Every cell of the grid on the definition partition only, at zero alpha, against the always-long baseline at the same geometry. Colour is trust, never profit. Nothing here is a verdict; a cell becomes a question only by the author's registration (M12.5).
grid 1716f9f1c506 · results f5b7a3470c77 · 38,976 cells · classifier 0434836a176a · computed on x86_64 Linux python 3.12.3 numpy 2.5.3 · Register record #9 (SurveyRecorded)
1 · The grid in words
grid-002, declared 2026-09-20, sha 1716f9f1c506; 1,920 families, 38,976 cells. A family is the cells that share everything but the swept geometry: the sweep a registered question would carry.
Mechanisms — every kind on the closed entry enum with a mechanism sentence:
breakout_high as a stop order, 4 parameterisations: lookback=1; lookback=5; lookback=20; lookback=50 · horizons multi_day, intraday
breakout_low as a limit order, 4 parameterisations: lookback=1; lookback=5; lookback=20; lookback=50 · horizons multi_day, intraday
close_above_ma as a market order, 4 parameterisations: lookback=10; lookback=20; lookback=50; lookback=200 · horizons multi_day
close_below_ma as a market order, 4 parameterisations: lookback=10; lookback=20; lookback=50; lookback=200 · horizons multi_day
down_run as a market order, 3 parameterisations: runs=2; runs=3; runs=4 · horizons multi_day
return_below as a market order, 4 parameterisations: lookback=1, percent=2; lookback=1, percent=3; lookback=5, percent=5; lookback=10, percent=8 · horizons multi_day
pullback_in_trend as a market order, 3 parameterisations: short=5, long=50; short=10, long=100; short=20, long=200 · horizons multi_day
Regimes: none, up, down, ranging (none is no gate; the rest read one frozen classifier). Holds 1, 3, 5, 10, 20; intraday: holds 1, targets 2R, 3R. Stops percent 2, 3, 5 · atr 1.5, 2, 3 over 14 bars. Targets none, 2R, 3R.
Universes, each a Register record with its calendar frozen:
index_etfs — a liquid US index ETF; regime read on SPY
sector_etfs — a US sector ETF; regime read on SPY
dow_30 — a Dow 30 constituent as listed on 2026-09-12; regime read on SPY
sp_100 — an S&P 100 constituent as listed on 2026-09-12; regime read on SPY
2 · Mechanism × universe, by trust
held EV and margin positive, positive in every era and with every name held out, ≥100 trades · positive EV and margin positive over ≥100 trades · margin only the margin over always-long is positive · refused the run was refused by name — a family with no auditor, or the whole-run refusal that preceded ADR-0041
mechanism · order
index_etfs
sector_etfs
dow_30
sp_100
breakout_high(lookback=1) · stop
heldbest margin +0.105 · 4 held
heldbest margin +0.041 · 1 held
positivebest margin +0.007 · 0 held
positivebest margin +0.007 · 0 held
breakout_high(lookback=5) · stop
positivebest margin +0.118 · 0 held
heldbest margin +0.073 · 3 held
heldbest margin +0.021 · 8 held
heldbest margin +0.015 · 6 held
breakout_high(lookback=20) · stop
positivebest margin +0.254 · 0 held
positivebest margin +0.071 · 0 held
heldbest margin +0.034 · 3 held
heldbest margin +0.021 · 6 held
breakout_high(lookback=50) · stop
positivebest margin +0.227 · 0 held
positivebest margin +0.106 · 0 held
heldbest margin +0.039 · 3 held
heldbest margin +0.014 · 3 held
breakout_low(lookback=1) · limit
heldbest margin +0.257 · 34 held
heldbest margin +0.106 · 2 held
heldbest margin +0.154 · 39 held
heldbest margin +0.142 · 25 held
breakout_low(lookback=5) · limit
heldbest margin +0.338 · 17 held
heldbest margin +0.131 · 32 held
heldbest margin +0.216 · 109 held
heldbest margin +0.218 · 143 held
breakout_low(lookback=20) · limit
heldbest margin +0.504 · 60 held
heldbest margin +0.210 · 67 held
heldbest margin +0.270 · 69 held
heldbest margin +0.260 · 146 held
breakout_low(lookback=50) · limit
heldbest margin +0.670 · 15 held
heldbest margin +0.245 · 53 held
heldbest margin +0.371 · 196 held
heldbest margin +0.315 · 228 held
close_above_ma(lookback=10) · market
positivebest margin +0.185 · 0 held
heldbest margin +0.078 · 3 held
heldbest margin +0.084 · 24 held
heldbest margin +0.017 · 1 held
close_above_ma(lookback=20) · market
heldbest margin +0.291 · 4 held
positivebest margin +0.087 · 0 held
heldbest margin +0.080 · 51 held
heldbest margin +0.027 · 3 held
close_above_ma(lookback=50) · market
heldbest margin +0.336 · 5 held
positivebest margin +0.054 · 0 held
heldbest margin +0.066 · 36 held
heldbest margin +0.047 · 26 held
close_above_ma(lookback=200) · market
positivebest margin +0.292 · 0 held
positivebest margin +0.050 · 0 held
heldbest margin +0.035 · 1 held
positivebest margin +0.020 · 0 held
close_below_ma(lookback=10) · market
heldbest margin +0.102 · 13 held
heldbest margin +0.083 · 21 held
heldbest margin +0.071 · 12 held
heldbest margin +0.069 · 64 held
close_below_ma(lookback=20) · market
heldbest margin +0.054 · 2 held
heldbest margin +0.135 · 52 held
heldbest margin +0.064 · 2 held
heldbest margin +0.059 · 35 held
close_below_ma(lookback=50) · market
positivebest margin +0.095 · 0 held
heldbest margin +0.066 · 21 held
heldbest margin +0.085 · 4 held
heldbest margin +0.050 · 12 held
close_below_ma(lookback=200) · market
heldbest margin +0.037 · 3 held
heldbest margin +0.187 · 29 held
heldbest margin +0.038 · 15 held
heldbest margin +0.047 · 25 held
down_run(runs=2) · market
heldbest margin +0.120 · 4 held
heldbest margin +0.111 · 27 held
heldbest margin +0.093 · 30 held
heldbest margin +0.095 · 93 held
down_run(runs=3) · market
heldbest margin +0.466 · 9 held
heldbest margin +0.268 · 75 held
heldbest margin +0.156 · 135 held
heldbest margin +0.162 · 155 held
down_run(runs=4) · market
heldbest margin +0.917 · 11 held
heldbest margin +0.294 · 33 held
heldbest margin +0.231 · 178 held
heldbest margin +0.256 · 203 held
return_below(lookback=1, percent=2) · market
heldbest margin +0.197 · 12 held
heldbest margin +0.188 · 8 held
heldbest margin +0.077 · 65 held
heldbest margin +0.081 · 80 held
return_below(lookback=1, percent=3) · market
heldbest margin +0.552 · 62 held
heldbest margin +0.760 · 17 held
heldbest margin +0.071 · 54 held
heldbest margin +0.119 · 83 held
return_below(lookback=5, percent=5) · market
heldbest margin +0.569 · 18 held
heldbest margin +0.287 · 39 held
heldbest margin +0.210 · 82 held
heldbest margin +0.144 · 175 held
return_below(lookback=10, percent=8) · market
positivebest margin +0.850 · 0 held
heldbest margin +0.472 · 17 held
heldbest margin +0.128 · 79 held
heldbest margin +0.130 · 143 held
pullback_in_trend(long=50, short=5) · market
positivebest margin +1.029 · 0 held
heldbest margin +0.170 · 5 held
heldbest margin +0.151 · 80 held
heldbest margin +0.108 · 85 held
pullback_in_trend(long=100, short=10) · market
positivebest margin +0.338 · 0 held
heldbest margin +0.123 · 19 held
heldbest margin +0.130 · 45 held
heldbest margin +0.129 · 86 held
pullback_in_trend(long=200, short=20) · market
positivebest margin +0.192 · 0 held
heldbest margin +0.227 · 37 held
positivebest margin +0.069 · 0 held
heldbest margin +0.105 · 12 held
Each cell is the best tier the mechanism reached in that universe across regimes and geometry, its best margin over always-long in net R, and how many of its cells are held. Refused means the fill auditor (M5.5) refused every cell.
