I wrote: <<Also, if NAQT had given out the secret formula, SCT hosts could have used it to compute rankings of the participating teams.>> To which ehillema replied: > Actually, our rankings can't be computed until all results from all > sectionals are in hand, due to the importance of the calculation of > each team's strength of schedule, which involves a comparison of the > average statistics of each team's actual collective opponents over the > course of their tournament with the statistics of the overall average > statistics for all teams from the same division playing on the same > questions in all tournaments. You need all of the results before you > can compute the exact percentage by which a given team's actual > schedule was statistically harder or easier than the national average, > and you can't complete our formula without knowing that. In other words, if two teams, A and B, are playing in the same division at the same SCT, then depending on the results of *other* SCTs, NAQT's formula might rank A above B, or B above A. This shows that the ranking system is absurd. If SCT hosts had all the data needed for rankings except for one missing parameter, which was "the exact percentage by which a given team's actual schedule was statistically harder or easier than the [inter]national average," then they could still compute local rankings with the formula if they made a guess of what this parameter might be. For example, a host could announce: "We haven't heard from other sectionals yet, but assuming that the field here is 30% stronger than average, the rankings are A, B, C, D, E. Assuming that the field here is 30% weaker than average, the rankings are A, C, B, E, D." > I've described the thing enough last year and this that the strategies > are clear enough: score as many points as possible in all of your > matches, and also (though admittedly less crucial until now unless an > automatic invitation via a title is at stake) win as many as > possible. These may be the strategies that the formula is *intended* to reward, but we do not know the weightings of the different factors.
This archive was generated by hypermail 2.4.0: Sat 12 Feb 2022 12:30:43 AM EST EST