harry/News-ML-Decoder
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Initial import: XAUUSD news sentiment models, evals, and data tooling

Browse Source 
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.
This commit is contained in:
Harwinder Sekhon
2026-09-27 19:22:30 -04:00
commit 01baef48ff
98 changed files with 1719757 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
models/candidates/xau_all_20260922.joblib
BIN
View File
Binary file not shown.
models/candidates/xau_all_20260922.metrics.json
+186
View File
@@ -0,0 +1,186 @@
{
"evaluation": {
"stratified_holdout": {
"accuracy": 0.6927480916030534,
"balanced_accuracy": 0.689958210237739,
"classification_report": {
"accuracy": 0.6927480916030534,
"bearish": {
"f1-score": 0.752,
"precision": 0.7305699481865285,
"recall": 0.7747252747252747,
"support": 182.0
},
"bullish": {
"f1-score": 0.6230529595015576,
"precision": 0.6369426751592356,
"recall": 0.6097560975609756,
"support": 164.0
},
"macro avg": {
"f1-score": 0.6894115925611253,
"precision": 0.6895540162110402,
"recall": 0.689958210237739,
"support": 524.0
},
"neutral": {
"f1-score": 0.6931818181818182,
"precision": 0.7011494252873564,
"recall": 0.6853932584269663,
"support": 178.0
},
"weighted avg": {
"f1-score": 0.6916623072416396,
"precision": 0.6912727614450616,
"recall": 0.6927480916030534,
"support": 524.0
}
},
"confusion_matrix": {
"labels": [
"bearish",
"neutral",
"bullish"
],
"matrix": [
[
141,
18,
23
],
[
22,
122,
34
],
[
30,
34,
100
]
]
},
"macro_f1": 0.6894115925611253,
"score_mae": 0.2644602373963052,
"test_examples": 524
},
"temporal_holdout": {
"accuracy": 0.666030534351145,
"balanced_accuracy": 0.6572494770227411,
"classification_report": {
"accuracy": 0.666030534351145,
"bearish": {
"f1-score": 0.7040498442367601,
"precision": 0.773972602739726,
"recall": 0.6457142857142857,
"support": 175.0
},
"bullish": {
"f1-score": 0.5647840531561462,
"precision": 0.5483870967741935,
"recall": 0.5821917808219178,
"support": 146.0
},
"macro avg": {
"f1-score": 0.6592513617287779,
"precision": 0.6664965814523229,
"recall": 0.6572494770227411,
"support": 524.0
},
"neutral": {
"f1-score": 0.7089201877934272,
"precision": 0.6771300448430493,
"recall": 0.7438423645320197,
"support": 203.0
},
"weighted avg": {
"f1-score": 0.6671335737104888,
"precision": 0.6736013754038612,
"recall": 0.666030534351145,
"support": 524.0
}
},
"confusion_matrix": {
"labels": [
"bearish",
"neutral",
"bullish"
],
"matrix": [
[
113,
31,
31
],
[
13,
151,
39
],
[
20,
41,
85
]
]
},
"macro_f1": 0.6592513617287779,
"score_mae": 0.24771861749332214,
"test_examples": 524,
"test_from": "2026-09-21T07:34:39+00:00",
"train_through": "2026-09-21T07:33:06+00:00"
}
},
"format_version": 3,
"model_path": "models/candidates/xau_all_20260922.joblib",
"model_type": "domain_augmented_tfidf_logistic_regression",
"score_definition": "P(XAUUSD bullish) - P(XAUUSD bearish)",
"sklearn_version": "1.9.1",
"target": "XAUUSD",
"trained_at": "2026-09-23T00:35:47.838884+00:00",
"training": {
"database": "newsdata.db",
"domain_features": [
"explicit_xau_direction",
"usd_direction",
"treasury_yield_direction",
"hawkish_or_dovish_policy"
],
"duplicates_collapsed": 17,
"examples": 2620,
"extra_labels": "eval/results/backfill_labels.jsonl",
"label_distribution": {
"bearish": 909,
"bullish": 818,
"neutral": 893
},
"label_priority": [
"xau_score",
"json",
"extra_labels"
],
"label_sources": {
"extra_labels": 933,
"json": 640,
"xau_score": 1064
},
"max_features": 50000,
"min_df": 2,
"neutral_threshold": 0.05,
"published_range": [
"2026-09-08T13:35:37+00:00",
"2026-09-22T18:13:29+00:00"
],
"random_state": 42,
"rows_without_target_label": 24,
"scope": "all",
"score_source": "llm",
"skipped_rows": 0,
"source_rows": 2661,
"target": "XAUUSD",
"target_labeled_rows": 2637,
"target_score_key": "XAU",
"test_feed_rows": 0,
"test_size": 0.2
}
}
models/candidates/xau_gold_20260922.joblib
BIN
View File
Binary file not shown.
