Files
News-ML-Decoder/eval/results/archive_20260923_0038_all.json
Harwinder Sekhon 01baef48ff Initial import: XAUUSD news sentiment models, evals, and data tooling
TF-IDF baseline (news_sentiment.py, models/xauusd_sentiment.joblib) and the
two-head DeBERTa-v3-small transformer (transformer/, news_transformer.py)
with its Google News crawler and DeepSeek labeller (data_collect/), the
golden-set and archive evals (eval/), and the 2026-09-22 and 2026-09-27
results and progress notes.

Transformer weights (model.pt, model.onnx) and the crawl database are not
tracked; rebuild with transformer/train.py and transformer/export_onnx.py.
The DeepSeek label cache in data/ is tracked.
2026-09-27 19:22:30 -04:00

1886 lines
40 KiB
JSON

{
"after_training": {
"all": {
"accuracy": 0.45872420262664165,
"confusion": [
[
186,
27,
106
],
[
74,
109,
192
],
[
129,
49,
194
]
],
"directional_accuracy": 0.5499276410998553,
"directional_hits": 380,
"directional_n": 691,
"hits": 489,
"macro_f1": 0.4545945057244492,
"n": 1066,
"sign_agreement": 0.6178861788617886,
"sign_hits": 380,
"sign_n": 615
},
"gold": {
"accuracy": 0.6822429906542056,
"confusion": [
[
84,
7,
19
],
[
8,
35,
16
],
[
42,
10,
100
]
],
"directional_accuracy": 0.7022900763358778,
"directional_hits": 184,
"directional_n": 262,
"hits": 219,
"macro_f1": 0.6720064440976077,
"n": 321,
"sign_agreement": 0.7510204081632653,
"sign_hits": 184,
"sign_n": 245
},
"macro": {
"accuracy": 0.3624161073825503,
"confusion": [
[
102,
20,
87
],
[
66,
74,
176
],
[
87,
39,
94
]
],
"directional_accuracy": 0.4568764568764569,
"directional_hits": 196,
"directional_n": 429,
"hits": 270,
"macro_f1": 0.36503325894343636,
"n": 745,
"sign_agreement": 0.5297297297297298,
"sign_hits": 196,
"sign_n": 370
}
},
"days": [
"2026-09-08",
"2026-09-09",
"2026-09-10",
"2026-09-11",
"2026-09-13",
"2026-09-14",
"2026-09-15",
"2026-09-16",
"2026-09-17",
"2026-09-18",
"2026-09-20",
"2026-09-21",
"2026-09-22"
],
"db": "../newsdata.db",
"extra_labels": "results/backfill_labels.jsonl",
"folds": [
{
"day": "2026-09-09",
"fit_seconds_full": 1.1599936260004142,
"fit_seconds_head": 0.02371588599999086,
"full": {
"accuracy": 0.5,
"confusion": [
[
4,
6,
0
],
[
3,
10,
0
],
[
5,
4,
4
]
],
"directional_accuracy": 0.34782608695652173,
"directional_hits": 8,
"directional_n": 23,
"hits": 18,
"macro_f1": 0.4800950683303624,
"n": 36,
"sign_agreement": 0.6153846153846154,
"sign_hits": 8,
"sign_n": 13
},
"head": {
"accuracy": 0.5277777777777778,
"confusion": [
[
5,
5,
0
],
[
1,
11,
1
],
[
3,
7,
3
]
],
"directional_accuracy": 0.34782608695652173,
"directional_hits": 8,
"directional_n": 23,
"hits": 19,
"macro_f1": 0.49678935901846116,
"n": 36,
"sign_agreement": 0.7272727272727273,
"sign_hits": 8,
"sign_n": 11
},
"keyword": {
"accuracy": 0.4166666666666667,
"confusion": [
[
2,
7,
1
],
[
0,
13,
0
],
[
1,
12,
0
]
],
"directional_accuracy": 0.08695652173913043,
"directional_hits": 2,
"directional_n": 23,
"hits": 15,
"macro_f1": 0.2951566951566951,
"n": 36,
"sign_agreement": 0.5,
"sign_hits": 2,
"sign_n": 4
},
"n_test": 36,
"n_train": 71
},
{
"day": "2026-09-10",
"fit_seconds_full": 1.522781863000091,
"fit_seconds_head": 0.026001763000294886,
"full": {
"accuracy": 0.49767441860465117,
"confusion": [
[
25,
42,
3
],
[
2,
60,
11
],
[
7,
43,
22
]
],
"directional_accuracy": 0.33098591549295775,