3 · index_etfs
4 names on the definition partition, 9,744 cells, 273 held; 6,140 days of measurement partition untouched.
Top 20 by margin over always-long
tier
regime
mechanism
geometry
trades
EV net R
always-long
margin
eras
min, one name out
N eff, measurement
lowest floor, k=4
margin
down
pullback_in_trend(long=50, short=5) · market
multi_day · hold 20 · stop percent 2 · no target
49
+0.852
-0.178
+1.029
+0.510 / +1.325 / +0.793
+0.785
137
—
margin
up
down_run(runs=4) · market
multi_day · hold 10 · stop percent 2 · no target
63
+0.750
-0.167
+0.917
+1.238 / +0.914 / +0.442
+0.590
165
—
margin
up
return_below(lookback=10, percent=8) · market
multi_day · hold 10 · stop percent 2 · no target
28
+0.684
-0.167
+0.850
— / +1.682 / +0.517
+0.509
81
—
margin
down
pullback_in_trend(long=50, short=5) · market
multi_day · hold 20 · stop percent 3 · no target
42
+0.672
-0.126
+0.798
+0.340 / +1.472 / +0.584
+0.587
107
—
margin
up
down_run(runs=4) · market
multi_day · hold 10 · stop percent 2 · target 3R
63
+0.640
-0.159
+0.798
+1.233 / +0.887 / +0.219
+0.448
151
—
margin
up
down_run(runs=4) · market
multi_day · hold 20 · stop percent 2 · target 3R
60
+0.662
-0.136
+0.797
+0.850 / +0.901 / +0.403
+0.524
149
—
margin
up
return_below(lookback=10, percent=8) · market
multi_day · hold 5 · stop percent 2 · no target
29
+0.557
-0.123
+0.680
— / +2.721 / +0.211
+0.348
79
—
margin
up
breakout_low(lookback=50) · limit
multi_day · hold 10 · stop percent 2 · no target
29
+0.503
-0.167
+0.670
+0.522 / +1.014 / +0.219
+0.273
74
—
margin
up
down_run(runs=4) · market
multi_day · hold 5 · stop percent 2 · no target
67
+0.544
-0.123
+0.667
+0.916 / +0.741 / +0.226
+0.362
166
—
margin
up
down_run(runs=4) · market
multi_day · hold 20 · stop percent 2 · no target
58
+0.484
-0.135
+0.619
+0.850 / +0.758 / +0.118
+0.215
167
—
margin
up
down_run(runs=4) · market
multi_day · hold 5 · stop percent 2 · target 3R
67
+0.485
-0.127
+0.613
+0.916 / +0.645 / +0.183
+0.315
167
—
margin
up
return_below(lookback=10, percent=8) · market
multi_day · hold 10 · stop percent 3 · no target
24
+0.534
-0.079
+0.612
— / +1.393 / +0.362
+0.355
69
—
margin
down
pullback_in_trend(long=50, short=5) · market
multi_day · hold 20 · stop percent 3 · target 3R
43
+0.464
-0.131
+0.595
+0.340 / +1.472 / +0.331
+0.382
107
—
margin
up
down_run(runs=4) · market
multi_day · hold 20 · stop percent 2 · target 2R
61
+0.448
-0.144
+0.592
+0.992 / +0.651 / +0.099
+0.205
165
—
margin
down
pullback_in_trend(long=50, short=5) · market
multi_day · hold 20 · stop atr 1.5 · no target
37
+0.492
-0.094
+0.586
+0.249 / +0.695 / +0.482
+0.405
86
—
margin
down
pullback_in_trend(long=50, short=5) · market
multi_day · hold 20 · stop percent 5 · no target
38
+0.505
-0.078
+0.583
+0.204 / +1.086 / +0.438
+0.356
93
—
margin
up
down_run(runs=4) · market
multi_day · hold 10 · stop atr 1.5 · target 3R
60
+0.537
-0.043
+0.581
+1.801 / +0.713 / -0.100
+0.181
150
—
margin
ranging
return_below(lookback=5, percent=5) · market
multi_day · hold 20 · stop percent 3 · target 3R
26
+0.694
+0.125
+0.569
+0.434 / -1.034 / +0.856
+0.633
77
—
margin
up
down_run(runs=4) · market
multi_day · hold 10 · stop percent 2 · target 2R
63
+0.428
-0.140
+0.568
+1.158 / +0.663 / -0.025
+0.222
153
—
margin
ranging
pullback_in_trend(long=50, short=5) · market
multi_day · hold 20 · stop percent 2 · target 3R
70
+0.507
-0.057
+0.564
+0.437 / +1.032 / +0.315
+0.410
184
—
Cells, in full
1 · index_etfs · up · breakout_low(lookback=20) · limit held
A buy-limit at the lowest low of the prior 20 bars, held for 10 bars with no regime gate, carries positive EV in net R on a liquid US index ETF because a fall to a new low is over-sold and bought back (reversal at a new low) — inside the up regime of the classifier frozen on SPY (F11)
105 trades on the definition partition; EV gross +0.397, net +0.346 at bounded costs; always-long at this geometry -0.159 over 505; margin +0.504. 4 names traded; weakest IWM -0.143 over 9, strongest SPY +0.463 over 49. Every name held out: at worst +0.243.
era 1 1993-01-29 → 1996-06-10 10 trades · EV +0.510 without it +0.329
era 2 1996-06-10 → 1999-10-21 35 trades · EV +0.688 without it +0.175
era 3 1999-10-21 → 2003-03-02 60 trades · EV +0.119 without it +0.649
Same-day ρ 0.24, design effect 1.07; the measurement partition would afford about 301 effective trades at this firing rate; at the cell's own σ (1.55) and k = 4, 0.05 R needs 14,316, 0.075 R needs 6,363, 0.1 R needs 3,579, 0.15 R needs 1,591, 0.2 R needs 895, 0.25 R needs 573.
