Best model: CFG_XAG_ASIAN_V6/run_20260601_221745_v6_asian

workspaces/CFG_XAG_ASIAN_V6/runs/run_20260601_221745_v6_asian/config.json

@@ -0,0 +1,100 @@
+{
+    "HF_RUN_ID": "run_20260601_221745_v6_asian",
+    "MT5_PATH": "D:\\mt5\\MetaTrader 5 EXNESS\\terminal64.exe",
+    "TARGET_SYMBOL": "XAGUSDm",
+    "EXECUTION_SYMBOL": "XAGUSDm",
+    "TARGET_PREFIX": "XAGUSD",
+    "CONFIG_ID": "CFG_XAG_ASIAN_V6",
+    "VERSION": "6.0",
+    "HF_CLOUD": {
+        "DATASET_REPO": "dung5k/argo_workspaces",
+        "MODEL_REPO": "dung5k/argo_workspaces",
+        "SYNC_CHUNKS": true
+    },
+    "FEATURE_ENGINEERING": {
+        "\u26a0\ufe0f_STRICT_WARNING_\u26a0\ufe0f": "T\u1ea4T C\u1ea2 MODULE MACRO_FEATURES L\u00c0 B\u1eaeT BU\u1ed8C.",
+        "TP_PCT": 0.006,
+        "SL_PCT": 0.004,
+        "MAX_HOLD_BARS": 20,
+        "FAST_HIT_BARS": 5,
+        "LABEL_MODE": "pct",
+        "PIP_SIZE": 0.001,
+        "lot_size": 0.1,
+        "CRYPTO_MODE": false,
+        "MTF_INPUTS": [
+            {
+                "SYMBOL": "XAGUSDm",
+                "TIMEFRAME": "5min",
+                "WINDOW_SIZE": 30,
+                "FEATURES": [
+                    "log_return_close",
+                    "body_pct",
+                    "bb_width",
+                    "rsi_14_scaled",
+                    "hour_sin",
+                    "hour_cos",
+                    "volatility_index",
+                    "order_flow_imbalance",
+                    "atr_14_scaled"
+                ]
+            },
+            {
+                "SYMBOL": "XAUUSD",
+                "TIMEFRAME": "5min",
+                "WINDOW_SIZE": 30,
+                "FEATURES": [
+                    "log_return_close",
+                    "volatility_index",
+                    "order_flow_imbalance"
+                ]
+            }
+        ],
+        "VOL_REGIME": true,
+        "ORDER_FLOW": true,
+        "STEP_SIZE": 1
+    },
+    "TRAINING": {
+        "D_MODEL": 128,
+        "N_HEAD": 8,
+        "NUM_LAYERS": 2,
+        "BATCH_SIZE": 32,
+        "WARMUP_EPOCHS": 15,
+        "FINETUNE_EPOCHS": 80,
+        "LEARNING_RATE": 1e-05,
+        "DROPOUT_RATE": 0.3,
+        "LR_SCHEDULER": "cosine_warm",
+        "POOLING": "mean",
+        "TRAIN_START": "2025-01-01",
+        "VAL_END": "2026-04-30"
+    },
+    "MT5_PATHS": {
+        "EXNESS": "D:\\mt5\\MetaTrader 5 EXNESS\\terminal64.exe"
+    },
+    "DATA_SOURCE": {
+        "RAW_LOCAL_DIR": "workspaces/CFG_XAG_ASIAN_V6/data/raw",
+        "DATASET_SUFFIX": "2026",
+        "TIMEFRAME": "M1",
+        "BROKERS": {
+            "EXNESS": "D:\\mt5\\MetaTrader 5 EXNESS\\terminal64.exe"
+        },
+        "ROUTING": {
+            "XAGUSDm": "EXNESS",
+            "XAUUSDm": "EXNESS"
+        }
+    },
+    "LIVE_BOT": {
+        "PAPER_TRADE": false,
+        "TRADE_PLATFORM": "EXNESS",
+        "MAX_ABSOLUTE_MSE": 0.8,
+        "MIN_PROBABILITY_THRESH": 0.59
+    },
+    "SESSION": "asian",
+    "SESSION_UTC": {
+        "START": "00:00",
+        "END": "07:00"
+    },
+    "RUN_ID": "run_20260531_103407_v6_asian",
+    "TRAIN": {
+        "LR_INIT": 0.0001
+    }
+}

workspaces/CFG_XAG_ASIAN_V6/runs/run_20260601_221745_v6_asian/prepare_dataset.log

