Check of the slow-drift hold (H2), 2026-10-13 12:51 Halifax
Plan: hold_check_plan.md (sha256 cd1a649bcebe06b3...), not edited after it was written
The cause test's locked files unchanged: True; the protected folders unchanged: True
Base runs reproduce their saved runs exactly: {13: True, 14: True, 15: True, 16: True, 17: True, 18: True}
Moments: T = 2,000 + o + 350 m; o by start {13: 225, 14: 185, 15: 220, 16: 245, 17: 210, 18: 5}

Choosing lambda (old starts 13 to 15, 30 moments, nothing after T)
  2026-10-13 12:20 Halifax: lambda = 0.05 not chosen: on starts 13 to 15 it misses F1, F2, F3a, F3b for the 500-unit hold
  2026-10-13 12:36 Halifax: lambda = 0.1 not chosen: on starts 13 to 15 it misses F1, F2, F3a, F3b for the 500-unit hold
  2026-10-13 12:48 Halifax: lambda = 0.2 not chosen: on starts 13 to 15 it misses F1, F2 for the 500-unit hold
  2026-10-13 12:48 Halifax: no candidate meets the criteria: H2 is not feasible at +-25% as specified; no confirmation
  lambda = 0.05, 500-unit hold (judged)
    carried out 26 of 30 (fewest per start 8); F1 NOT met
      not carried out: start 13, moment 7 (T = 4675): kick 53 at t = 4435: down B infeasible
      not carried out: start 13, moment 8 (T = 5025): kick 28 at t = 4660: down B infeasible
      not carried out: start 14, moment 5 (T = 3935): kick 6 at t = 3460: up A infeasible
      not carried out: start 15, moment 0 (T = 2220): kick 6 at t = 1745: down B infeasible
    F2 NOT met: both fractions within 0.75 to 1.25 in 4% of moments; A_up 0.734 (SD 0.597, -0.589 to 2.093), A_down 0.776 (SD 0.438, -0.145 to 1.680)
    F3a NOT met: every block keeps at least a quarter of the planned difference in 65% of moments (lowest block difference -0.38)
    F3b NOT met: block means over the moments, up [0.785, 0.765, 0.659, 0.707, 0.731], down [0.554, 0.766, 0.877, 0.815, 0.847] (difference [0.67, 0.766, 0.768, 0.761, 0.789])
    F4a met: sham's sum of squared moves / the mean of up's and down's 1.000 (SD 0.000, 1.000 to 1.001)
    F4c met: sham's sum of |moves| / the mean of up's and down's 1.000 (SD 0.000, 0.999 to 1.001); the same kick times in every moment: True
    F4b (reported) within -0.25 to +0.25: A_sham 0.096 (SD 0.558, -0.902 to 2.254)
    the draft's standard (every 100-unit block fraction of up and down within 0.5 to 1.5): 0% of moments; the first check's F3 (25-unit blocks): 0%
  lambda = 0.05, 100-unit hold (reported)
    carried out 29 of 30 (fewest per start 9); F1 met
      not carried out: start 15, moment 2 (T = 2920): kick 18 at t = 2905: down B infeasible
    F2 NOT met: both fractions within 0.75 to 1.25 in 3% of moments; A_up 0.692 (SD 0.535, -0.365 to 1.529), A_down 0.541 (SD 0.431, -0.229 to 1.333)
    F3a NOT met: every block keeps at least a quarter of the planned difference in 90% of moments (lowest block difference 0.04)
    F3b NOT met: block means over the moments, up [0.66], down [0.56] (difference [0.61])
    F4a met: sham's sum of squared moves / the mean of up's and down's 1.000 (SD 0.000, 1.000 to 1.001)
    F4c met: sham's sum of |moves| / the mean of up's and down's 1.000 (SD 0.000, 1.000 to 1.000); the same kick times in every moment: True
    F4b (reported) within -0.25 to +0.25: A_sham -0.009 (SD 0.128, -0.237 to 0.365)
    the draft's standard (every 100-unit block fraction of up and down within 0.5 to 1.5): 31% of moments; the first check's F3 (25-unit blocks): 3%
  lambda = 0.1, 500-unit hold (judged)
    carried out 24 of 30 (fewest per start 7); F1 NOT met
      not carried out: start 13, moment 1 (T = 2575): kick 64 at t = 2390: down B infeasible
