams of relatively similar strength, B goes to B, etc, but seeded too.

I'm not a tournament director, so correct me where wrong.

[The Predictable Consequences]

Self-selecting divisions (with the TD holding final discretion) seem to make the most sense. It enables teams to get the most games possible against other teams of similar skill level.

[Matt Weiner]

There is pretty much no reason to ever do a card system or other form of swiss pair/power matching outside of something like HSNCT where it's both literally impossible to do anything else and you take an enormous number of teams to the playoffs. I remain unconvinced that such a system accomplishes anything besides creating moral hazards (ie penalizing winning and creating a huge advantage for losing at the right time) or that anyone in quizbowl would think it was a good thing were it not used in other, incommensurable activities/a good way to show off your irrelevant mathematical prowess. Please do not use power matching.

[Black-throated Antshrike]

There is pretty much no reason to ever do a card system or other form of swiss pair/power matching outside of something like HSNCT where it's both literally impossible to do anything else and you take an enormous number of teams to the playoffs. I remain unconvinced that such a system accomplishes anything besides creating moral hazards (ie penalizing winning and creating a huge advantage for losing at the right time) or that anyone in quizbowl would think it was a good thing were it not used in other, incommensurable activities/a good way to show off your irrelevant mathematical prowess. Please do not use power matching.

If you preseeded the cards around where you think they will place, this would work out I think (I am not an expert on this at all so I could be absolutely wrong and incorrect). Tressler told me at some point that he did something like this. Honestly, the card system is way to easy to mess up if you don't know what you're doing. If done absolutely properly, I believe it could work out well enough without people getting screwed over.

[cvdwightw]

I'm having a lot of trouble understanding how "TD-derived schedules" are more fair to teams. Suppose that we have 6 teams ranked 1-6 at equal ability spacing. Let's take a "difficulty-balanced" five-round schedule:

Team 1: 5-4-3-2-2
Team 2: 6-3-4-1-1
Team 3: 4-2-1-5-6
Team 4: 3-1-2-6-5
Team 5: 1-6-6-3-4
Team 6: 2-5-5-4-3

Yes, Teams 1 and 2 have a harder schedule than Teams 5 and 6, but from hardest to easiest schedules, it goes 2,1,4,3,6,5. So while Team 5 is going 2-3 against a relatively easy schedule, Team 4 is going 2-3 against a tougher schedule - and similarly, Team 2 and 3 are going 3-2 against different schedules. It gets even worse if, say, Team 2 and 3 are mis-seeded and end up playing each other's schedules. Team 2 is now going 4-1 with Team 3's schedule and Team 3 is going 2-3, the same as Team 4 and Team 5. And unless you're breaking ties by something like head-to-head or strength of schedule, you've got a major issue trying to break that three-way tie (i.e. total points is going to do a terrible job of promoting the best team).

If you're the worst team in the bracket, you have a harder schedule because you don't get to play yourself. This is common sense or something. There is a monotonic relationship between your ability and your strength of schedule, assuming seeds are correct. If the TD tries to re-jigger this such that worse teams play easier schedules and better teams play harder schedules, then there just isn't going to be the same relationship between ability and strength of schedule, even if seeds are correct. In other words, for middle of the pack teams trying to eke out a playoff spot in a relatively forgiving playoff, brackets don't care if you're a "good" middle team or a "bad" middle team - you'll beat the bad teams, lose to the good teams, and advance if you beat the other middle team(s). For individual schedule making, there can be inherent advantages to being seeded in certain parts of the "middle of the road" - you get to play more "bad" teams, meaning it's more likely you'll finish with one more win than a similar-ability team that has to play more "good" teams just because it was seeded a few places higher.

By the way, my general philosophy is "anything that quizbowl has discarded for not working probably shouldn't be reintroduced." This includes things like ladder play, variable-value bonuses, and "TD makes individual schedules for each team," which was abandoned by NAQT after requiring a 5-way tiebreaker to figure out the last playoff spot at the 2001 HSNCT.