@@ -0,0 +1,41 @@
+
+======================================================================
+🔥 CHUẨN BỊ DỮ LIỆU V6 MTF HETEROGENEOUS: CFG_XAG_ASIAN_V6
+======================================================================
+
+[LOAD] Quét thư mục: workspaces\CFG_XAG_ASIAN_V6\data\raw
+  + XAGUSDM: Đã gộp 1 file → 71,381 nến (từ 2024)
+  + XAUUSD: Đã gộp 1 file → 71,346 nến (từ 2024)
+  ✅ df_raw tổng hợp: 71,569 dòng × 14 cột
+
+[1] Gắn nhãn Triple-Barrier trên khung Base (1m) cho XAGUSDm...
+  Phân bố TRAIN sau Downsample: {2: 1045, 0: 1045, 1: 700}
+  Phân bố TEST (Giữ nguyên thực tế): {2: 11401, 1: 17, 0: 16}
+
+[2] Feature Engineering Độc Lập Cho Từng MTF_INPUTS...
+  -> Nhánh 0: XAGUSDm | 5min | Window 30
+     [Cảnh báo] Feature 'volatility_index' không tồn tại. Bỏ qua.
+     [Cảnh báo] Feature 'order_flow_imbalance' không tồn tại. Bỏ qua.
+     [Cảnh báo] Feature 'atr_14_scaled' không tồn tại. Bỏ qua.
+     Đã lọc 6 features: ['log_return_close', 'body_pct', 'bb_width']...
+     [DEBUG SHAPE] len(df_f) = 14346
+  -> Nhánh 1: XAUUSD | 5min | Window 30
+     [Cảnh báo] Feature 'volatility_index' không tồn tại. Bỏ qua.
+     [Cảnh báo] Feature 'order_flow_imbalance' không tồn tại. Bỏ qua.
+     Đã lọc 1 features: ['log_return_close']...
+     [DEBUG SHAPE] len(df_f) = 14346
+
+[DEBUG] Bắt đầu ráp Tensor MTF với window_sizes=[30, 30], step_size=1, weekly_split=True
+[DEBUG] max_idx (số lượng nến Base có thể dùng): 71539
+
+  Bỏ qua (ngoài session): 49,745
+  Bỏ qua (clean mask):    16,726
+  Bỏ qua (NaN):           8
+  Bỏ qua (Embargo):       0
+  Giữ lại (Valid Samples):5,060
+
+👉 KẾT QUẢ: Y=(5060,)
+  X_tf0 (XAGUSDm-5min): (5060, 30, 6)
+  X_tf1 (XAUUSD-5min): (5060, 30, 1)
+
+✅ Đã lưu tensor MTF Heterogeneous cục bộ tại: workspaces\CFG_XAG_ASIAN_V6\data\tensors/

workspaces/CFG_XAG_ASIAN_V6/runs/run_20260601_221745_v6_asian/results/config.json

