- Ladder play/power matching: The tournament format leaves a lot to be desired. I think the Friday night/Saturday morning sections are good. One question though, is why not have a bracketed round-robin Friday night? I don't see the point of having the seeded-not-bracketed system. The power-matching formula seemed to work pretty well, though I have some questions/suggestions. Most importantly: what was the effect of the bye on 6 teams (including ours) that played team 42 on Friday? Were the scores multiplied by 6/5? Was the bye round counted as a win against a winless opponent (which means that scores were unchanged)? The modified Swiss-pairs portion works very well, as in previous years. I would suggest that the organizers print out a few more columns of the spreadsheet as they send out the schedule: without us knowing the win/loss records of our opponents, we can't check that an error wasn't made in the seeding. I appreciate that the schedulers waited to check that all was correct before posting the next round, but errors can still occur, and apparently did (since the schedule was changed a few times). Ladder play is useless. Let me correct that: ladder play allows NAQT to put teams 1-42 (or 1-26, for Div II) in order easily. I've said this often and will say it again, trying to make my point. I can probably put it best in computer science terms: Ladder play is essentially a bubble sort, where you only have a slightly better than 50% chance of making the "right" sort each time, if you call the team that will more often win in a certain matchup the "right" one. This is probably the MOST inefficient way of sorting teams top to bottom. As I said previously, I believe that the modified Swiss pairs works very well in sorting teams to similar ability. Thus you end up playing the team just above or below you, and will end up playing very close games... games hinging on a single question likely (as all of our ladder play games did). A single loss or win in these games doesn't mean much, but the total number of wins/losses vs. similarly rated teams would work better. A four-game round-robin among groups of 5 teams would be much better at determining their relative position. If you dislike not allowing teams to move between brackets, then make brackets of 4 which then seeds for an offset round robin. - Other comments: Though I'm happy with our final placing, and believe it was probably (approximately) correct, I do have a niggling doubt or two. Penn State lost 5 matches, by a combined total of less than 150 points! If we had been blown out by Illinois or Harvard or whomever we would have known more concretely that we weren't suited for such lofty places. On the other hand, the fact that we lost to teams who weren't so highly ranked means we weren't that good. When you win is more (equally?) important than who you beat: we lost to Carleton in the final round of Swiss pairing, dropping our record to 7-3. We beat them twice in the ladder play, however. Had we beaten them the first time, we would have been seeded ~7th instead of 11th. Illinois beat Chicago twice before losing to them once, and hence had to beat them twice in the finals to win the whole thing. This is an artificial effect of the seeding process, and probably should be addressed in future setups. -On the whole, I enjoyed the tournament, but think that the fancy-shmancy methods of ranking teams have hidden pitfalls. With more mundane methods (such as what every other tournament uses!) you know what you are getting into beforehand. Rob
This archive was generated by hypermail 2.4.0: Sat 12 Feb 2022 12:30:42 AM EST EST