"directional_hits": 47,
"directional_n": 142,
"hits": 107,
"macro_f1": 0.4795451179243228,
"n": 215,
"sign_agreement": 0.8245614035087719,
"sign_hits": 47,
"sign_n": 57
},
"head": {
"accuracy": 0.5720930232558139,
"confusion": [
[
34,
29,
7
],
[
1,
59,
13
],
[
8,
34,
30
]
],
"directional_accuracy": 0.4507042253521127,
"directional_hits": 64,
"directional_n": 142,
"hits": 123,
"macro_f1": 0.5662337984403848,
"n": 215,
"sign_agreement": 0.810126582278481,
"sign_hits": 64,
"sign_n": 79
},
"keyword": {
"accuracy": 0.4558139534883721,
"confusion": [
[
23,
46,
1
],
[
3,
69,
1
],
[
11,
55,
6
]
],
"directional_accuracy": 0.20422535211267606,
"directional_hits": 29,
"directional_n": 142,
"hits": 98,
"macro_f1": 0.38260259220799203,
"n": 215,
"sign_agreement": 0.7073170731707317,
"sign_hits": 29,
"sign_n": 41
},
"n_test": 215,
"n_train": 107
},
{
"day": "2026-09-11",
"fit_seconds_full": 2.209334836999915,
"fit_seconds_head": 1.8741989730001478,
"full": {
"accuracy": 0.45365853658536587,
"confusion": [
[
32,
23,
15
],
[
11,
38,
14
],
[
19,
30,
23
]
],
"directional_accuracy": 0.3873239436619718,
"directional_hits": 55,
"directional_n": 142,
"hits": 93,
"macro_f1": 0.44977424009682077,
"n": 205,
"sign_agreement": 0.6179775280898876,
"sign_hits": 55,
"sign_n": 89
},
"head": {
"accuracy": 0.5365853658536586,
"confusion": [
[
39,
14,
17
],
[
7,
37,
19
],
[
18,
20,
34
]
],
"directional_accuracy": 0.5140845070422535,
"directional_hits": 73,
"directional_n": 142,
"hits": 110,
"macro_f1": 0.5377338658818583,
"n": 205,
"sign_agreement": 0.6759259259259259,
"sign_hits": 73,
"sign_n": 108
},
"keyword": {
"accuracy": 0.40487804878048783,
"confusion": [
[
24,
43,
3
],
[
7,
54,
2
],
[
15,
52,
5
]
],
"directional_accuracy": 0.20422535211267606,
"directional_hits": 29,
"directional_n": 142,
"hits": 83,
"macro_f1": 0.3483927617415407,
"n": 205,
"sign_agreement": 0.6170212765957447,
"sign_hits": 29,
"sign_n": 47
},
"n_test": 205,
"n_train": 322
},
{
"day": "2026-09-13",
"fit_seconds_full": 3.299125021999771,
"fit_seconds_head": 3.1570361300000513,
"full": {
"accuracy": 0.4,
"confusion": [
[
1,
0,
0
],
[
0,
1,
2
],
[
1,
0,
0
]
],
"directional_accuracy": 0.5,
"directional_hits": 1,
"directional_n": 2,
"hits": 2,
"macro_f1": 0.38888888888888884,
"n": 5,
"sign_agreement": 0.5,
"sign_hits": 1,
"sign_n": 2
},
"head": {
"accuracy": 0.6,
"confusion": [
[
1,
0,
0
],
[
0,
2,
1
],
[
0,
1,
0
]
],
"directional_accuracy": 0.5,
"directional_hits": 1,
"directional_n": 2,
"hits": 3,
"macro_f1": 0.5555555555555555,
"n": 5,
"sign_agreement": 1.0,
"sign_hits": 1,
"sign_n": 1
},
"keyword": {
"accuracy": 0.8,
"confusion": [
[
1,
0,
0
],
[
0,
3,
0
],
[
0,
1,
0
]
],
"directional_accuracy": 0.5,
"directional_hits": 1,
"directional_n": 2,
"hits": 4,
"macro_f1": 0.6190476190476191,
"n": 5,
"sign_agreement": 1.0,
"sign_hits": 1,
"sign_n": 1
},
"n_test": 5,
"n_train": 527
},
{
"day": "2026-09-14",
"fit_seconds_full": 3.3643084619998262,
"fit_seconds_head": 3.809506984000109,
"full": {
"accuracy": 0.6278195488721805,
"confusion": [
[
87,
35,
22
],
[
4,
59,
14
],
[
10,
14,
21
]
],
"directional_accuracy": 0.5714285714285714,
"directional_hits": 108,
"directional_n": 189,
"hits": 167,
"macro_f1": 0.5866022084509479,
"n": 266,
"sign_agreement": 0.7714285714285715,
"sign_hits": 108,
"sign_n": 140