models/candidates/xau_gold_20260922.metrics.json
+185
View File
@@ -0,0 +1,185 @@
{
"evaluation": {
"stratified_holdout": {
"accuracy": 0.8042328042328042,
"balanced_accuracy": 0.7887686387686387,
"classification_report": {
"accuracy": 0.8042328042328042,
"bearish": {
"f1-score": 0.8435374149659864,
"precision": 0.8266666666666667,
"recall": 0.8611111111111112,
"support": 72.0
},
"bullish": {
"f1-score": 0.8,
"precision": 0.7948717948717948,
"recall": 0.8051948051948052,
"support": 77.0
},
"macro avg": {
"f1-score": 0.7934598400763816,
"precision": 0.7997720797720796,
"recall": 0.7887686387686387,
"support": 189.0
},
"neutral": {
"f1-score": 0.7368421052631579,
"precision": 0.7777777777777778,
"recall": 0.7,
"support": 40.0
},
"weighted avg": {
"f1-score": 0.8032189316829489,
"precision": 0.8033663455885677,
"recall": 0.8042328042328042,
"support": 189.0
}
},
"confusion_matrix": {
"labels": [
"bearish",
"neutral",
"bullish"
],
"matrix": [
[
62,
3,
7
],
[
3,
28,
9
],
[
10,
5,
62
]
]
},
"macro_f1": 0.7934598400763816,
"score_mae": 0.24237639972402897,
"test_examples": 189
},
"temporal_holdout": {
"accuracy": 0.746031746031746,
"balanced_accuracy": 0.7388071895424838,
"classification_report": {
"accuracy": 0.746031746031746,
"bearish": {
"f1-score": 0.8070175438596491,
"precision": 0.8518518518518519,
"recall": 0.7666666666666667,
"support": 90.0
},
"bullish": {
"f1-score": 0.6833333333333333,
"precision": 0.5942028985507246,
"recall": 0.803921568627451,
"support": 51.0
},
"macro avg": {
"f1-score": 0.7343315184513006,
"precision": 0.7469755150914571,
"recall": 0.7388071895424838,
"support": 189.0
},
"neutral": {
"f1-score": 0.7126436781609196,
"precision": 0.7948717948717948,
"recall": 0.6458333333333334,
"support": 48.0
},
"weighted avg": {
"f1-score": 0.7496744735401722,
"precision": 0.7678564055375648,
"recall": 0.746031746031746,
"support": 189.0
}
},
"confusion_matrix": {
"labels": [
"bearish",
"neutral",
"bullish"
],
"matrix": [
[
69,
5,
16
],
[
5,
31,
12
],
[
7,
3,
41
]
]
},
"macro_f1": 0.7343315184513006,
"score_mae": 0.27303151729050784,
"test_examples": 189,
"test_from": "2026-09-21T02:41:59+00:00",
"train_through": "2026-09-21T02:41:59+00:00"
}
},
"format_version": 3,
"model_path": "models/candidates/xau_gold_20260922.joblib",
"model_type": "domain_augmented_tfidf_logistic_regression",
"score_definition": "P(XAUUSD bullish) - P(XAUUSD bearish)",
"sklearn_version": "1.9.1",
"target": "XAUUSD",
"trained_at": "2026-09-23T00:35:57.482787+00:00",
"training": {
"database": "newsdata.db",
"domain_features": [
"explicit_xau_direction",
"usd_direction",
"treasury_yield_direction",
"hawkish_or_dovish_policy"
],
"duplicates_collapsed": 14,
"examples": 945,
"extra_labels": null,
"label_distribution": {
"bearish": 359,
"bullish": 384,
"neutral": 202
},
"label_priority": [
"xau_score",
"json",
"extra_labels"
],
"label_sources": {
"json": 640,
"xau_score": 319
},
"max_features": 50000,
"min_df": 2,
"neutral_threshold": 0.05,
"published_range": [
"2026-09-08T13:46:00+00:00",
"2026-09-22T18:13:29+00:00"
],
"random_state": 42,
"rows_without_target_label": 1702,
"scope": "gold",
"score_source": "llm",
"skipped_rows": 0,
"source_rows": 2661,
"target": "XAUUSD",
"target_labeled_rows": 959,
"target_score_key": "XAU",
"test_feed_rows": 0,
"test_size": 0.2
}
}