caught  reversal keeps the sign of gross  (FAIL as worked)
caught  reversal pays no cost  (FAIL as worked)
caught  reversal pays the cost twice  (FAIL as worked)
caught  t with n instead of n - 1  (FAIL t-statistic)
caught  profit factor inverted  (FAIL profit factor)
caught  eligible at 25 trades  (FAIL 29 development trades: not eligible)
caught  eligible on development only  (FAIL 29 later trades: not eligible)
caught  rule A on gross  (FAIL a version losing only to costs: rule A, not B)
caught  rule B on net  (FAIL a version losing only to costs: rule A, not B)
caught  rule C at -1  (FAIL a mild loser with t -1.48: rule B, not rule C)
caught  index cut a year early  (FAIL index futures)
caught  gold validation ends 2020  (FAIL spot gold)
caught  bitcoin cut moved  (FAIL bitcoin)
caught  SPY cut moved  (FAIL SPY)
caught  GC treated as gold  (FAIL market groups)
caught  later excludes the holdout  (FAIL reversed: development 40, later -60)
caught  gate 1 at 1.0  (FAIL validation PF 1.10: gate 1 fails)
caught  gate 2 at 25  (FAIL under 50 trades in a period: gate 2 fails)
caught  gate 3 at 1.0  (FAIL development t 1.48: gate 3 fails)
caught  FX cut moved  (FAIL FX)
caught  shuffle p in one direction only  (FAIL perfect separation the other way: difference -1, p 1)
caught  gate 4 reads the percentile as is  (FAIL all pass)
caught  gate 5 keeps the top 10  (FAIL positive only through its 10 best trades: gate 5 fails)
caught  gate 6 on all trades  (FAIL holdout negative: gate 6 fails)
caught  shares by count, not by test  (FAIL test-weighted 45%, not 9 of 11)
caught  shuffle across tests  (FAIL with one version per test, the labels cannot move)
26 of 26 caught