@@ -0,0 +1,100 @@
+{
+    "HF_RUN_ID": "run_20260601_221745_v6_asian",
+    "MT5_PATH": "D:\\mt5\\MetaTrader 5 EXNESS\\terminal64.exe",
+    "TARGET_SYMBOL": "XAGUSDm",
+    "EXECUTION_SYMBOL": "XAGUSDm",
+    "TARGET_PREFIX": "XAGUSD",
+    "CONFIG_ID": "CFG_XAG_ASIAN_V6",
+    "VERSION": "6.0",
+    "HF_CLOUD": {
+        "DATASET_REPO": "dung5k/argo_workspaces",
+        "MODEL_REPO": "dung5k/argo_workspaces",
+        "SYNC_CHUNKS": true
+    },
+    "FEATURE_ENGINEERING": {
+        "\u26a0\ufe0f_STRICT_WARNING_\u26a0\ufe0f": "T\u1ea4T C\u1ea2 MODULE MACRO_FEATURES L\u00c0 B\u1eaeT BU\u1ed8C.",
+        "TP_PCT": 0.006,
+        "SL_PCT": 0.004,
+        "MAX_HOLD_BARS": 20,
+        "FAST_HIT_BARS": 5,
+        "LABEL_MODE": "pct",
+        "PIP_SIZE": 0.001,
+        "lot_size": 0.1,
+        "CRYPTO_MODE": false,
+        "MTF_INPUTS": [
+            {
+                "SYMBOL": "XAGUSDm",
+                "TIMEFRAME": "5min",
+                "WINDOW_SIZE": 30,
+                "FEATURES": [
+                    "log_return_close",
+                    "body_pct",
+                    "bb_width",
+                    "rsi_14_scaled",
+                    "hour_sin",
+                    "hour_cos",
+                    "volatility_index",
+                    "order_flow_imbalance",
+                    "atr_14_scaled"
+                ]
+            },
+            {
+                "SYMBOL": "XAUUSD",
+                "TIMEFRAME": "5min",
+                "WINDOW_SIZE": 30,
+                "FEATURES": [
+                    "log_return_close",
+                    "volatility_index",
+                    "order_flow_imbalance"
+                ]
+            }
+        ],
+        "VOL_REGIME": true,
+        "ORDER_FLOW": true,
+        "STEP_SIZE": 1
+    },
+    "TRAINING": {
+        "D_MODEL": 128,
+        "N_HEAD": 8,
+        "NUM_LAYERS": 2,
+        "BATCH_SIZE": 32,
+        "WARMUP_EPOCHS": 15,
+        "FINETUNE_EPOCHS": 80,
+        "LEARNING_RATE": 1e-05,
+        "DROPOUT_RATE": 0.3,
+        "LR_SCHEDULER": "cosine_warm",
+        "POOLING": "mean",
+        "TRAIN_START": "2025-01-01",
+        "VAL_END": "2026-04-30"
+    },
+    "MT5_PATHS": {
+        "EXNESS": "D:\\mt5\\MetaTrader 5 EXNESS\\terminal64.exe"
+    },
+    "DATA_SOURCE": {
+        "RAW_LOCAL_DIR": "workspaces/CFG_XAG_ASIAN_V6/data/raw",
+        "DATASET_SUFFIX": "2026",
+        "TIMEFRAME": "M1",
+        "BROKERS": {
+            "EXNESS": "D:\\mt5\\MetaTrader 5 EXNESS\\terminal64.exe"
+        },
+        "ROUTING": {
+            "XAGUSDm": "EXNESS",
+            "XAUUSDm": "EXNESS"
+        }
+    },
+    "LIVE_BOT": {
+        "PAPER_TRADE": false,
+        "TRADE_PLATFORM": "EXNESS",
+        "MAX_ABSOLUTE_MSE": 0.8,
+        "MIN_PROBABILITY_THRESH": 0.59
+    },
+    "SESSION": "asian",
+    "SESSION_UTC": {
+        "START": "00:00",
+        "END": "07:00"
+    },
+    "RUN_ID": "run_20260531_103407_v6_asian",
+    "TRAIN": {
+        "LR_INIT": 0.0001
+    }
+}

workspaces/CFG_XAG_ASIAN_V6/runs/run_20260601_221745_v6_asian/results/train_v3.log