Registering it: not registrable as it stands: no floor in the table is affordable on the measurement partition. The question would carry the family's sweep of 30 cells; the accountant prices it at 0.3000 alpha on the regime axis (rate × 30 cells).
2 · index_etfs · up · breakout_low(lookback=20) · limit held
A buy-limit at the lowest low of the prior 20 bars, held for 10 bars with no regime gate, carries positive EV in net R on a liquid US index ETF because a fall to a new low is over-sold and bought back (reversal at a new low) — inside the up regime of the classifier frozen on SPY (F11)
107 trades on the definition partition; EV gross +0.317, net +0.266 at bounded costs; always-long at this geometry -0.140 over 553; margin +0.406. 4 names traded; weakest IWM -0.196 over 9, strongest SPY +0.407 over 50. Every name held out: at worst +0.143.
era 1 1993-01-29 → 1996-06-10 10 trades · EV +0.560 without it +0.236
era 2 1996-06-10 → 1999-10-21 36 trades · EV +0.597 without it +0.099
era 3 1999-10-21 → 2003-03-02 61 trades · EV +0.023 without it +0.589
Same-day ρ 0.35, design effect 1.10; the measurement partition would afford about 299 effective trades at this firing rate; at the cell's own σ (1.34) and k = 4, 0.05 R needs 10,744, 0.075 R needs 4,775, 0.1 R needs 2,686, 0.15 R needs 1,194, 0.2 R needs 672, 0.25 R needs 430.
Registering it: not registrable as it stands: no floor in the table is affordable on the measurement partition. The question would carry the family's sweep of 30 cells; the accountant prices it at 0.3000 alpha on the regime axis (rate × 30 cells).
3 · index_etfs · up · down_run(runs=3) · market held
After 3 consecutive lower closes, a liquid US index ETF carries positive EV in net R over the next 20 bars because a streak of declines exhausts short-horizon sellers and the next days correct (reversal after a streak; ADR-0037) — inside the up regime of the classifier frozen on SPY (F11)
140 trades on the definition partition; EV gross +0.317, net +0.266 at bounded costs; always-long at this geometry -0.136 over 418; margin +0.401. 4 names traded; weakest QQQ +0.110 over 31, strongest IWM +0.504 over 8. Every name held out: at worst +0.170.
era 1 1993-01-29 → 1996-06-10 17 trades · EV +0.469 without it +0.238
era 2 1996-06-10 → 1999-10-21 54 trades · EV +0.434 without it +0.160
era 3 1999-10-21 → 2003-03-02 69 trades · EV +0.084 without it +0.443
Same-day ρ 0.46, design effect 1.17; the measurement partition would afford about 367 effective trades at this firing rate; at the cell's own σ (1.72) and k = 4, 0.05 R needs 17,688, 0.075 R needs 7,861, 0.1 R needs 4,422, 0.15 R needs 1,966, 0.2 R needs 1,106, 0.25 R needs 708.
Registering it: not registrable as it stands: no floor in the table is affordable on the measurement partition. The question would carry the family's sweep of 30 cells; the accountant prices it at 0.3000 alpha on the regime axis (rate × 30 cells).
4 · index_etfs · up · breakout_low(lookback=20) · limit held
A buy-limit at the lowest low of the prior 20 bars, held for 3 bars with no regime gate, carries positive EV in net R on a liquid US index ETF because a fall to a new low is over-sold and bought back (reversal at a new low) — inside the up regime of the classifier frozen on SPY (F11)
136 trades on the definition partition; EV gross +0.284, net +0.233 at bounded costs; always-long at this geometry -0.108 over 1,010; margin +0.341. 4 names traded; weakest IWM -0.109 over 11, strongest QQQ +0.473 over 27. Every name held out: at worst +0.173.
era 1 1993-01-29 → 1996-06-10 13 trades · EV +0.182 without it +0.238
era 2 1996-06-10 → 1999-10-21 47 trades · EV +0.281 without it +0.207
era 3 1999-10-21 → 2003-03-02 76 trades · EV +0.212 without it +0.259
Same-day ρ 0.35, design effect 1.13; the measurement partition would afford about 370 effective trades at this firing rate; at the cell's own σ (0.99) and k = 4, 0.05 R needs 5,841, 0.075 R needs 2,596, 0.1 R needs 1,461, 0.15 R needs 649, 0.2 R needs 366, 0.25 R needs 234.
Registering it: registrable in principle: three names or more, the sweep holds the plateau, and the measurement partition affords a floor of 0.2 R at k = 4. The question would carry the family's sweep of 30 cells; the accountant prices it at 0.3000 alpha on the regime axis (rate × 30 cells).
4 · sector_etfs
9 names on the definition partition, 9,744 cells, 561 held; 5,063 days of measurement partition untouched.
Missed trades: the fill auditor refused an entry in 304 cells, 482 entries in all, not opened and counted beside the trade count (ADR-0041) — by name, XLB 348, XLI 134; 5,152 in the always-long baselines.
