Part 1: false-positive rate (p < 0.05), no true difference, 20000 runs per row n = 6 per group: log-rank 0.0607 Cox Wald 0.0370 Cox LR 0.0564 Cox score 0.0607 n = 8 per group: log-rank 0.0571 Cox Wald 0.0422 Cox LR 0.0536 Cox score 0.0571 n = 10 per group: log-rank 0.0566 Cox Wald 0.0457 Cox LR 0.0525 Cox score 0.0566 n = 20 per group: log-rank 0.0531 Cox Wald 0.0473 Cox LR 0.0504 Cox score 0.0531 Part 2: crossing curves (treated hazard 3x before day 20, 0.25x after), 5000 runs per row n = 10 per group: share with p < 0.05 log-rank 0.1782 RMST(60) 0.2882 n = 20 per group: share with p < 0.05 log-rank 0.2676 RMST(60) 0.4362 n = 40 per group: share with p < 0.05 log-rank 0.4426 RMST(60) 0.6880 True RMST(60): control 35.106, treated 22.761 days