Okay, last aside: TERRIBLE SPORTS ANALOGY time. This "difficulty-balanced schedule making" is exactly what the NFL uses. The NFL likes this because it helps promote the illusion of parity - a bad team that suddenly gets good is a lot more likely to make the playoffs because it gets to beat up on two bad teams, while a team in the same division that was already good replaces those games with games against one or two good teams. Yes, the best teams (Patriots, Steelers, Packers, etc.) will almost invariably make the playoffs. But that second tier of teams is hugely affected by their schedule: two teams can split their games, have the same record against common opponents, and have their final standing determined by how difficult their opponents in those two games are. As an example, the wild card order for the AFC in 2007 was determined like this. Jacksonville and Tennessee split their games and went 8-4 against common opponents. However, Jacksonville earned the #5 seed based on beating Buffalo and Pittsburgh, while Tennessee split with NYJ and Cincinnati and was relegated to the #6 seed. (incidentally, Buffalo and Pittsburgh were by far the harder opponents that season; nevertheless, that order would have been flipped had Jacksonville lost one of the two games against teams Tennessee didn't play)

The point of the above is that the NFL likes it because it creates excitement, but it isn't always fair because one team can be promoted over another based on games the second team didn't have the opportunity to play. I remain unconvinced of the merits of "difficulty-balanced" scheduling in any activity in which the purpose is to fairly delineate the distinction between playoff and not-playoff teams.

[Standard-winged Nightjar]

If you're the worst team in the bracket, you have a harder schedule because you don't get to play yourself. This is common sense or something. .... For individual schedule making, there can be inherent advantages to being seeded in certain parts of the "middle of the road" - you get to play more "bad" teams, meaning it's more likely you'll finish with one more win than a similar-ability team that has to play more "good" teams just because it was seeded a few places higher.

I get the not playing yourself bit. But it's still regressive - there's no getting around that. And when I set matches, keep in mind it's not that I just divide teams into 3 piles of good, medium, and bad. It's more a continuum, at least in theory. But even with bracketing, it's often hard to gauge the strengths of new teams. For NHBB it's even harder since I don't have as many straight history stats to go off of.

By the way, my general philosophy is "anything that quizbowl has discarded for not working probably shouldn't be reintroduced." This includes things like ladder play, variable-value bonuses, and "TD makes individual schedules for each team," which was abandoned by NAQT after requiring a 5-way tiebreaker to figure out the last playoff spot at the 2001 HSNCT.

This assumes that every quizbowl innovation has been a positive development. But this isn't always true, especially when it's not clear what the point of the exercise is. When the imperatives of fairness and appealing to new teams conflict, does fairness have to win out every single time? This would mean all tournaments would look like NSC. At some level, especially in prelim matches at non-national tournaments, taking the desires of new teams into account (including not getting clobbered as much) helps.
Moreover, NHBB tournaments are often of amounts of teams that don't lend themselves well to brackets. I'm not turning away teams just because 26 is an awkward number. Recruitment might matter more for us than others in qb, but again, I think everyone benefits.

[Kenneth Widmerpool]

I think that you're engaging in some dangerous, unnecessary psychologizing of new teams here. Rather than try to speak for them, the best course of action is to offer them the fairest format, which will leave them happy over the average of a great number of instances. I'm certain that most new teams won't want this kind of preferential treatment.

[ether a go-go]

I think that there sometimes is a rush to do all tournaments the same way, and TDs are better off if they have various options available based on the specifics of their tournament. All systems have flaws, and anybody who thinks the NSC system works perfectly must not have been paying attention in 2010 or 2011. (Of course, in 2011, the use of pools was not the only flaw.)

A few weeks ago, we had a situation were there were 64 teams, teams that didn't want to play until dinner time, a shortage of staffers, and bounceback bonuses. Under those circumstances, the options were pool play or cancelling the tournament.

The notion that the only reason for pool play is showing off mathematical prowess is absurd. For one thing, there are resourceson the internet for setting up such schedules, so nobody should be impressed by a TD who uses power matching, and for another thing there have been plenty of tournaments where the TD has used power matching and credited somebody else. As to moral hazards, it depends on the details--the google doc I just linked to has a schedule with a grand total of exactly zero moral hazards, and at the HSNCT there were moral hazards in fewer than one percent of the matches (and I'm hoping to decrease if not eliminate such matches in the future).

As far as Madden's scheduling method, I would have to see the details before making a judgment. As to his fairness comments about standard pools, he is absolutely correct. A team with a good reputation that shows up shorthanded without having played or practiced in the past month has a major advantage in pool play over a team without that reputation that finally started reading packets regularly and showed up at full strength and on that day is every bit as good as the other team, assuming they are placed in different pools (which probably will happen). The former team still has an advantage in power matching, but that advantage is decreased.