},
"head": {
"accuracy": 0.6390977443609023,
"confusion": [
[
89,
25,
30
],
[
2,
64,
11
],
[
4,
24,
17
]
],
"directional_accuracy": 0.5608465608465608,
"directional_hits": 106,
"directional_n": 189,
"hits": 170,
"macro_f1": 0.5828503907939814,
"n": 266,
"sign_agreement": 0.7571428571428571,
"sign_hits": 106,
"sign_n": 140
},
"keyword": {
"accuracy": 0.5075187969924813,
"confusion": [
[
61,
77,
6
],
[
7,
69,
1
],
[
5,
35,
5
]
],
"directional_accuracy": 0.3492063492063492,
"directional_hits": 66,
"directional_n": 189,
"hits": 135,
"macro_f1": 0.42417809966276027,
"n": 266,
"sign_agreement": 0.8571428571428571,
"sign_hits": 66,
"sign_n": 77
},
"n_test": 266,
"n_train": 532
},
{
"day": "2026-09-15",
"fit_seconds_full": 3.2497777419998783,
"fit_seconds_head": 3.444847923999987,
"full": {
"accuracy": 0.576036866359447,
"confusion": [
[
57,
17,
16
],
[
6,
54,
8
],
[
24,
21,
14
]
],
"directional_accuracy": 0.47651006711409394,
"directional_hits": 71,
"directional_n": 149,
"hits": 125,
"macro_f1": 0.5359091968082008,
"n": 217,
"sign_agreement": 0.6396396396396397,
"sign_hits": 71,
"sign_n": 111
},
"head": {
"accuracy": 0.6082949308755761,
"confusion": [
[
55,
18,
17
],
[
3,
50,
15
],
[
10,
22,
27
]
],
"directional_accuracy": 0.5503355704697986,
"directional_hits": 82,
"directional_n": 149,
"hits": 132,
"macro_f1": 0.5955803475649003,
"n": 217,
"sign_agreement": 0.7522935779816514,
"sign_hits": 82,
"sign_n": 109
},
"keyword": {
"accuracy": 0.4009216589861751,
"confusion": [
[
24,
59,
7
],
[
4,
59,
5
],
[
11,
44,
4
]
],
"directional_accuracy": 0.18791946308724833,
"directional_hits": 28,
"directional_n": 149,
"hits": 87,
"macro_f1": 0.3306010560611167,
"n": 217,
"sign_agreement": 0.6086956521739131,
"sign_hits": 28,
"sign_n": 46
},
"n_test": 217,
"n_train": 798
},
{
"day": "2026-09-16",
"fit_seconds_full": 4.366315348999706,
"fit_seconds_head": 4.552804755000125,
"full": {
"accuracy": 0.5375,
"confusion": [
[
69,
19,
17
],
[
22,
63,
24
],
[
36,
30,
40
]
],
"directional_accuracy": 0.5165876777251185,
"directional_hits": 109,
"directional_n": 211,
"hits": 172,
"macro_f1": 0.5309236064814159,
"n": 320,
"sign_agreement": 0.6728395061728395,
"sign_hits": 109,
"sign_n": 162
},
"head": {
"accuracy": 0.5625,
"confusion": [
[
68,
18,
19
],
[
20,
66,
23
],
[
32,
28,
46
]
],
"directional_accuracy": 0.5402843601895735,
"directional_hits": 114,
"directional_n": 211,
"hits": 180,
"macro_f1": 0.5586521054804863,
"n": 320,
"sign_agreement": 0.6909090909090909,
"sign_hits": 114,
"sign_n": 165
},
"keyword": {
"accuracy": 0.41875,
"confusion": [
[
32,
67,
6
],
[
11,
90,
8
],
[
15,
79,
12
]
],
"directional_accuracy": 0.20853080568720378,
"directional_hits": 44,
"directional_n": 211,
"hits": 134,
"macro_f1": 0.36539844968759344,
"n": 320,
"sign_agreement": 0.676923076923077,
"sign_hits": 44,
"sign_n": 65
},
"n_test": 320,
"n_train": 1015
},
{
"day": "2026-09-17",
"fit_seconds_full": 4.3225185520000196,
"fit_seconds_head": 5.175150255000062,
"full": {
"accuracy": 0.6,
"confusion": [
[
67,
22,
19
],
[
15,
55,
13
],
[
19,
24,
46
]
],
"directional_accuracy": 0.5736040609137056,
"directional_hits": 113,
"directional_n": 197,
"hits": 168,
"macro_f1": 0.5966242053027293,
"n": 280,
"sign_agreement": 0.7483443708609272,
"sign_hits": 113,
"sign_n": 151
},
"head": {
"accuracy": 0.5964285714285714,
