I would like to calculate statistical significance between two samples of sports data - one is actual_wins and the other is expected_wins. actual_wins is a binary data series (0 = loss, 1 = win) and expected_wins is a data series of continuous probabilities.

I have read about t-tests and chi-squared/McNemar (related groups) tests but this data violates the assumptions of both methods as the former assumes a normal distribution and continuous data while the latter assumes categorical/binary data.

Is there any method to calculate significance with this data?

  • $\begingroup$ Try searching this site for Brier score. $\endgroup$
    – mdewey
    Jun 6, 2020 at 15:12
  • $\begingroup$ Yes, thank you @mdewey. $\endgroup$
    – ferdiaoh
    Jun 14, 2020 at 13:38


Browse other questions tagged or ask your own question.