Having said all that, we will use pools at our tournament next week.

[mithokie]

Here are the various methods of tournament play that I have seen used in quizbowl tournaments:
1. Single Elimination Tournament
2. Double Elimination Tournament
3. Full Round Robin
4. Randomly assigned schedule followed by single elimination
5. Pooled Prelims followed by single elimination tounrament
6. Pooled Prelims followed by Rebracketed Pools
7. Power Matching followed by Double Elimination Tournament
8. Pooled Prelims followed by Power Matching for lower seeds and pool play for upper seeds
9. Pooled Pelims followed by Playoff Pools followed by Superplayoff pools
10. Pooled Prelims followed by Separate Division Playoff Pools with Cross-Bracket Games
11. 2 pool prelim games followed by cross bracket placement game

The one method that I have not personally seen, and that I think might warrant a fair shake is to use power matching for morning prelims, followed by round-robin pool play. Power matching would allow teams to play teams of like ability more quickly than they would while waiting for a 6 or 8 team prelim pool where half of the games could be blow outs. If you then use the results of the AM power matched results to rebracket teams into homegeneous ability group brackets, and then have a round robin pool play within each group, teams will get more matches against teams of like ability than any of the other methods mentioned above.

This obviously works most cleanly if there are 2^k teams playing, but can still be done with other numbers of teams.
With 32 teams, teams would finish the prelims with a card of number 1 through 32 roughly approximating their ability.
Card 1 would have a 5-0 record
Cards 2-6 would have 4-1 records
Cards 7-16 would have 3-2 records
Cards 17-26 would have 2-3 records
Cards 27-31 would have 1-4 records
Card 32 would have an 0-5 record

After prelims, teams would be sorted first by record, and then by PPB (since teams will have played different schedules).

If you had time for 10 total rounds (plus 1 or 2 championship matches), then group the teams into playoff groups of size 6-6-6-6-8*(You would play an incomplete round robin in the bottom group).

If you had time for 12 total rounds (plus 1 or 2 championship maches), then group the teams into playoff groups of size 8-8-8-8.

Thoughts...

[mtimmons]

The one method that I have not personally seen, and that I think might warrant a fair shake is to use power matching for morning prelims, followed by round-robin pool play. Power matching would allow teams to play teams of like ability more quickly than they would while waiting for a 6 or 8 team prelim pool where half of the games could be blow outs. If you then use the results of the AM power matched results to rebracket teams into homegeneous ability group brackets, and then have a round robin pool play within each group, teams will get more matches against teams of like ability than any of the other methods mentioned above.

I think has happened before. I think this is what happened at the 2009 New Trier Varsity http://www.hsquizbowl.org/forums/viewto ... =20&t=8003 which generated lots of debate. I agree that it is a good idea though for 32-team tournaments as it both gives team many games against teams with similar skills and requires a team to lose twice before being eliminated from championship contention something which does not happen in double round robin with this many teams unless a very large number of rounds are played. I'm pretty skeptical that this a good idea for tournaments that are neither very large nor have 2^n teams though.

[cvdwightw]

If you're the worst team in the bracket, you have a harder schedule because you don't get to play yourself. This is common sense or something. .... For individual schedule making, there can be inherent advantages to being seeded in certain parts of the "middle of the road" - you get to play more "bad" teams, meaning it's more likely you'll finish with one more win than a similar-ability team that has to play more "good" teams just because it was seeded a few places higher.

I get the not playing yourself bit. But it's still regressive - there's no getting around that. And when I set matches, keep in mind it's not that I just divide teams into 3 piles of good, medium, and bad. It's more a continuum, at least in theory. But even with bracketing, it's often hard to gauge the strengths of new teams. For NHBB it's even harder since I don't have as many straight history stats to go off of.

I will always favor a strictly regressive schedule over a schedule that provides perverse incentives to underperform. Say I'm a good but not elite team at an NHB tournament, and I get a favorable playoff packet such that I've clinched the game after tossup 3 of the fourth quarter. I have a perverse incentive not to buzz for the rest of the game, in order to look worse than I really am and thus get an easier schedule at a future NHB tournament, compared to someone that may have advanced to the same level at a different NHB qualifying tournament. Similarly, if I'm that team's playoff opponent and there's no way I'm coming back, I too have a perverse incentive not to buzz and, therefore, look worse than I really am. Once one team or the other has clinched a win, neither team has an incentive to put up more points - both teams have an incentive to make themselves look worse and, in doing so, will make their opponent's strength of schedule appear worse as well.