Top 20 by margin over always-long
tier
regime
mechanism
geometry
trades
EV net R
always-long
margin
eras
min, one name out
N eff, measurement
lowest floor, k=4
margin
ranging
return_below(lookback=1, percent=3) · market
multi_day · hold 20 · stop percent 2 · no target
49
+0.780
+0.020
+0.760
+1.448 / +0.027 / -1.051
+0.530
94
—
margin
ranging
return_below(lookback=1, percent=3) · market
multi_day · hold 10 · stop percent 2 · no target
50
+0.484
-0.002
+0.485
+0.800 / +0.304 / -1.051
+0.298
86
—
positive
down
return_below(lookback=10, percent=8) · market
multi_day · hold 20 · stop percent 2 · no target
206
+0.551
+0.079
+0.472
+0.502 / +0.717 / -1.051
+0.466
248
—
margin
ranging
return_below(lookback=10, percent=8) · market
multi_day · hold 20 · stop percent 2 · no target
43
+0.484
+0.020
+0.464
+0.736 / +1.142 / -1.101
+0.225
85
—
margin
up
return_below(lookback=10, percent=8) · market
multi_day · hold 5 · stop percent 2 · no target
66
+0.394
-0.044
+0.437
+0.449 / +0.137 / +0.863
+0.352
116
—
margin
ranging
return_below(lookback=10, percent=8) · market
multi_day · hold 20 · stop percent 3 · no target
39
+0.388
-0.025
+0.413
+0.608 / +1.221 / -1.056
+0.208
69
—
margin
ranging
return_below(lookback=10, percent=8) · market
multi_day · hold 20 · stop percent 2 · target 3R
47
+0.437
+0.031
+0.406
+0.382 / +2.149 / -0.514
+0.357
91
—
margin
ranging
return_below(lookback=10, percent=8) · market
multi_day · hold 20 · stop atr 2 · target 3R
30
+0.380
-0.009
+0.389
+0.604 / +0.562 / -1.022
+0.236
61
—
margin
ranging
return_below(lookback=1, percent=3) · market
multi_day · hold 20 · stop percent 2 · target 3R
52
+0.414
+0.031
+0.383
+0.734 / +0.175 / -1.051
+0.340
103
—
margin
ranging
return_below(lookback=1, percent=3) · market
multi_day · hold 20 · stop percent 3 · no target
49
+0.334
-0.025
+0.359
+0.698 / +0.014 / -1.034
+0.145
91
—
margin
ranging
return_below(lookback=10, percent=8) · market
multi_day · hold 20 · stop atr 2 · no target
30
+0.323
-0.022
+0.345
+0.522 / +0.562 / -1.022
+0.244
61
—
margin
up
return_below(lookback=10, percent=8) · market
multi_day · hold 5 · stop percent 3 · target 3R
60
+0.307
-0.037
+0.344
+0.354 / +0.144 / +0.492
+0.289
101
—
margin
ranging
return_below(lookback=1, percent=3) · market
multi_day · hold 10 · stop percent 2 · target 3R
52
+0.357
+0.014
+0.343
+0.665 / +0.127 / -1.051
+0.230
100
—
margin
up
return_below(lookback=10, percent=8) · market
multi_day · hold 5 · stop percent 3 · no target
60
+0.294
-0.039
+0.333
+0.334 / +0.144 / +0.492
+0.265
101
—
margin
ranging
return_below(lookback=1, percent=3) · market
multi_day · hold 20 · stop atr 1.5 · no target
44
+0.258
-0.048
+0.306
+0.550 / -0.027 / -1.044
+0.104
81
—
margin
ranging
return_below(lookback=10, percent=8) · market
multi_day · hold 20 · stop atr 2 · target 2R
30
+0.328
+0.024
+0.305
+0.500 / +0.687 / -1.022
+0.213
61
—
positive
none
return_below(lookback=10, percent=8) · market
multi_day · hold 20 · stop percent 2 · no target
300
+0.326
+0.026
+0.300
+0.281 / +0.524 / -1.079
+0.267
440
—
margin
up
return_below(lookback=10, percent=8) · market
multi_day · hold 5 · stop percent 2 · target 3R
67
+0.251
-0.046
+0.297
+0.240 / +0.137 / +0.863
+0.207
132
—
held
down
down_run(runs=4) · market
multi_day · hold 20 · stop atr 1.5 · target 3R
140
+0.340
+0.046
+0.294
+0.245 / +0.370 / +0.607
+0.306
205
—
margin
ranging
return_below(lookback=10, percent=8) · market
multi_day · hold 20 · stop percent 3 · target 2R
42
+0.295
+0.003
+0.293
+0.327 / +1.366 / -0.605
+0.227
79
—
Cells, in full
1 · sector_etfs · down · down_run(runs=4) · market held
After 4 consecutive lower closes, a US sector ETF carries positive EV in net R over the next 20 bars because a streak of declines exhausts short-horizon sellers and the next days correct (reversal after a streak; ADR-0037) — inside the down regime of the classifier frozen on SPY (F11)
140 trades on the definition partition; EV gross +0.369, net +0.340 at bounded costs; always-long at this geometry +0.046 over 601; margin +0.294. 9 names traded; weakest XLU -0.021 over 14, strongest XLP +0.623 over 12. Every name held out: at worst +0.306.
era 1 1998-12-22 → 2001-09-29 58 trades · EV +0.245 without it +0.408
era 2 2001-09-29 → 2004-07-08 69 trades · EV +0.370 without it +0.311
era 3 2004-07-08 → 2007-04-17 13 trades · EV +0.607 without it +0.313
Same-day ρ 0.64, design effect 1.39; the measurement partition would afford about 205 effective trades at this firing rate; at the cell's own σ (1.75) and k = 4, 0.05 R needs 18,228, 0.075 R needs 8,102, 0.1 R needs 4,557, 0.15 R needs 2,026, 0.2 R needs 1,140, 0.25 R needs 730.
Registering it: not registrable as it stands: no floor in the table is affordable on the measurement partition. The question would carry the family's sweep of 30 cells; the accountant prices it at 0.3000 alpha on the regime axis (rate × 30 cells).
2 · sector_etfs · down · return_below(lookback=5, percent=5) · market held
After a decline of at least 5 % over 5 bars, a US sector ETF carries positive EV in net R over the next 20 bars because a fall of that size over-corrects (reversal after a decline of size; ADR-0037) — inside the down regime of the classifier frozen on SPY (F11)
cell 7bb79763dcaf0cd4 · spec 7de742e53d12 · multi_day · hold 20 · stop percent 2 · no target · axis regime
287 trades on the definition partition; EV gross +0.417, net +0.366 at bounded costs; always-long at this geometry +0.079 over 801; margin +0.287. 9 names traded; weakest XLK -0.121 over 76, strongest XLP +1.221 over 15. Every name held out: at worst +0.298.
era 1 1998-12-22 → 2001-09-29 139 trades · EV +0.187 without it +0.534
era 2 2001-09-29 → 2004-07-08 130 trades · EV +0.407 without it +0.332
era 3 2004-07-08 → 2007-04-17 18 trades · EV +1.454 without it +0.293
Same-day ρ 0.47, design effect 1.45; the measurement partition would afford about 402 effective trades at this firing rate; at the cell's own σ (3.38) and k = 4, 0.05 R needs 68,460, 0.075 R needs 30,427, 0.1 R needs 17,115, 0.15 R needs 7,607, 0.2 R needs 4,279, 0.25 R needs 2,739.
Registering it: not registrable as it stands: no floor in the table is affordable on the measurement partition. The question would carry the family's sweep of 15 cells; the accountant prices it at 0.1500 alpha on the regime axis (rate × 15 cells).
