Parabolic SAR self-test, 2026-10-22T08:06:11.329Z

1. The SAR by hand (bars as high / low / close)
   b0 10/9/9.5, b1 11/9.5/10.5 -> rising, prior 9, EP 11; SAR2 = 9.04 capped at low[0] = 9; b2 12/10 -> EP 12, AF 0.04; SAR3 = 9 + 0.04 x 3 = 9.12;
   b3 12.5/11.5 -> EP 12.5, AF 0.06; SAR4 = 9.12 + 0.06 x 3.38 = 9.3228; b4 low 9.3 reaches it -> reversal; SAR5 = the old EP 12.5; b5 11/9 -> EP 9, AF 0.04;
   SAR6 = 12.5 - 0.04 x 3.5 = 12.36, raised to high[4] = 12.4
  [ok] SAR 9.0000, 9.1200, 9.3228, 12.5000, 12.4000; reversal at bar 4
   AF cap: 40 bars each making a new high, lows far above the SAR: the last steps use AF = 0.20
  [ok] step / (EP - SAR) = 0.200000

2. Fills and variants on hand-made bars (one minute per bar; reversals placed at bars 505 and 508)
  [ok] PS-SAR: short +39 pips net, then long to the end +19 (39.0, 19.0)
  [ok] PS-200: the short is refused (close above the EMA), the long taken
  [ok] PS-ADX: the short taken (ADX 30), the long refused (ADX 20)

3. Engine against the independent reference
  [ok] synthetic series: PS-SAR 2/2, PS-200 0/0, PS-ADX 2/2, identical
  real data: 8326700 minutes 2004-01-01 00:03 .. 2026-09-28 15:46 New York; 141042 1-hour bars
  [ok] PS-SAR: 11941 trades, reference 11941, identical
  [ok] PS-200: 5327 trades, reference 5327, identical
  [ok] PS-ADX: 6072 trades, reference 6072, identical

4. Control at offset 0 (same bars, directions and lengths; prices at bar opens by definition)
  [ok] PS-SAR: 11940 control trades on the real bars
  [ok] PS-200: 5327 control trades on the real bars
  [ok] PS-ADX: 6072 control trades on the real bars

PASSED