[Mechanical Beasts]

Now, I believe that fair formats ought to be used. But aren't

the fairest format... will leave them happy over the average of a great number of instances

and

most new teams won't want... preferential treatment.

both

dangerous, unnecessary psychologizing of new teams

or what?

[Howard]

There are numerous good points in this thread, and I agree with many of them. Regardless, sitting around discussing on a message board why teams are not attending tournaments, although a necessary and useful exercise, is still largely speculation.

Is it possible to take the team list from a tournament like Centennial's, which drew a large number of teams not regularly seen on the circuit, and simply survey the coach to find out why they don't attend more tournaments? I suspect most coaches will be quite frank with us.

[Standard-winged Nightjar]

If you're the worst team in the bracket, you have a harder schedule because you don't get to play yourself. This is common sense or something. .... For individual schedule making, there can be inherent advantages to being seeded in certain parts of the "middle of the road" - you get to play more "bad" teams, meaning it's more likely you'll finish with one more win than a similar-ability team that has to play more "good" teams just because it was seeded a few places higher.

I get the not playing yourself bit. But it's still regressive - there's no getting around that. And when I set matches, keep in mind it's not that I just divide teams into 3 piles of good, medium, and bad. It's more a continuum, at least in theory. But even with bracketing, it's often hard to gauge the strengths of new teams. For NHBB it's even harder since I don't have as many straight history stats to go off of.

I will always favor a strictly regressive schedule over a schedule that provides perverse incentives to underperform. Say I'm a good but not elite team at an NHB tournament, and I get a favorable playoff packet such that I've clinched the game after tossup 3 of the fourth quarter. I have a perverse incentive not to buzz for the rest of the game, in order to look worse than I really am and thus get an easier schedule at a future NHB tournament, compared to someone that may have advanced to the same level at a different NHB qualifying tournament. Similarly, if I'm that team's playoff opponent and there's no way I'm coming back, I too have a perverse incentive not to buzz and, therefore, look worse than I really am. Once one team or the other has clinched a win, neither team has an incentive to put up more points - both teams have an incentive to make themselves look worse and, in doing so, will make their opponent's strength of schedule appear worse as well.

I understand this point in theory, though I'd be surprised if it really happened much at all. It can also be countered relatively effectively by looking at NHBB Bee stats, and a general knowledge of teams' strengths. This is a potential downside, but the net result of it is likely to be very slight in the grand scheme of things in the relatively unlikely instances when it occurs.

On John's point about surveying teams, yes, yes, yes. If I get to it, I will put something like this together for the Half Hollow Hills tournament, which typically brings in 30+ novice teams that don't do many other tournaments. Assuming I get to it, I will try and make sure every team fills it out, so as to avoid selection bias. The question arises though, what to do if the prerogatives of novice teams conflict somewhat with the prerogatives of top teams? Let's see what the teams say, but keep open minds about diverging viewpoints and be ready to compromise somewhat if the data seems to indicate that might not be a bad idea.

[The Predictable Consequences]

I think yesterday's Lower Hudson NHB scheduling highlighted the major problem with difficulty adjusted scheduling. Take Ardsley B and Irvington A--two solid teams from last year, both with a history of program continuity and strength. Thus, they received a "slightly" harder schedule than other teams. Practically, what this meant was that each team struggled through the prelims at 2-3, losing only to teams with combined records of 15-0 and 14-1. That strength of schedule, coupled with the decision to rank by W-L record first, destroyed any chance those teams had at a competitive playoff game. Or, for that matter, a Nationals bid, since going 3-2 or winning a game in the playoff would have required beating one of the top 6 teams at the tournament for Ardsley B or Irvington A. Actually, Ardsley B ended up having played every one of the semifinalists of the Bowl. While I like the idea in the abstract of adjusting the schedule based on supposed strength of team, it needs to be done much much more carefully and much much more slightly if it is to be done at all.