3 · sector_etfs · down · down_run(runs=3) · market held
After 3 consecutive lower closes, a US sector ETF carries positive EV in net R over the next 20 bars because a streak of declines exhausts short-horizon sellers and the next days correct (reversal after a streak; ADR-0037) — inside the down regime of the classifier frozen on SPY (F11)
cell 767ae04fffd6579b · spec 2f87f53ecc0a · multi_day · hold 20 · stop percent 2 · no target · axis regime
341 trades on the definition partition; EV gross +0.398, net +0.347 at bounded costs; always-long at this geometry +0.079 over 801; margin +0.268. 9 names traded; weakest XLK -0.123 over 58, strongest XLF +1.131 over 37. Every name held out: at worst +0.252.
era 1 1998-12-22 → 2001-09-29 155 trades · EV +0.272 without it +0.409
era 2 2001-09-29 → 2004-07-08 142 trades · EV +0.095 without it +0.527
era 3 2004-07-08 → 2007-04-17 44 trades · EV +1.423 without it +0.188
Same-day ρ 0.79, design effect 1.71; the measurement partition would afford about 406 effective trades at this firing rate; at the cell's own σ (2.88) and k = 4, 0.05 R needs 49,431, 0.075 R needs 21,970, 0.1 R needs 12,358, 0.15 R needs 5,493, 0.2 R needs 3,090, 0.25 R needs 1,978.
Registering it: not registrable as it stands: no floor in the table is affordable on the measurement partition. The question would carry the family's sweep of 15 cells; the accountant prices it at 0.1500 alpha on the regime axis (rate × 15 cells).
4 · sector_etfs · down · down_run(runs=3) · market held
After 3 consecutive lower closes, a US sector ETF carries positive EV in net R over the next 20 bars because a streak of declines exhausts short-horizon sellers and the next days correct (reversal after a streak; ADR-0037) — inside the down regime of the classifier frozen on SPY (F11)
313 trades on the definition partition; EV gross +0.296, net +0.262 at bounded costs; always-long at this geometry -0.000 over 764; margin +0.263. 9 names traded; weakest XLU -0.009 over 32, strongest XLV +0.451 over 31. Every name held out: at worst +0.242.
era 1 1998-12-22 → 2001-09-29 138 trades · EV +0.158 without it +0.345
era 2 2001-09-29 → 2004-07-08 133 trades · EV +0.172 without it +0.329
era 3 2004-07-08 → 2007-04-17 42 trades · EV +0.892 without it +0.165
Same-day ρ 0.49, design effect 1.39; the measurement partition would afford about 459 effective trades at this firing rate; at the cell's own σ (1.73) and k = 4, 0.05 R needs 17,783, 0.075 R needs 7,904, 0.1 R needs 4,446, 0.15 R needs 1,976, 0.2 R needs 1,112, 0.25 R needs 712.
Registering it: not registrable as it stands: no floor in the table is affordable on the measurement partition. The question would carry the family's sweep of 30 cells; the accountant prices it at 0.3000 alpha on the regime axis (rate × 30 cells).
5 · dow_30
27 names on the definition partition, 9,744 cells, 1,320 held; 6,152 days of measurement partition untouched.
Missed trades: the fill auditor refused an entry in 3,157 cells, 5,143 entries in all, not opened and counted beside the trade count (ADR-0041) — by name, DIS 1,484, SHW 1,408, AMGN 802, CVX 728, PG 351, CAT 244, KO 126; 18,246 in the always-long baselines.
Top 20 by margin over always-long
tier
regime
mechanism
geometry
trades
EV net R
always-long
margin
eras
min, one name out
N eff, measurement
lowest floor, k=4
held
up
breakout_low(lookback=50) · limit
multi_day · hold 20 · stop percent 2 · target 2R
869
+0.296
-0.075
+0.371
+0.364 / +0.294 / +0.248
+0.275
1,600
0.15 R
held
up
breakout_low(lookback=50) · limit
multi_day · hold 10 · stop percent 2 · target 2R
875
+0.294
-0.073
+0.366
+0.346 / +0.290 / +0.260
+0.275
1,614
0.15 R
held
up
breakout_low(lookback=50) · limit
multi_day · hold 5 · stop percent 2 · target 2R
900
+0.257
-0.069
+0.326
+0.261 / +0.226 / +0.298
+0.240
1,671
0.15 R
held
ranging
breakout_low(lookback=20) · limit
multi_day · hold 20 · stop percent 2 · target 3R
681
+0.274
+0.004
+0.270
+0.351 / +0.596 / +0.059
+0.249
1,051
0.25 R
held
up
breakout_low(lookback=50) · limit
multi_day · hold 3 · stop percent 2 · target 2R
955
+0.197
-0.069
+0.266
+0.181 / +0.178 / +0.238
+0.176
1,742
0.15 R
held
up
breakout_low(lookback=50) · limit
multi_day · hold 5 · stop percent 2 · target 3R
843
+0.218
-0.039
+0.256
+0.239 / +0.242 / +0.166
+0.197
1,514
0.2 R
held
none
breakout_low(lookback=50) · limit
multi_day · hold 20 · stop percent 2 · target 2R
2,201
+0.168
-0.087
+0.254
+0.302 / +0.206 / +0.103
+0.154
3,892
0.1 R
held
up
breakout_low(lookback=50) · limit
multi_day · hold 10 · stop percent 2 · target 3R
793
+0.217
-0.036
+0.253
+0.281 / +0.298 / +0.051
+0.196
1,409
0.2 R
held
none
breakout_low(lookback=50) · limit
multi_day · hold 10 · stop percent 2 · target 2R
2,208
+0.165
-0.086
+0.251
+0.281 / +0.203 / +0.106
+0.152
3,909
0.1 R
held
none
breakout_low(lookback=50) · limit
multi_day · hold 5 · stop percent 2 · target 2R
2,260
+0.155
-0.080
+0.235
+0.254 / +0.156 / +0.119
+0.142
3,859
0.1 R
held
ranging
breakout_low(lookback=20) · limit
multi_day · hold 5 · stop percent 2 · target 3R
739
+0.224
-0.007
+0.231
+0.297 / +0.361 / +0.096
+0.208
1,103
0.2 R
positive
up
down_run(runs=4) · market
multi_day · hold 20 · stop percent 2 · target 3R
1,229 · 1 missed
+0.197
-0.034
+0.231
+0.464 / +0.202 / -0.032
+0.169
2,024
0.2 R
held
up
breakout_low(lookback=50) · limit
multi_day · hold 20 · stop percent 3 · target 2R
673
+0.217
-0.012
+0.230
+0.400 / +0.240 / +0.045
