Performance of a contestant is the rating, which if the contestant had at the start of the contest, the rating change for the contestant after the contest would be zero.
The performance calculation is not perfect, but it's close. The 100% correct approach is too slow to be practical.
Testing on a few contests, the calculated performance can be seen to be between 0 and 4 greater than the true performance.
Note: The issue described in #18 remains, the performance cannot be expected to be very accurate as long as I am forced to work with fake ratings.
Add a performance column.
Performance of a contestant is the rating, which if the contestant had at the start of the contest, the rating change for the contestant after the contest would be zero.
The performance calculation is not perfect, but it's close. The 100% correct approach is too slow to be practical. Testing on a few contests, the calculated performance can be seen to be between 0 and 4 greater than the true performance.
Note: The issue described in #18 remains, the performance cannot be expected to be very accurate as long as I am forced to work with fake ratings.