[Standard-winged Nightjar]

Here were the schedules and here was my reasoning for the two teams you mentioned:
Ardsley B
1. Stuyvesant (a strong team, that won the tournament but was hard to gauge because they had never played NHBB before and don't play all that many tournaments)
2. Croton-Harmon A (an A team, but of a team that didn't play NHBB last year, and I don't associate with being one of the stronger Westchester schools)
3. Nanuet (went 3-2 last year off a weak schedule, but otherwise doesn't play quizbowl)
4. Scarsdale A (very good team)
5. White Plains A (very good team)

In retrospect, this was a tough schedule, and I apologize for sending Ardsley B up against 3 really strong teams. I'd consider Ardsley B a middle of the pack team in this field, and they should have had someone other than White Plains A in their final match to give them a better shot of going 3-2. If anything, this schedule was more regressive than what pool play would have likely generated.

Irvington A
1. Woodlands B (should have been an easy win)
2. Kellenberg A (tough match, could have gone either way, I thought)
3. Wilton A (probable win, since Wilton doesn't really do other quiz bowl)
4. Ardsley A (tough match, maybe 20% chance of pulling off an upset)
5. White Plains C (should have been an easy win)

I make fewer apologies for Irvington - this schedule to me should have produced something like a 3-1-1 record. Since I would have ranked Irvington about 8 in this field of 26, that doesn't sound off.

Beyond this, two points. First, this was an abnormally strong an competitive field. There were something like 7 teams going in with a shot at the title, I thought:
1. White Plains
2. Stuyvesant
3. Ardsley
4. Scarsdale
5. Kellenberg
6. Hastings
7. Irvington or Horace Greeley

Next, consider that Eastchester, Blair Academy and Greens Farms all went to Nationals, and that Horace Greeley was last year's JV champ. Next consider that a number of these teams had B teams, which are often hard to gauge difficulty for. Ardsley B's schedule was somewhat too hard, but the fact remained that this was a very competitive field. Last year, I ended up instituting a wild card policy for NHBB Nationals if teams went 2-3 against a tough schedule. I tend to be disinclined to do that this year, at least yet. But the imperfections of schedule making may make me have a second look at it in early 2012. Any wild card policy would include a point threshold, in any case.

[The Predictable Consequences]

I get the difficulties in seeding some teams that rarely play quiz bowl, and certainly I didn't expect Wilton to play as well as they did. However, I think your estimation of the title contenders was off--there were four, maybe five (semifinalists plus Kellenberg) teams that could reasonably have won. It's unreasonable to expect a TD to predict these things, so why should we exacerbate the problem by tinkering with prelim schedules, especially in with a single-elimination playoff where the prelims matter tremendously in determining a team's eventual success?

[Standard-winged Nightjar]

I'm not sure though; Irvington had 2 playoff teams in DC at last year's Nationals. Hastings has had really good results this year. And Greeley was last year's JV champ. So I stand by my estimate. Keep in mind that often it's also a matter of who shows up, and it's also difficult to estimate how this being a History only tournament stacks up. Beyond that, the bigger issue against using brackets at NHBB is that we are locked in to having 5 prelim games. So really, only mutliples of 6 make sense. Sometimes this happens, as it did in NE Ohio. Most times it doesn't. And the difficulty in predicting how teams will do in an all history format, especially with regards to teams that don't play much qb, and the perennial problem of estimating B/C team and new team (like Wilton) strength, and there's no easy way to do this. I will do a better job going team by team and making sure they don't have 3 games that they have little hope of winning though (a la Ardsley B yesterday).

[Scaled Flowerpiercer]

Just to comment upon this point, especially as Ryan was largely vouching in defense of Irvington, I feel it is appropriate to offer my perspective on it.

While the schedule did end up being very difficult (in the sense that we went 2-3...) I don't think that it was unreasonable for it to formed as it was.

Woodlands B / White Plains D were the easy wins they were expected to be
Ardsley A / Kellenberg A were always going to be difficult matches, though by no means automatic losses; Our loss to Ardsley occurred in a tie-break question - it doesn't get any closer than that - and though our loss to Kellenberg was greater (I believe about 70 points) we didn't play our greatest and though our chances of winning might have been somewhat low, they weren't nil.

Really, the wrench in the works was Wilton, there was almost no data on them from previous tournaments, there was no way to know they would be strong contenders and would therefore ruin our chances of a good showing (beating us in another close match, 220-240)

So even though it did end up hurting Irvington, I think that it was a fair / reasonable schedule and Dave and/or his system are not at fault here.

Of course, perhaps there could be the argument for seeding based on total points over W/L...but I think that question is somewhat open-ended (especially as going by personal experience, it was the first points, then W-L policy that kept Irvington out of the playoffs at last year's state history bowl)