@@ -0,0 +1,5 @@
+[PACK] Kèm theo scaler file vào: D:\DungLA\Argo\workspaces\CFG_XAG_ASIAN_V6\runs\run_20260601_221745_v6_asian\brains
+--- 🚀 BẮT ĐẦU VÒNG LẶP FINE-TUNING ĐA NHIỆM (Infinite Loop) ---
+[Epoch 1] Loss(MSE:0.0353/CE:1.6596) | LR=1.00e-05 | Val V6 Simulator Score [0.000] - Loss(MSE:0.0470/CE:1.7200) |   >=53%: WR=0.0% | Score=0.000 | N=30 (30B/0S Bal:0.00) | >=54%: WR=0.0% | Score=0.000 | N=20 (20B/0S Bal:0.00) | >=55%: WR=0.0% | Score=0.000 | N=5 (5B/0S Bal:0.00) | >=56%: WR=0.0% | Score=0.000 | N=5 (5B/0S Bal:0.00)
+  [ARGO2] MO HINH HOI TU TOT NHAT! Val CE Loss = 1.7647 (WR: 0.0%). Luu model...
+[22:19:03] PUSH RUN: run_20260601_221745_v6_asian -> HF (new/changed files only)...

workspaces/CFG_XAG_ASIAN_V6/runs/run_20260601_221745_v6_asian/results/training_metrics_v3.json

@@ -0,0 +1,86 @@
+{
+    "target": "xagusd",
+    "version": "Transformer_V3",
+    "dimensions": {
+        "num_features_target": 0,
+        "num_features_macro": 38
+    },
+    "sessions": {
+        "asian": {
+            "BEST_VLOSS": {
+                "epoch": 1,
+                "max_threshold": 0.56,
+                "composite_score": 0.0,
+                "val_loss": 1.7669982618215132,
+                "threshold_metrics": [
+                    {
+                        "threshold": 0.53,
+                        "total_signals": 30,
+                        "win_rate": 0.0,
+                        "avg_win_return": 0.001,
+                        "avg_loss_return": 0.001,
+                        "ev_score": 0.0,
+                        "sharpe_score": 0.0,
+                        "tus_score": 0.0,
+                        "total_buy": 30,
+                        "total_sell": 0
+                    },
+                    {
+                        "threshold": 0.54,
+                        "total_signals": 20,
+                        "win_rate": 0.0,
+                        "avg_win_return": 0.001,
+                        "avg_loss_return": 0.001,
+                        "ev_score": 0.0,
+                        "sharpe_score": 0.0,
+                        "tus_score": 0.0,
+                        "total_buy": 20,
+                        "total_sell": 0
+                    },
+                    {
+                        "threshold": 0.55,
+                        "total_signals": 5,
+                        "win_rate": 0.0,
+                        "avg_win_return": 0.001,
+                        "avg_loss_return": 0.001,
+                        "ev_score": 0.0,
+                        "sharpe_score": 0.0,
+                        "tus_score": 0.0,
+                        "total_buy": 5,
+                        "total_sell": 0
+                    },
+                    {
+                        "threshold": 0.56,
+                        "total_signals": 5,
+                        "win_rate": 0.0,
+                        "avg_win_return": 0.001,
+                        "avg_loss_return": 0.001,
+                        "ev_score": 0.0,
+                        "sharpe_score": 0.0,
+                        "tus_score": 0.0,
+                        "total_buy": 5,
+                        "total_sell": 0
+                    }
+                ],
+                "win_rates": [
+                    0.0,
+                    0.0,
+                    0.0,
+                    0.0
+                ],
+                "thresholds": [
+                    0.53,
+                    0.54,
+                    0.55,
+                    0.56
+                ],
+                "totals": [
+                    30,
+                    20,
+                    5,
+                    5
+                ]
+            }
+        }
+    }
+}