"confusion": [
[
72,
19,
17
],
[
12,
44,
27
],
[
18,
20,
51
]
],
"directional_accuracy": 0.6243654822335025,
"directional_hits": 123,
"directional_n": 197,
"hits": 167,
"macro_f1": 0.5900608645763178,
"n": 280,
"sign_agreement": 0.7784810126582279,
"sign_hits": 123,
"sign_n": 158
},
"keyword": {
"accuracy": 0.4107142857142857,
"confusion": [
[
37,
65,
6
],
[
13,
66,
4
],
[
11,
66,
12
]
],
"directional_accuracy": 0.24873096446700507,
"directional_hits": 49,
"directional_n": 197,
"hits": 115,
"macro_f1": 0.3751715367099983,
"n": 280,
"sign_agreement": 0.7424242424242424,
"sign_hits": 49,
"sign_n": 66
},
"n_test": 280,
"n_train": 1335
},
{
"day": "2026-09-18",
"fit_seconds_full": 5.254859737000061,
"fit_seconds_head": 4.997993596000015,
"full": {
"accuracy": 0.5835777126099707,
"confusion": [
[
43,
15,
17
],
[
10,
56,
37
],
[
34,
29,
100
]
],
"directional_accuracy": 0.6008403361344538,
"directional_hits": 143,
"directional_n": 238,
"hits": 199,
"macro_f1": 0.5711677206534405,
"n": 341,
"sign_agreement": 0.7371134020618557,
"sign_hits": 143,
"sign_n": 194
},
"head": {
"accuracy": 0.5894428152492669,
"confusion": [
[
47,
13,
15
],
[
11,
59,
33
],
[
40,
28,
95
]
],
"directional_accuracy": 0.5966386554621849,
"directional_hits": 142,
"directional_n": 238,
"hits": 201,
"macro_f1": 0.5818494740043826,
"n": 341,
"sign_agreement": 0.7208121827411168,
"sign_hits": 142,
"sign_n": 197
},
"keyword": {
"accuracy": 0.41055718475073316,
"confusion": [
[
28,
34,
13
],
[
5,
89,
9
],
[
20,
120,
23
]
],
"directional_accuracy": 0.21428571428571427,
"directional_hits": 51,
"directional_n": 238,
"hits": 140,
"macro_f1": 0.3910349044019564,
"n": 341,
"sign_agreement": 0.6071428571428571,
"sign_hits": 51,
"sign_n": 84
},
"n_test": 341,
"n_train": 1615
},
{
"day": "2026-09-20",
"fit_seconds_full": 4.983563467999829,
"fit_seconds_head": 3.135302303000117,
"full": {
"accuracy": 0.6538461538461539,
"confusion": [
[
6,
0,
1
],
[
0,
8,
0
],
[
2,
6,
3
]
],
"directional_accuracy": 0.5,
"directional_hits": 9,
"directional_n": 18,
"hits": 17,
"macro_f1": 0.6424242424242425,
"n": 26,
"sign_agreement": 0.75,
"sign_hits": 9,
"sign_n": 12
},
"head": {
"accuracy": 0.7307692307692307,
"confusion": [
[
3,
0,
4
],
[
0,
8,
0
],
[
1,
2,
8
]
],
"directional_accuracy": 0.6111111111111112,
"directional_hits": 11,
"directional_n": 18,
"hits": 19,
"macro_f1": 0.7099985360854926,
"n": 26,
"sign_agreement": 0.6875,
"sign_hits": 11,
"sign_n": 16
},
"keyword": {
"accuracy": 0.46153846153846156,
"confusion": [
[
0,
7,
0
],
[
0,
8,
0
],
[
0,
7,
4
]
],
"directional_accuracy": 0.2222222222222222,
"directional_hits": 4,
"directional_n": 18,
"hits": 12,
"macro_f1": 0.35555555555555557,
"n": 26,
"sign_agreement": 1.0,
"sign_hits": 4,
"sign_n": 4
},
"n_test": 26,
"n_train": 1956
},
{
"day": "2026-09-21",
"fit_seconds_full": 5.172674763000032,
"fit_seconds_head": 3.7032702040000913,
"full": {
"accuracy": 0.6430517711171662,
"confusion": [
[
84,
19,
29
],
[
16,
96,
31
],
[
10,
26,
56
]
],
"directional_accuracy": 0.625,
"directional_hits": 140,
"directional_n": 224,
"hits": 236,
"macro_f1": 0.6362442508409218,
"n": 367,
"sign_agreement": 0.7821229050279329,
"sign_hits": 140,
"sign_n": 179
},
"head": {
"accuracy": 0.6566757493188011,
"confusion": [
[
80,
16,
36
],
[
8,
105,
30
],
[
8,
28,
56
]
],