      not carried out: start 13, moment 6 (T = 4325): kick 27 at t = 3955: down B infeasible
      not carried out: start 14, moment 1 (T = 2535): kick 54 at t = 2300: down B infeasible
      not carried out: start 14, moment 3 (T = 3235): kick 55 at t = 3005: down B infeasible
      not carried out: start 14, moment 4 (T = 3585): kick 33 at t = 3245: down B infeasible
      not carried out: start 15, moment 3 (T = 3270): kick 20 at t = 2865: down A infeasible
    F2 NOT met: both fractions within 0.75 to 1.25 in 21% of moments; A_up 0.709 (SD 0.382, 0.018 to 1.528), A_down 0.887 (SD 0.278, 0.287 to 1.454)
    F3a NOT met: every block keeps at least a quarter of the planned difference in 83% of moments (lowest block difference -0.08)
    F3b NOT met: block means over the moments, up [0.64, 0.71, 0.547, 0.91, 0.766], down [0.712, 0.933, 0.967, 0.847, 0.868] (difference [0.676, 0.821, 0.757, 0.878, 0.817])
    F4a met: sham's sum of squared moves / the mean of up's and down's 1.000 (SD 0.000, 0.999 to 1.001)
    F4c met: sham's sum of |moves| / the mean of up's and down's 1.000 (SD 0.000, 0.999 to 1.001); the same kick times in every moment: True
    F4b (reported) within -0.25 to +0.25: A_sham 0.023 (SD 0.414, -0.790 to 1.006)
    the draft's standard (every 100-unit block fraction of up and down within 0.5 to 1.5): 4% of moments; the first check's F3 (25-unit blocks): 0%
  lambda = 0.1, 100-unit hold (reported)
    carried out 29 of 30 (fewest per start 9); F1 met
      not carried out: start 15, moment 2 (T = 2920): kick 10 at t = 2865: down A infeasible
    F2 NOT met: both fractions within 0.75 to 1.25 in 14% of moments; A_up 0.746 (SD 0.444, -0.221 to 1.459), A_down 0.628 (SD 0.369, -0.087 to 1.277)
    F3a met: every block keeps at least a quarter of the planned difference in 97% of moments (lowest block difference 0.08)
    F3b NOT met: block means over the moments, up [0.715], down [0.647] (difference [0.681])
    F4a met: sham's sum of squared moves / the mean of up's and down's 1.000 (SD 0.000, 1.000 to 1.001)
    F4c met: sham's sum of |moves| / the mean of up's and down's 1.000 (SD 0.000, 0.999 to 1.000); the same kick times in every moment: True
    F4b (reported) within -0.25 to +0.25: A_sham -0.008 (SD 0.129, -0.238 to 0.345)
    the draft's standard (every 100-unit block fraction of up and down within 0.5 to 1.5): 41% of moments; the first check's F3 (25-unit blocks): 10%
  lambda = 0.2, 500-unit hold (judged)
    carried out 17 of 30 (fewest per start 4); F1 NOT met
      not carried out: start 13, moment 0 (T = 2225): kick 25 at t = 1845: up B infeasible
      not carried out: start 13, moment 2 (T = 2925): kick 29 at t = 2565: down A infeasible
      not carried out: start 13, moment 5 (T = 3975): kick 13 at t = 3535: down B infeasible
      not carried out: start 13, moment 6 (T = 4325): kick 43 at t = 4035: down A infeasible
      not carried out: start 13, moment 7 (T = 4675): kick 24 at t = 4290: down B infeasible
      not carried out: start 13, moment 8 (T = 5025): kick 29 at t = 4665: down A infeasible
      not carried out: start 14, moment 0 (T = 2185): kick 98 at t = 2170: down B infeasible
      not carried out: start 14, moment 3 (T = 3235): kick 83 at t = 3145: down A infeasible
      not carried out: start 14, moment 4 (T = 3585): kick 17 at t = 3165: down A infeasible
      not carried out: start 14, moment 7 (T = 4635): kick 81 at t = 4535: down B infeasible
      not carried out: start 15, moment 0 (T = 2220): kick 33 at t = 1880: up B infeasible
      not carried out: start 15, moment 3 (T = 3270): kick 63 at t = 3080: down B infeasible