workspaces/CFG_XAG_ASIAN_V6/runs/run_20260601_221745_v6_asian/train_v6.log

@@ -0,0 +1,36 @@
+Loading config from: workspaces/CFG_XAG_ASIAN_V6\runs\run_20260601_221745_v6_asian\config.json
+Loaded run-specific overrides from: D:\DungLA\Argo\workspaces\CFG_XAG_ASIAN_V6\runs\run_20260601_221745_v6_asian\config.json
+☁️ Đang tải Dataset Tensor từ HuggingFace HUB...
+✅ Tensor đã sẵn sàng!
+✅ Tải thành công! Y: (5060,), T: (5060,)
+   X_tf0: (5060, 30, 6)
+   X_tf1: (5060, 30, 1)
+[DATA CHECK] X abs_max=15.0000 | abs_mean=0.5260
+✅ Dữ liệu đã được scale chuẩn (abs_max < 100).
+[DATA CHECK] Phân bố nhãn Y (Tổng): {0: 339, 1: 210, 2: 4511}
+[DATA SPLIT] Dùng S_tensor (Chronological Split trước khi Downsample) để chia Train/Val.
+[DATA CHECK] Train: 3501 mẫu | Val: 1559 mẫu (Giữ nguyên thực tế, không downsample)
+[CLASS WEIGHTS] Trọng số cân bằng lớp (tự động): [5.748768472906404, 8.774436090225564, 0.3687203791469194]
+💻 Đang Train trên nền tảng: cpu
+[MODEL] Init with: d_model=128, nhead=8, num_layers=2, dropout=0.3, d_ff=512
+
+[INHERIT] Bỏ qua kế thừa theo cờ --scratch. Đào tạo mới hoàn toàn!
+✅ [TELEGRAM] Đã gửi thông báo khởi động!
+[LOSS] Khởi tạo AAMT_JointLoss với focal_gamma = 0.0 | mse_gate = 0.0 | lambda_recon = 1.0 | label_smoothing = 0.15
+[SCHEDULER] CosineAnnealingWarmRestarts (T_0=15, T_mult=2)
+--- 🚀 BẮT ĐẦU WARM-UP AUTOENCODER (10 Epochs) ---
+[Warm-up] Epoch 1/10 | Recon_MSE_Loss: 0.298913
+[Warm-up] Epoch 2/10 | Recon_MSE_Loss: 0.148549
+[Warm-up] Epoch 3/10 | Recon_MSE_Loss: 0.099415
+[Warm-up] Epoch 4/10 | Recon_MSE_Loss: 0.077255
+[Warm-up] Epoch 5/10 | Recon_MSE_Loss: 0.063904
+[Warm-up] Epoch 6/10 | Recon_MSE_Loss: 0.054706
+[Warm-up] Epoch 7/10 | Recon_MSE_Loss: 0.047966
+[Warm-up] Epoch 8/10 | Recon_MSE_Loss: 0.043042
+[Warm-up] Epoch 9/10 | Recon_MSE_Loss: 0.038782
+[Warm-up] Epoch 10/10 | Recon_MSE_Loss: 0.035690
+[PACK] Kèm theo scaler file vào: D:\DungLA\Argo\workspaces\CFG_XAG_ASIAN_V6\runs\run_20260601_221745_v6_asian\brains
+--- 🚀 BẮT ĐẦU VÒNG LẶP FINE-TUNING ĐA NHIỆM (Infinite Loop) ---
+[Epoch 1] Loss(MSE:0.0353/CE:1.6596) | LR=1.00e-05 | Val V6 Simulator Score [0.000] - Loss(MSE:0.0470/CE:1.7200) |   >=53%: WR=0.0% | Score=0.000 | N=30 (30B/0S Bal:0.00) | >=54%: WR=0.0% | Score=0.000 | N=20 (20B/0S Bal:0.00) | >=55%: WR=0.0% | Score=0.000 | N=5 (5B/0S Bal:0.00) | >=56%: WR=0.0% | Score=0.000 | N=5 (5B/0S Bal:0.00)
+  [ARGO2] MO HINH HOI TU TOT NHAT! Val CE Loss = 1.7647 (WR: 0.0%). Luu model...
+[22:19:03] PUSH RUN: run_20260601_221745_v6_asian -> HF (new/changed files only)...