"directional_accuracy": 0.6071428571428571,
"directional_hits": 136,
"directional_n": 224,
"hits": 241,
"macro_f1": 0.6480989847126672,
"n": 367,
"sign_agreement": 0.7555555555555555,
"sign_hits": 136,
"sign_n": 180
},
"keyword": {
"accuracy": 0.4986376021798365,
"confusion": [
[
53,
72,
7
],
[
4,
125,
14
],
[
5,
82,
5
]
],
"directional_accuracy": 0.25892857142857145,
"directional_hits": 58,
"directional_n": 224,
"hits": 183,
"macro_f1": 0.4078515256335195,
"n": 367,
"sign_agreement": 0.8285714285714286,
"sign_hits": 58,
"sign_n": 70
},
"n_test": 367,
"n_train": 1982
},
{
"day": "2026-09-22",
"fit_seconds_full": 4.346579720999671,
"fit_seconds_head": 6.362624880000112,
"full": {
"accuracy": 0.6597222222222222,
"confusion": [
[
60,
16,
17
],
[
7,
85,
20
],
[
16,
22,
45
]
],
"directional_accuracy": 0.5965909090909091,
"directional_hits": 105,
"directional_n": 176,
"hits": 190,
"macro_f1": 0.6502256608639586,
"n": 288,
"sign_agreement": 0.7608695652173914,
"sign_hits": 105,
"sign_n": 138
},
"head": {
"accuracy": 0.6527777777777778,
"confusion": [
[
56,
18,
19
],
[
2,
86,
24
],
[
18,
19,
46
]
],
"directional_accuracy": 0.5795454545454546,
"directional_hits": 102,
"directional_n": 176,
"hits": 188,
"macro_f1": 0.6431735026794594,
"n": 288,
"sign_agreement": 0.7338129496402878,
"sign_hits": 102,
"sign_n": 139
},
"keyword": {
"accuracy": 0.4895833333333333,
"confusion": [
[
37,
49,
7
],
[
4,
99,
9
],
[
17,
61,
5
]
],
"directional_accuracy": 0.23863636363636365,
"directional_hits": 42,
"directional_n": 176,
"hits": 141,
"macro_f1": 0.40101416707531706,
"n": 288,
"sign_agreement": 0.6363636363636364,
"sign_hits": 42,
"sign_n": 66
},
"n_test": 288,
"n_train": 2349
}
],
"generated_at": "20260923_0038_all",
"in_sample": {
"headline_description": {
"accuracy": 0.5608646188850968,
"confusion": [
[
624,
73,
219
],
[
182,
335,
379
],
[
222,
83,
520
]
],
"directional_accuracy": 0.6570936243538197,
"directional_hits": 1144,
"directional_n": 1741,
"hits": 1479,
"macro_f1": 0.5534290089278285,
"mae": 0.2846768854759196,
"n": 2637,
"sign_agreement": 0.7217665615141956,
"sign_hits": 1144,
"sign_n": 1585,
"spearman": 0.5336683695481723
},
"headline_only": {
"accuracy": 0.5733788395904437,
"confusion": [
[
652,
43,
221
],
[
152,
304,
440
],
[
210,
59,
556
]
],
"directional_accuracy": 0.6938541068351523,
"directional_hits": 1208,
"directional_n": 1741,
"hits": 1512,
"macro_f1": 0.5623952358379628,
"mae": 0.30869570231323473,
"n": 2637,
"sign_agreement": 0.7370347773032336,
"sign_hits": 1208,
"sign_n": 1639,
"spearman": 0.554711637369288
}
},
"label_distribution": {
"bearish": 916,
"bullish": 825,
"neutral": 896
},
"model": {
"label_distribution": {
"bearish": 250,
"bullish": 221,
"neutral": 135
},
"model_path": "/home/harwinder/Development/ML-NEWS/models/xauusd_sentiment.joblib",
"model_type": "domain_augmented_tfidf_logistic_regression",
"neutral_threshold": 0.05,
"sklearn_version": "1.9.1",
"trained_at": "2026-09-17T18:54:46.714699+00:00",
"training_examples": 606
},
"pooled_oof": {
"full": {
"accuracy": 0.5822291504286827,
"confusion": [
[
535,
214,
156
],
[
96,
585,
174
],
[
183,
249,
374
]
],
"directional_accuracy": 0.5312682641729982,
"directional_hits": 909,
"directional_n": 1711,
"hits": 1494,
"macro_f1": 0.5775459537878285,
"mae": 0.2729630943102105,
"n": 2566,