+0.186
1,211
0.2 R
positive
none
down_run(runs=4) · market
multi_day · hold 20 · stop percent 2 · target 3R
2,345 · 1 missed
+0.168
-0.061
+0.229
+0.391 / +0.221 / -0.007
+0.148
3,900
0.15 R
held
up
breakout_low(lookback=20) · limit
multi_day · hold 20 · stop percent 2 · target 2R
1,905
+0.153
-0.075
+0.228
+0.228 / +0.150 / +0.102
+0.135
3,151
0.15 R
held
up
breakout_low(lookback=20) · limit
multi_day · hold 10 · stop percent 2 · target 2R
1,918
+0.155
-0.073
+0.228
+0.228 / +0.155 / +0.100
+0.137
3,215
0.15 R
held
up
breakout_low(lookback=50) · limit
multi_day · hold 20 · stop percent 2 · target 3R
770
+0.194
-0.034
+0.228
+0.260 / +0.274 / +0.033
+0.168
1,361
0.2 R
held
up
breakout_low(lookback=50) · limit
multi_day · hold 5 · stop percent 3 · target 2R
758
+0.191
-0.026
+0.217
+0.297 / +0.176 / +0.122
+0.172
1,426
0.15 R
held
down
breakout_low(lookback=5) · limit
multi_day · hold 20 · stop percent 2 · target 2R
2,421
+0.111
-0.105
+0.216
+0.385 / +0.127 / +0.077
+0.099
3,126
0.15 R
held
down
breakout_low(lookback=5) · limit
multi_day · hold 10 · stop percent 2 · target 2R
2,432
+0.110
-0.105
+0.215
+0.370 / +0.114 / +0.079
+0.098
3,187
0.15 R
Cells, in full
1 · dow_30 · up · breakout_low(lookback=50) · limit held
A buy-limit at the lowest low of the prior 50 bars, held for 20 bars with no regime gate, carries positive EV in net R on a Dow 30 constituent as listed on 2026-09-12 because a fall to a new low is over-sold and bought back (reversal at a new low) — inside the up regime of the classifier frozen on SPY (F11)
869 trades on the definition partition; EV gross +0.347, net +0.296 at bounded costs; always-long at this geometry -0.075 over 12,140; margin +0.371. 27 names traded; weakest AMZN -0.384 over 18, strongest GS +0.949 over 9. Every name held out: at worst +0.275.
era 1 1993-01-04 → 1996-05-18 203 trades · EV +0.364 without it +0.275
era 2 1996-05-18 → 1999-09-30 388 trades · EV +0.294 without it +0.297
era 3 1999-09-30 → 2003-02-12 278 trades · EV +0.248 without it +0.318
Same-day ρ 0.09, design effect 1.07; the measurement partition would afford about 1,600 effective trades at this firing rate; at the cell's own σ (1.51) and k = 4, 0.05 R needs 13,553, 0.075 R needs 6,024, 0.1 R needs 3,389, 0.15 R needs 1,506, 0.2 R needs 848, 0.25 R needs 543.
Registering it: registrable in principle: three names or more, the sweep holds the plateau, and the measurement partition affords a floor of 0.15 R at k = 4. The question would carry the family's sweep of 30 cells; the accountant prices it at 0.3000 alpha on the regime axis (rate × 30 cells).
2 · dow_30 · up · breakout_low(lookback=50) · limit held
A buy-limit at the lowest low of the prior 50 bars, held for 10 bars with no regime gate, carries positive EV in net R on a Dow 30 constituent as listed on 2026-09-12 because a fall to a new low is over-sold and bought back (reversal at a new low) — inside the up regime of the classifier frozen on SPY (F11)
875 trades on the definition partition; EV gross +0.345, net +0.294 at bounded costs; always-long at this geometry -0.073 over 12,466; margin +0.366. 27 names traded; weakest AMZN -0.384 over 18, strongest GS +0.949 over 9. Every name held out: at worst +0.275.
era 1 1993-01-04 → 1996-05-18 206 trades · EV +0.346 without it +0.277
era 2 1996-05-18 → 1999-09-30 390 trades · EV +0.290 without it +0.297
era 3 1999-09-30 → 2003-02-12 279 trades · EV +0.260 without it +0.309
Same-day ρ 0.09, design effect 1.07; the measurement partition would afford about 1,614 effective trades at this firing rate; at the cell's own σ (1.49) and k = 4, 0.05 R needs 13,324, 0.075 R needs 5,922, 0.1 R needs 3,331, 0.15 R needs 1,481, 0.2 R needs 833, 0.25 R needs 533.
Registering it: registrable in principle: three names or more, the sweep holds the plateau, and the measurement partition affords a floor of 0.15 R at k = 4. The question would carry the family's sweep of 30 cells; the accountant prices it at 0.3000 alpha on the regime axis (rate × 30 cells).
3 · dow_30 · up · breakout_low(lookback=50) · limit held
A buy-limit at the lowest low of the prior 50 bars, held for 5 bars with no regime gate, carries positive EV in net R on a Dow 30 constituent as listed on 2026-09-12 because a fall to a new low is over-sold and bought back (reversal at a new low) — inside the up regime of the classifier frozen on SPY (F11)
900 trades on the definition partition; EV gross +0.308, net +0.257 at bounded costs; always-long at this geometry -0.069 over 13,931; margin +0.326. 27 names traded; weakest AMZN -0.384 over 18, strongest GS +0.877 over 9. Every name held out: at worst +0.240.
era 1 1993-01-04 → 1996-05-18 217 trades · EV +0.261 without it +0.256
era 2 1996-05-18 → 1999-09-30 399 trades · EV +0.226 without it +0.282
era 3 1999-09-30 → 2003-02-12 284 trades · EV +0.298 without it +0.238
Same-day ρ 0.08, design effect 1.06; the measurement partition would afford about 1,671 effective trades at this firing rate; at the cell's own σ (1.44) and k = 4, 0.05 R needs 12,326, 0.075 R needs 5,479, 0.1 R needs 3,082, 0.15 R needs 1,370, 0.2 R needs 771, 0.25 R needs 494.
Registering it: registrable in principle: three names or more, the sweep holds the plateau, and the measurement partition affords a floor of 0.15 R at k = 4. The question would carry the family's sweep of 30 cells; the accountant prices it at 0.3000 alpha on the regime axis (rate × 30 cells).