      not carried out: start 15, moment 9 (T = 5370): kick 80 at t = 5265: down A infeasible
    F2 NOT met: both fractions within 0.75 to 1.25 in 82% of moments; A_up 0.858 (SD 0.120, 0.606 to 1.077), A_down 0.975 (SD 0.160, 0.613 to 1.195)
    F3a met: every block keeps at least a quarter of the planned difference in 94% of moments (lowest block difference 0.13)
    F3b met: block means over the moments, up [0.775, 0.841, 0.888, 0.942, 0.837], down [0.874, 1.006, 1.044, 0.971, 0.935] (difference [0.825, 0.924, 0.966, 0.956, 0.886])
    F4a met: sham's sum of squared moves / the mean of up's and down's 1.000 (SD 0.001, 0.999 to 1.002)
    F4c met: sham's sum of |moves| / the mean of up's and down's 1.000 (SD 0.000, 0.999 to 1.001); the same kick times in every moment: True
    F4b (reported) within -0.25 to +0.25: A_sham 0.051 (SD 0.560, -0.982 to 1.095)
    the draft's standard (every 100-unit block fraction of up and down within 0.5 to 1.5): 41% of moments; the first check's F3 (25-unit blocks): 0%
  lambda = 0.2, 100-unit hold (reported)
    carried out 29 of 30 (fewest per start 9); F1 met
      not carried out: start 15, moment 2 (T = 2920): kick 10 at t = 2865: down A infeasible
    F2 NOT met: both fractions within 0.75 to 1.25 in 21% of moments; A_up 0.810 (SD 0.339, -0.002 to 1.364), A_down 0.729 (SD 0.313, 0.092 to 1.264)
    F3a met: every block keeps at least a quarter of the planned difference in 97% of moments (lowest block difference 0.20)
    F3b NOT met: block means over the moments, up [0.78], down [0.747] (difference [0.764])
    F4a met: sham's sum of squared moves / the mean of up's and down's 1.000 (SD 0.000, 1.000 to 1.000)
    F4c met: sham's sum of |moves| / the mean of up's and down's 1.000 (SD 0.000, 0.999 to 1.001); the same kick times in every moment: True
    F4b (reported) within -0.25 to +0.25: A_sham 0.005 (SD 0.143, -0.236 to 0.499)
    the draft's standard (every 100-unit block fraction of up and down within 0.5 to 1.5): 62% of moments; the first check's F3 (25-unit blocks): 3%

No lambda was chosen: H2 is NOT FEASIBLE at +-25% as specified. The confirmation was not run.
Reported separately (choosing lambda, old starts 13 to 15)
  lambda = 0.05, 500-unit hold
    history: dM = 0.00500 (SD 0.00203, 0.00106 to 0.00822) against the planned 0.00648; dM / the natural SD of 500-unit history means (0.00548) = 0.91; block SDs over the moments: up [0.89, 1.14, 0.94, 0.66, 0.43], down [0.65, 0.56, 0.6, 0.55, 0.57], difference [0.31, 0.57, 0.37, 0.39, 0.32]
      ratio to target just before a kick: drift up -0.0397 (SD 0.1142, -0.2921 to 0.2339), down 0.1040 (SD 0.1594, -0.2503 to 0.4339); spread up 0.410, down 0.567; 5th to 95th percentile up [-0.654, 0.678], down [-0.691, 1.142]
    effort: 101 kicks; sum of squared moves up 4.11 (SD 3.97, 0.27 to 14.34), down 4.36 (SD 5.93, 0.08 to 28.79), sham 4.24 (SD 4.02, 0.96 to 21.53) (the first check's pinned hold: up 216.3, down 109.8, sham 131.5; one pulse at +-25%: 0.92)
      sum of |moves| up 458.3 (SD 112.2, 267.7 to 728.1), down 363.9 (SD 133.1, 145.2 to 834.4), sham 411.1 (SD 104.8, 261.3 to 780.7); largest move up 0.073, down 0.085, sham 0.085
      first kick's share up 0.823, down 0.819, sham 0.901; median later kick (sum of squared moves) up 0.0015, down 0.0008
      each kick's sham-to-arms ratio: sum of squared moves median 1.000 (0.957 to 1.043); sum of |moves| median 1.000 (0.973 to 1.025)
  lambda = 0.05, 100-unit hold
    history: dM = 0.00429 (SD 0.00257, 0.00018 to 0.00961) against the planned 0.00655; dM / the natural SD of 100-unit history means (0.00702) = 0.61; block SDs over the moments: up [0.47], down [0.42], difference [0.27]