"sign_agreement": 0.7283653846153846,
"sign_hits": 909,
"sign_n": 1248,
"spearman": 0.45128845313804333
},
"full_gold": {
"accuracy": 0.6508620689655172,
"confusion": [
[
241,
34,
82
],
[
23,
122,
47
],
[
78,
60,
241
]
],
"directional_accuracy": 0.654891304347826,
"directional_hits": 482,
"directional_n": 736,
"hits": 604,
"macro_f1": 0.643706808194913,
"mae": 0.32061842672413793,
"n": 928,
"sign_agreement": 0.7507788161993769,
"sign_hits": 482,
"sign_n": 642,
"spearman": 0.5534196552017715
},
"full_macro": {
"accuracy": 0.5433455433455433,
"confusion": [
[
294,
180,
74
],
[
73,
463,
127
],
[
105,
189,
133
]
],
"directional_accuracy": 0.4379487179487179,
"directional_hits": 427,
"directional_n": 975,
"hits": 890,
"macro_f1": 0.5151362201299858,
"mae": 0.24596422466422468,
"n": 1638,
"sign_agreement": 0.7046204620462047,
"sign_hits": 427,
"sign_n": 606,
"spearman": 0.34905990065793735
},
"full_since_2026-09-18_all": {
"accuracy": 0.6281800391389433,
"confusion": [
[
193,
50,
64
],
[
33,
245,
88
],
[
62,
83,
204
]
],
"directional_accuracy": 0.6051829268292683,
"directional_hits": 397,
"directional_n": 656,
"hits": 642,
"macro_f1": 0.6286883501970909,
"n": 1022,
"sign_agreement": 0.7590822179732314,
"sign_hits": 397,
"sign_n": 523
},
"full_since_2026-09-18_gold": {
"accuracy": 0.7188498402555911,
"confusion": [
[
75,
6,
28
],
[
8,
36,
15
],
[
21,
10,
114
]
],
"directional_accuracy": 0.7440944881889764,
"directional_hits": 189,
"directional_n": 254,
"hits": 225,
"macro_f1": 0.7026136293928477,
"n": 313,
"sign_agreement": 0.7941176470588235,
"sign_hits": 189,
"sign_n": 238
},
"full_since_2026-09-18_macro": {
"accuracy": 0.5881523272214386,
"confusion": [
[
118,
44,
36
],
[
25,
209,
73
],
[
41,
73,
90
]
],
"directional_accuracy": 0.5174129353233831,
"directional_hits": 208,
"directional_n": 402,
"hits": 417,
"macro_f1": 0.5749329075109032,
"n": 709,
"sign_agreement": 0.7298245614035088,
"sign_hits": 208,
"sign_n": 285
},
"full_train_ge_300": {
"accuracy": 0.5913606911447085,
"confusion": [
[
506,
166,
153
],
[
91,
515,
163
],
[
171,
202,
348
]
],
"directional_accuracy": 0.5523932729624839,
"directional_hits": 854,
"directional_n": 1546,
"hits": 1369,
"macro_f1": 0.5870977019215816,
"mae": 0.2741298056155508,
"n": 2315,
"sign_agreement": 0.7249575551782682,
"sign_hits": 854,
"sign_n": 1178,
"spearman": 0.4632028749594082
},
"head": {
"accuracy": 0.6052221356196414,
"confusion": [
[
549,
175,
181
],
[
67,
591,
197
],
[
160,
233,
413
]
],
"directional_accuracy": 0.5622443015780245,
"directional_hits": 962,
"directional_n": 1711,
"hits": 1553,
"macro_f1": 0.6026476231875058,
"mae": 0.3004310210444271,
"n": 2566,
"sign_agreement": 0.7382962394474291,
"sign_hits": 962,
"sign_n": 1303,
"spearman": 0.4780602704994956
},
"keyword": {
"accuracy": 0.4469992205767732,
"confusion": [
[
322,
526,
57
],
[
58,
744,
53
],
[
111,
614,
81
]
],
"directional_accuracy": 0.23553477498538866,
"directional_hits": 403,
"directional_n": 1711,
"hits": 1147,
"macro_f1": 0.38902315963793493,
"n": 2566,
"sign_agreement": 0.7057793345008757,
"sign_hits": 403,
"sign_n": 571
}
},
"reprobe": {},
"routing": [
{
"confidence_min": 0.0,
"hybrid_agreement": 0.5822291504286827,
"kept": 2566,
"kept_accuracy": 0.5822291504286827,
"kept_share": 1.0,
"llm_calls_saved_share": 1.0,
"margin_min": 0.0
},