4 · dow_30 · ranging · breakout_low(lookback=20) · limit held
A buy-limit at the lowest low of the prior 20 bars, held for 20 bars with no regime gate, carries positive EV in net R on a Dow 30 constituent as listed on 2026-09-12 because a fall to a new low is over-sold and bought back (reversal at a new low) — inside the ranging regime of the classifier frozen on SPY (F11)
681 trades on the definition partition; EV gross +0.325, net +0.274 at bounded costs; always-long at this geometry +0.004 over 3,743; margin +0.270. 27 names traded; weakest PG -0.682 over 18, strongest AXP +1.044 over 21. Every name held out: at worst +0.249.
era 1 1993-01-04 → 1996-05-18 228 trades · EV +0.351 without it +0.235
era 2 1996-05-18 → 1999-09-30 148 trades · EV +0.596 without it +0.184
era 3 1999-09-30 → 2003-02-12 305 trades · EV +0.059 without it +0.448
Same-day ρ 0.18, design effect 1.28; the measurement partition would afford about 1,051 effective trades at this firing rate; at the cell's own σ (2.00) and k = 4, 0.05 R needs 23,850, 0.075 R needs 10,600, 0.1 R needs 5,963, 0.15 R needs 2,650, 0.2 R needs 1,491, 0.25 R needs 954.
Registering it: registrable in principle: three names or more, the sweep holds the plateau, and the measurement partition affords a floor of 0.25 R at k = 4. The question would carry the family's sweep of 30 cells; the accountant prices it at 0.3000 alpha on the regime axis (rate × 30 cells).
6 · sp_100
83 names on the definition partition, 9,744 cells, 1,838 held; 6,152 days of measurement partition untouched.
Missed trades: the fill auditor refused an entry in 8,888 cells, 383,551 entries in all, not opened and counted beside the trade count (ADR-0041) — by name, C 351,080, AMD 8,052, DHR 3,202, GILD 2,688, COF 1,648, MO 1,529, DIS 1,484, SPG 1,330; 1,365,750 in the always-long baselines.
Top 20 by margin over always-long
tier
regime
mechanism
geometry
trades
EV net R
always-long
margin
eras
min, one name out
N eff, measurement
lowest floor, k=4
held
up
breakout_low(lookback=50) · limit
multi_day · hold 20 · stop percent 2 · target 2R
2,695 · 6 missed
+0.230
-0.085
+0.315
+0.340 / +0.209 / +0.156
+0.222
4,905
0.1 R
held
up
breakout_low(lookback=50) · limit
multi_day · hold 10 · stop percent 2 · target 2R
2,706 · 6 missed
+0.227
-0.083
+0.309
+0.320 / +0.212 / +0.157
+0.219
4,941
0.1 R
held
up
breakout_low(lookback=50) · limit
multi_day · hold 5 · stop percent 2 · target 2R
2,775 · 6 missed
+0.206
-0.079
+0.285
+0.285 / +0.181 / +0.164
+0.199
5,038
0.1 R
held
up
breakout_low(lookback=50) · limit
multi_day · hold 20 · stop percent 2 · target 3R
2,421 · 6 missed
+0.226
-0.044
+0.271
+0.374 / +0.246 / +0.044
+0.217
4,455
0.15 R
held
none
breakout_low(lookback=50) · limit
multi_day · hold 20 · stop percent 2 · target 2R
6,596 · 6 missed
+0.173
-0.093
+0.266
+0.332 / +0.187 / +0.100
+0.167
11,135
0.075 R
held
up
breakout_low(lookback=50) · limit
multi_day · hold 10 · stop percent 2 · target 3R
2,464 · 6 missed
+0.216
-0.047
+0.263
+0.344 / +0.244 / +0.036
+0.206
4,508
0.15 R
held
none
breakout_low(lookback=50) · limit
multi_day · hold 10 · stop percent 2 · target 2R
6,621 · 6 missed
+0.168
-0.092
+0.260
+0.311 / +0.184 / +0.100
+0.162
11,141
0.075 R
held
up
breakout_low(lookback=20) · limit
multi_day · hold 20 · stop percent 2 · target 2R
5,897 · 16 missed
+0.175
-0.085
+0.260
+0.327 / +0.152 / +0.075
+0.166
8,840
0.075 R
held
up
breakout_low(lookback=50) · limit
multi_day · hold 3 · stop percent 2 · target 2R
2,935 · 5 missed
+0.184
-0.075
+0.259
+0.244 / +0.161 / +0.159
+0.177
5,154
0.075 R
held
up
breakout_low(lookback=20) · limit
multi_day · hold 10 · stop percent 2 · target 2R
5,945 · 16 missed
+0.174
-0.083
+0.256
+0.320 / +0.152 / +0.076
+0.165
8,907
0.075 R
positive
up
down_run(runs=4) · market
multi_day · hold 20 · stop percent 2 · target 3R
3,660 · 18 missed
+0.212
-0.044
+0.256
+0.374 / +0.289 / -0.031
+0.203
5,650
0.1 R
held
up
breakout_low(lookback=50) · limit
multi_day · hold 20 · stop percent 3 · target 2R
2,138 · 5 missed
+0.225
-0.025
+0.251
+0.365 / +0.251 / +0.039
+0.218
3,870
0.1 R
held
up
breakout_low(lookback=50) · limit
multi_day · hold 5 · stop percent 2 · target 3R
2,579 · 6 missed
+0.203
-0.045
+0.248
+0.278 / +0.232 / +0.071
+0.194
4,657
0.1 R
held
none
breakout_low(lookback=20) · limit
multi_day · hold 20 · stop percent 2 · target 2R
12,069 · 17 missed
+0.154
-0.093
+0.247
+0.299 / +0.155 / +0.080
+0.148
17,754
0.05 R
held
ranging
breakout_low(lookback=20) · limit
multi_day · hold 20 · stop percent 2 · target 3R
1,982 · 1 missed
+0.233
-0.010
+0.243
+0.242 / +0.516 / +0.056
+0.224
2,644
0.15 R
held
none
breakout_low(lookback=20) · limit
multi_day · hold 10 · stop percent 2 · target 2R
12,164 · 17 missed
+0.151
-0.092
+0.243
+0.287 / +0.154 / +0.079
+0.145
17,960
0.05 R
held
none
breakout_low(lookback=50) · limit
multi_day · hold 5 · stop percent 2 · target 2R
6,769 · 6 missed
+0.153
-0.087
+0.240
+0.271 / +0.158 / +0.099
+0.147
11,287
0.075 R
held
up
down_run(runs=4) · market
multi_day · hold 5 · stop percent 2 · no target
3,730 · 18 missed
+0.255
+0.016
+0.238
+0.254 / +0.386 / +0.048
+0.237
5,669
0.15 R
held
none
breakout_low(lookback=50) · limit
multi_day · hold 20 · stop percent 2 · target 3R
5,926 · 6 missed
+0.183
-0.054
+0.238
+0.353 / +0.271 / +0.064
+0.170
9,681
0.075 R
held
ranging
breakout_low(lookback=50) · limit
multi_day · hold 5 · stop percent 2 · target 3R
1,175
+0.217
-0.020
+0.237
+0.258 / +0.368 / +0.106
+0.202
1,747
0.2 R
Cells, in full
1 · sp_100 · up · breakout_low(lookback=50) · limit held
A buy-limit at the lowest low of the prior 50 bars, held for 20 bars with no regime gate, carries positive EV in net R on an S&P 100 constituent as listed on 2026-09-12 because a fall to a new low is over-sold and bought back (reversal at a new low) — inside the up regime of the classifier frozen on SPY (F11)
2,695 trades on the definition partition; EV gross +0.282, net +0.230 at bounded costs; always-long at this geometry -0.085 over 38,837; margin +0.315. 81 names traded; weakest MDLZ -4.039 over 2, strongest ACN +1.199 over 4. 6 entries missed — the fill auditor refused them and no position was opened (ADR-0041): C 5, DE 1. Every name held out: at worst +0.222.
