2019 D-Value Revision
Prior to its 2019 Sectional Championship Tournaments (SCTs), ÎÞÓǶÌÊÓƵ has made minor revisions to the D-Value system used to compare the performance of teams at different SCTs.
The strength-of-schedule (SOS) factor in the D-Value calculation has been altered to better reflect the performance of teams (or, more precisely, teams' opponents) that played in combined fields at the SCT (that is, when low registration required Division I and Division II teams to compete against each other on the same questions). The revision does not apply to the D-Values for the 2019 Community College Sectional Championship Tournaments, but will be applied to those events in future years. (Since CC SCTs do not have combined fields in which some teams play on the "wrong" question set, the changes to those values are small.)
SOS was previously calculated by taking a team's opponents' average tossup points per tossup heard (TUPPTUHopp) and dividing it by the average team's tossup points per tossup heard across all SCT sites (TUPPTUHavg). Under the revised calculation, SOS is calculated by taking a team's TUPPTUHopp plus the average points per bonus of the team's opponents (PPBopp) over the sum of the TUPPTUHavg and average team's points per bonus across all SCT sites (PPBavg). In short, the factor now includes bonus performance.
For SCT teams that play in a combined field, the SOS calculation will now use "corrected" values. Specifically, when calculating TUPPTUHopp, TUPPTUHavg, PPBopp, and PPBavg, the average will be taken over teams' corrected values: Corrected values are the raw values multipled by the appropriate (tossup or bonus) correction factor. (Previously, the SOS calculation was made on raw values, with the correction factors applied in the final D-Value calculation.)
In conjunction with this change, the bonus correction factor (DCB) will now be 0.82 for Division I teams that played on Division II questions and 1.22 for Division II teams that played on Division I questions; the tossup correction factor (DCT) will remain unchanged.
Based on research on past SCT and ICT performances, ÎÞÓǶÌÊÓƵ found that this revision improved the predictiveness of D-Values by a modest, but still significant, amount. We wish to thank Ryan Rosenberg for his suggestion that spurred this research and revision.
Questions and comments about these changes can be directed to ÎÞÓǶÌÊÓƵ at [email protected].