{
"confidence_min": 0.0,
"hybrid_agreement": 0.6874512860483243,
"kept": 2096,
"kept_accuracy": 0.6173664122137404,
"kept_share": 0.8168355416991426,
"llm_calls_saved_share": 0.8168355416991426,
"margin_min": 0.1
},
{
"confidence_min": 0.0,
"hybrid_agreement": 0.7727981293842556,
"kept": 1700,
"kept_accuracy": 0.6570588235294118,
"kept_share": 0.6625097427903351,
"llm_calls_saved_share": 0.6625097427903351,
"margin_min": 0.2
},
{
"confidence_min": 0.0,
"hybrid_agreement": 0.8347622759158223,
"kept": 1358,
"kept_accuracy": 0.6877761413843888,
"kept_share": 0.5292283710054559,
"llm_calls_saved_share": 0.5292283710054559,
"margin_min": 0.3
},
{
"confidence_min": 0.0,
"hybrid_agreement": 0.886983632112237,
"kept": 1052,
"kept_accuracy": 0.7243346007604563,
"kept_share": 0.4099766173031956,
"llm_calls_saved_share": 0.4099766173031956,
"margin_min": 0.4
},
{
"confidence_min": 0.5,
"hybrid_agreement": 0.7603273577552611,
"kept": 1786,
"kept_accuracy": 0.6556550951847704,
"kept_share": 0.696024941543258,
"llm_calls_saved_share": 0.696024941543258,
"margin_min": 0.0
},
{
"confidence_min": 0.5,
"hybrid_agreement": 0.7626656274356975,
"kept": 1770,
"kept_accuracy": 0.6559322033898305,
"kept_share": 0.6897895557287607,
"llm_calls_saved_share": 0.6897895557287607,
"margin_min": 0.1
},
{
"confidence_min": 0.5,
"hybrid_agreement": 0.7848791893998441,
"kept": 1654,
"kept_accuracy": 0.6662636033857315,
"kept_share": 0.6445830085736555,
"llm_calls_saved_share": 0.6445830085736555,
"margin_min": 0.2
},
{
"confidence_min": 0.5,
"hybrid_agreement": 0.8347622759158223,
"kept": 1358,
"kept_accuracy": 0.6877761413843888,
"kept_share": 0.5292283710054559,
"llm_calls_saved_share": 0.5292283710054559,
"margin_min": 0.3
},
{
"confidence_min": 0.5,
"hybrid_agreement": 0.886983632112237,
"kept": 1052,
"kept_accuracy": 0.7243346007604563,
"kept_share": 0.4099766173031956,
"llm_calls_saved_share": 0.4099766173031956,
"margin_min": 0.4
},
{
"confidence_min": 0.6,
"hybrid_agreement": 0.8604832424006236,
"kept": 1240,
"kept_accuracy": 0.7112903225806452,
"kept_share": 0.48324240062353857,
"llm_calls_saved_share": 0.48324240062353857,
"margin_min": 0.0
},
{
"confidence_min": 0.6,
"hybrid_agreement": 0.8604832424006236,
"kept": 1240,
"kept_accuracy": 0.7112903225806452,
"kept_share": 0.48324240062353857,
"llm_calls_saved_share": 0.48324240062353857,
"margin_min": 0.1
},
{
"confidence_min": 0.6,
"hybrid_agreement": 0.8604832424006236,
"kept": 1240,
"kept_accuracy": 0.7112903225806452,
"kept_share": 0.48324240062353857,
"llm_calls_saved_share": 0.48324240062353857,
"margin_min": 0.2
},
{
"confidence_min": 0.6,
"hybrid_agreement": 0.8628215120810601,
"kept": 1218,
"kept_accuracy": 0.7110016420361248,
"kept_share": 0.47466874512860485,
"llm_calls_saved_share": 0.47466874512860485,
"margin_min": 0.3
},
{
"confidence_min": 0.6,
"hybrid_agreement": 0.886983632112237,
"kept": 1052,
"kept_accuracy": 0.7243346007604563,
"kept_share": 0.4099766173031956,
"llm_calls_saved_share": 0.4099766173031956,
"margin_min": 0.4
},
{
"confidence_min": 0.7,
"hybrid_agreement": 0.9368667186282151,
"kept": 747,
"kept_accuracy": 0.7831325301204819,
"kept_share": 0.29111457521434136,
"llm_calls_saved_share": 0.29111457521434136,
"margin_min": 0.0
},
{
"confidence_min": 0.7,
"hybrid_agreement": 0.9368667186282151,