      ratio to target just before a kick: drift up -0.0627 (SD 0.1133, -0.2627 to 0.1599), down 0.1734 (SD 0.1559, -0.1149 to 0.4831); spread up 0.359, down 0.548; 5th to 95th percentile up [-0.59, 0.529], down [-0.588, 1.064]
    effort: 21 kicks; sum of squared moves up 2.63 (SD 2.59, 0.13 to 11.11), down 1.84 (SD 1.51, 0.13 to 5.80), sham 2.23 (SD 1.49, 0.25 to 6.37) (the first check's pinned hold: up 11.3, down 7.8, sham 8.3; one pulse at +-25%: 0.92)
      sum of |moves| up 146.2 (SD 60.9, 62.4 to 342.0), down 128.3 (SD 37.6, 56.3 to 206.4), sham 137.2 (SD 40.7, 71.7 to 257.4); largest move up 0.066, down 0.046, sham 0.066
      first kick's share up 0.970, down 0.962, sham 0.985; median later kick (sum of squared moves) up 0.0008, down 0.0007
      each kick's sham-to-arms ratio: sum of squared moves median 1.000 (0.977 to 1.016); sum of |moves| median 1.000 (0.984 to 1.014)
Reported separately (choosing lambda, old starts 13 to 15)
  lambda = 0.1, 500-unit hold
    history: dM = 0.00563 (SD 0.00189, 0.00114 to 0.00853) against the planned 0.00648; dM / the natural SD of 500-unit history means (0.00548) = 1.03; block SDs over the moments: up [0.71, 0.73, 0.51, 0.5, 0.42], down [0.52, 0.54, 0.5, 0.29, 0.45], difference [0.32, 0.42, 0.32, 0.27, 0.28]
      ratio to target just before a kick: drift up -0.0417 (SD 0.0722, -0.1740 to 0.1142), down 0.0733 (SD 0.1024, -0.1578 to 0.2755); spread up 0.385, down 0.547; 5th to 95th percentile up [-0.632, 0.615], down [-0.692, 1.059]
    effort: 101 kicks; sum of squared moves up 4.51 (SD 4.16, 0.72 to 15.81), down 5.29 (SD 6.51, 0.37 to 31.22), sham 4.90 (SD 4.49, 1.26 to 23.51) (the first check's pinned hold: up 216.3, down 109.8, sham 131.5; one pulse at +-25%: 0.92)
      sum of |moves| up 729.7 (SD 144.9, 499.8 to 1096.6), down 589.4 (SD 219.3, 300.2 to 1408.8), sham 659.5 (SD 160.4, 409.0 to 1251.2); largest move up 0.073, down 0.085, sham 0.085
      first kick's share up 0.604, down 0.682, sham 0.730; median later kick (sum of squared moves) up 0.0051, down 0.0029
      each kick's sham-to-arms ratio: sum of squared moves median 1.000 (0.961 to 1.040); sum of |moves| median 1.000 (0.974 to 1.021)
  lambda = 0.1, 100-unit hold
    history: dM = 0.00476 (SD 0.00260, 0.00033 to 0.01078) against the planned 0.00655; dM / the natural SD of 100-unit history means (0.00702) = 0.68; block SDs over the moments: up [0.38], down [0.36], difference [0.25]
      ratio to target just before a kick: drift up -0.0525 (SD 0.0978, -0.2574 to 0.1521), down 0.1457 (SD 0.1396, -0.1085 to 0.3786); spread up 0.361, down 0.546; 5th to 95th percentile up [-0.577, 0.546], down [-0.61, 1.047]
    effort: 21 kicks; sum of squared moves up 2.69 (SD 2.62, 0.18 to 11.34), down 1.89 (SD 1.51, 0.15 to 5.81), sham 2.29 (SD 1.51, 0.29 to 6.56) (the first check's pinned hold: up 11.3, down 7.8, sham 8.3; one pulse at +-25%: 0.92)
      sum of |moves| up 182.5 (SD 71.3, 86.1 to 419.3), down 156.9 (SD 37.8, 76.0 to 224.1), sham 169.7 (SD 47.3, 98.7 to 314.8); largest move up 0.066, down 0.046, sham 0.066
      first kick's share up 0.919, down 0.908, sham 0.952; median later kick (sum of squared moves) up 0.0026, down 0.0021
      each kick's sham-to-arms ratio: sum of squared moves median 1.000 (0.978 to 1.018); sum of |moves| median 1.000 (0.985 to 1.013)
Reported separately (choosing lambda, old starts 13 to 15)
  lambda = 0.2, 500-unit hold