era 1 1993-01-04 → 1996-05-18 738 trades · EV +0.340 without it +0.189
era 2 1996-05-18 → 1999-09-30 1,225 trades · EV +0.209 without it +0.248
era 3 1999-09-30 → 2003-02-12 732 trades · EV +0.156 without it +0.258
Same-day ρ 0.10, design effect 1.20; the measurement partition would afford about 4,905 effective trades at this firing rate; at the cell's own σ (1.51) and k = 4, 0.05 R needs 13,565, 0.075 R needs 6,029, 0.1 R needs 3,392, 0.15 R needs 1,508, 0.2 R needs 848, 0.25 R needs 543.
Registering it: registrable in principle: three names or more, the sweep holds the plateau, and the measurement partition affords a floor of 0.1 R at k = 4. The question would carry the family's sweep of 30 cells; the accountant prices it at 0.3000 alpha on the regime axis (rate × 30 cells).
2 · sp_100 · up · breakout_low(lookback=50) · limit held
A buy-limit at the lowest low of the prior 50 bars, held for 10 bars with no regime gate, carries positive EV in net R on an S&P 100 constituent as listed on 2026-09-12 because a fall to a new low is over-sold and bought back (reversal at a new low) — inside the up regime of the classifier frozen on SPY (F11)
2,706 trades on the definition partition; EV gross +0.278, net +0.227 at bounded costs; always-long at this geometry -0.083 over 39,774; margin +0.309. 81 names traded; weakest MDLZ -4.039 over 2, strongest ACN +1.199 over 4. 6 entries missed — the fill auditor refused them and no position was opened (ADR-0041): C 5, DE 1. Every name held out: at worst +0.219.
era 1 1993-01-04 → 1996-05-18 743 trades · EV +0.320 without it +0.191
era 2 1996-05-18 → 1999-09-30 1,229 trades · EV +0.212 without it +0.239
era 3 1999-09-30 → 2003-02-12 734 trades · EV +0.157 without it +0.252
Same-day ρ 0.09, design effect 1.19; the measurement partition would afford about 4,941 effective trades at this firing rate; at the cell's own σ (1.49) and k = 4, 0.05 R needs 13,350, 0.075 R needs 5,934, 0.1 R needs 3,338, 0.15 R needs 1,484, 0.2 R needs 835, 0.25 R needs 534.
Registering it: registrable in principle: three names or more, the sweep holds the plateau, and the measurement partition affords a floor of 0.1 R at k = 4. The question would carry the family's sweep of 30 cells; the accountant prices it at 0.3000 alpha on the regime axis (rate × 30 cells).
3 · sp_100 · up · breakout_low(lookback=50) · limit held
A buy-limit at the lowest low of the prior 50 bars, held for 5 bars with no regime gate, carries positive EV in net R on an S&P 100 constituent as listed on 2026-09-12 because a fall to a new low is over-sold and bought back (reversal at a new low) — inside the up regime of the classifier frozen on SPY (F11)
2,775 trades on the definition partition; EV gross +0.257, net +0.206 at bounded costs; always-long at this geometry -0.079 over 43,782; margin +0.285. 81 names traded; weakest MDLZ -4.039 over 2, strongest ACN +1.199 over 4. 6 entries missed — the fill auditor refused them and no position was opened (ADR-0041): C 5, DE 1. Every name held out: at worst +0.199.
era 1 1993-01-04 → 1996-05-18 774 trades · EV +0.285 without it +0.175
era 2 1996-05-18 → 1999-09-30 1,256 trades · EV +0.181 without it +0.226
era 3 1999-09-30 → 2003-02-12 745 trades · EV +0.164 without it +0.221
Same-day ρ 0.10, design effect 1.20; the measurement partition would afford about 5,038 effective trades at this firing rate; at the cell's own σ (1.44) and k = 4, 0.05 R needs 12,443, 0.075 R needs 5,530, 0.1 R needs 3,111, 0.15 R needs 1,383, 0.2 R needs 778, 0.25 R needs 498.
Registering it: registrable in principle: three names or more, the sweep holds the plateau, and the measurement partition affords a floor of 0.1 R at k = 4. The question would carry the family's sweep of 30 cells; the accountant prices it at 0.3000 alpha on the regime axis (rate × 30 cells).
4 · sp_100 · up · breakout_low(lookback=50) · limit held
A buy-limit at the lowest low of the prior 50 bars, held for 20 bars with no regime gate, carries positive EV in net R on an S&P 100 constituent as listed on 2026-09-12 because a fall to a new low is over-sold and bought back (reversal at a new low) — inside the up regime of the classifier frozen on SPY (F11)
2,421 trades on the definition partition; EV gross +0.278, net +0.226 at bounded costs; always-long at this geometry -0.044 over 30,785; margin +0.271. 81 names traded; weakest MDLZ -4.039 over 2, strongest XOM +1.416 over 19. 6 entries missed — the fill auditor refused them and no position was opened (ADR-0041): C 5, DE 1. Every name held out: at worst +0.217.
era 1 1993-01-04 → 1996-05-18 656 trades · EV +0.374 without it +0.172
era 2 1996-05-18 → 1999-09-30 1,113 trades · EV +0.246 without it +0.210
era 3 1999-09-30 → 2003-02-12 652 trades · EV +0.044 without it +0.294
Same-day ρ 0.10, design effect 1.18; the measurement partition would afford about 4,455 effective trades at this firing rate; at the cell's own σ (1.88) and k = 4, 0.05 R needs 21,228, 0.075 R needs 9,435, 0.1 R needs 5,307, 0.15 R needs 2,359, 0.2 R needs 1,327, 0.25 R needs 850.
Registering it: registrable in principle: three names or more, the sweep holds the plateau, and the measurement partition affords a floor of 0.15 R at k = 4. The question would carry the family's sweep of 30 cells; the accountant prices it at 0.3000 alpha on the regime axis (rate × 30 cells).