"kept": 747,
"kept_accuracy": 0.7831325301204819,
"kept_share": 0.29111457521434136,
"llm_calls_saved_share": 0.29111457521434136,
"margin_min": 0.1
},
{
"confidence_min": 0.7,
"hybrid_agreement": 0.9368667186282151,
"kept": 747,
"kept_accuracy": 0.7831325301204819,
"kept_share": 0.29111457521434136,
"llm_calls_saved_share": 0.29111457521434136,
"margin_min": 0.2
},
{
"confidence_min": 0.7,
"hybrid_agreement": 0.9368667186282151,
"kept": 747,
"kept_accuracy": 0.7831325301204819,
"kept_share": 0.29111457521434136,
"llm_calls_saved_share": 0.29111457521434136,
"margin_min": 0.3
},
{
"confidence_min": 0.7,
"hybrid_agreement": 0.9368667186282151,
"kept": 747,
"kept_accuracy": 0.7831325301204819,
"kept_share": 0.29111457521434136,
"llm_calls_saved_share": 0.29111457521434136,
"margin_min": 0.4
},
{
"confidence_min": 0.8,
"hybrid_agreement": 0.9828526890101325,
"kept": 409,
"kept_accuracy": 0.8924205378973105,
"kept_share": 0.15939204988308653,
"llm_calls_saved_share": 0.15939204988308653,
"margin_min": 0.0
},
{
"confidence_min": 0.8,
"hybrid_agreement": 0.9828526890101325,
"kept": 409,
"kept_accuracy": 0.8924205378973105,
"kept_share": 0.15939204988308653,
"llm_calls_saved_share": 0.15939204988308653,
"margin_min": 0.1
},
{
"confidence_min": 0.8,
"hybrid_agreement": 0.9828526890101325,
"kept": 409,
"kept_accuracy": 0.8924205378973105,
"kept_share": 0.15939204988308653,
"llm_calls_saved_share": 0.15939204988308653,
"margin_min": 0.2
},
{
"confidence_min": 0.8,
"hybrid_agreement": 0.9828526890101325,
"kept": 409,
"kept_accuracy": 0.8924205378973105,
"kept_share": 0.15939204988308653,
"llm_calls_saved_share": 0.15939204988308653,
"margin_min": 0.3
},
{
"confidence_min": 0.8,
"hybrid_agreement": 0.9828526890101325,
"kept": 409,
"kept_accuracy": 0.8924205378973105,
"kept_share": 0.15939204988308653,
"llm_calls_saved_share": 0.15939204988308653,
"margin_min": 0.4
},
{
"confidence_min": 0.9,
"hybrid_agreement": 0.9964925954793453,
"kept": 152,
"kept_accuracy": 0.9407894736842105,
"kept_share": 0.059236165237724084,
"llm_calls_saved_share": 0.059236165237724084,
"margin_min": 0.0
},
{
"confidence_min": 0.9,
"hybrid_agreement": 0.9964925954793453,
"kept": 152,
"kept_accuracy": 0.9407894736842105,
"kept_share": 0.059236165237724084,
"llm_calls_saved_share": 0.059236165237724084,
"margin_min": 0.1
},
{
"confidence_min": 0.9,
"hybrid_agreement": 0.9964925954793453,
"kept": 152,
"kept_accuracy": 0.9407894736842105,
"kept_share": 0.059236165237724084,
"llm_calls_saved_share": 0.059236165237724084,
"margin_min": 0.2
},
{
"confidence_min": 0.9,
"hybrid_agreement": 0.9964925954793453,
"kept": 152,
"kept_accuracy": 0.9407894736842105,
"kept_share": 0.059236165237724084,
"llm_calls_saved_share": 0.059236165237724084,
"margin_min": 0.3
},
{
"confidence_min": 0.9,
"hybrid_agreement": 0.9964925954793453,
"kept": 152,
"kept_accuracy": 0.9407894736842105,
"kept_share": 0.059236165237724084,
"llm_calls_saved_share": 0.059236165237724084,
"margin_min": 0.4
}
],
"rows": 2637,
"rows_per_day": {
"2026-09-08": 71,
"2026-09-09": 36,
"2026-09-10": 215,
"2026-09-11": 205,
"2026-09-13": 5,
"2026-09-14": 266,
"2026-09-15": 217,
"2026-09-16": 320,
"2026-09-17": 280,
"2026-09-18": 341,
"2026-09-20": 26,
"2026-09-21": 367,
"2026-09-22": 288
},
"threshold": 0.05,
"train_scope": "all"
}