    history: dM = 0.00661 (SD 0.00149, 0.00286 to 0.00903) against the planned 0.00648; dM / the natural SD of 500-unit history means (0.00548) = 1.21; block SDs over the moments: up [0.41, 0.31, 0.31, 0.19, 0.25], down [0.43, 0.3, 0.23, 0.26, 0.2], difference [0.23, 0.22, 0.2, 0.09, 0.2]
      ratio to target just before a kick: drift up -0.0164 (SD 0.0280, -0.0741 to 0.0298), down 0.0323 (SD 0.0706, -0.0870 to 0.1967); spread up 0.366, down 0.507; 5th to 95th percentile up [-0.591, 0.604], down [-0.697, 0.942]
    effort: 101 kicks; sum of squared moves up 6.50 (SD 3.16, 2.86 to 12.43), down 5.78 (SD 3.51, 1.62 to 12.93), sham 6.14 (SD 1.81, 3.22 to 9.93) (the first check's pinned hold: up 216.3, down 109.8, sham 131.5; one pulse at +-25%: 0.92)
      sum of |moves| up 1285.2 (SD 215.8, 984.7 to 1639.0), down 900.9 (SD 212.2, 463.5 to 1160.9), sham 1093.1 (SD 170.6, 785.9 to 1322.3); largest move up 0.054, down 0.060, sham 0.060
      first kick's share up 0.309, down 0.568, sham 0.494; median later kick (sum of squared moves) up 0.0211, down 0.0081
      each kick's sham-to-arms ratio: sum of squared moves median 1.000 (0.966 to 1.042); sum of |moves| median 1.000 (0.981 to 1.026)
  lambda = 0.2, 100-unit hold
    history: dM = 0.00528 (SD 0.00260, 0.00084 to 0.01242) against the planned 0.00655; dM / the natural SD of 100-unit history means (0.00702) = 0.75; block SDs over the moments: up [0.28], down [0.3], difference [0.21]
      ratio to target just before a kick: drift up -0.0416 (SD 0.0825, -0.2259 to 0.1408), down 0.1138 (SD 0.1267, -0.1051 to 0.3253); spread up 0.364, down 0.544; 5th to 95th percentile up [-0.574, 0.564], down [-0.618, 1.022]
    effort: 21 kicks; sum of squared moves up 2.91 (SD 2.75, 0.33 to 12.20), down 2.03 (SD 1.50, 0.21 to 5.87), sham 2.47 (SD 1.59, 0.43 to 7.20) (the first check's pinned hold: up 11.3, down 7.8, sham 8.3; one pulse at +-25%: 0.92)
      sum of |moves| up 245.6 (SD 96.4, 128.6 to 564.0), down 203.6 (SD 42.9, 95.7 to 267.0), sham 224.6 (SD 61.3, 119.8 to 411.3); largest move up 0.066, down 0.046, sham 0.066
      first kick's share up 0.814, down 0.804, sham 0.868; median later kick (sum of squared moves) up 0.0085, down 0.0052
      each kick's sham-to-arms ratio: sum of squared moves median 1.000 (0.974 to 1.020); sum of |moves| median 1.000 (0.986 to 1.012)

Pieces: 13, 12:01 to 12:48 Halifax, 39 minutes of running (the plan said about 30 to 45 minutes in all, with a confirmation); tasks took up to about 190 s with seven running at once
Parts matching the run log: 96 of 96; one numerical environment in every piece: True (INTEL(R) XEON(R) PLATINUM 8581C CPU @ 2.30GHz, Python 3.11.14, glibc 2.40, numpy 2.4.3, BLAS threads: the library default (numpy loaded first))
Seen before the check ran (dryruns/hold_check_dryrun.txt; old start 1, one moment at T = 600, lambda = 0.10; counts toward nothing): 500-unit hold A_up 0.759, A_down 0.853, A_sham +0.025, up's 100-unit blocks [0.43, 0.84, 0.94, 0.86, 0.55]; 100-unit hold A_up 1.255, A_down 0.369. Both were carried out.
Changes to the report code after the run, reporting only (no criterion, candidate or number changed): before the report was first made, the plan's 'reported separately' items were added for the candidates tried, since no confirmation ran, and the 100-unit hold's verdict was worded as not judged; after it was first made, these last lines (pieces, parts, environment and the dry run's numbers) were added.

Verdict: the 500-unit hold under H2 is NOT FEASIBLE at +-25% as specified; the 100-unit hold was not judged, since the plan judges it on the confirmation, which did not run (its numbers on starts 13 to 15 are above, reported only).