{ "deepseek_self_agreement": { "accuracy": 0.7, "confusion": [ [ 4, 0, 1 ], [ 0, 0, 0 ], [ 1, 1, 3 ] ], "directional_accuracy": 0.7, "directional_hits": 7, "directional_n": 10, "hits": 7, "identical": 2, "macro_f1": 0.48888888888888893, "mae": 0.25, "n": 10, "sign_agreement": 0.7777777777777778, "sign_hits": 7, "sign_n": 9, "spearman": 0.5569257132640542 }, "generated_at": "20260917_2241", "latency": { "deepseek_ms_per_headline": 56.414902748947796, "model_ms_mean": 1.0686842105263157 }, "log_rows": 76, "n": 76, "subsets": { "all headlines": { "keyword": { "accuracy": 0.3815789473684211, "confusion": [ [ 9, 17, 0 ], [ 0, 16, 0 ], [ 2, 28, 4 ] ], "directional_accuracy": 0.21666666666666667, "directional_hits": 13, "directional_n": 60, "hits": 29, "macro_f1": 0.37086573928679195, "n": 76, "sign_agreement": 0.8666666666666667, "sign_hits": 13, "sign_n": 15 }, "model_full": { "accuracy": 0.6052631578947368, "confusion": [ [ 16, 3, 7 ], [ 5, 6, 5 ], [ 7, 3, 24 ] ], "directional_accuracy": 0.6666666666666666, "directional_hits": 40, "directional_n": 60, "hits": 46, "macro_f1": 0.5689594356261023, "mae": 0.33246710526315787, "n": 76, "sign_agreement": 0.7407407407407407, "sign_hits": 40, "sign_n": 54, "spearman": 0.4808996156923549 }, "model_head": { "accuracy": 0.6842105263157895, "confusion": [ [ 20, 0, 6 ], [ 2, 6, 8 ], [ 6, 2, 26 ] ], "directional_accuracy": 0.7666666666666667, "directional_hits": 46, "directional_n": 60, "hits": 52, "macro_f1": 0.6478144811478145, "mae": 0.3404934210526316, "n": 76, "sign_agreement": 0.7931034482758621, "sign_hits": 46, "sign_n": 58, "spearman": 0.5639368037865334 }, "n": 76 }, "mentions gold (XAU keyword)": { "keyword": { "accuracy": 0.3404255319148936, "confusion": [ [ 5, 6, 0 ], [ 0, 9, 0 ], [ 0, 25, 2 ] ], "directional_accuracy": 0.18421052631578946, "directional_hits": 7, "directional_n": 38, "hits": 16, "macro_f1": 0.37675932441942295, "n": 47, "sign_agreement": 1.0, "sign_hits": 7, "sign_n": 7 }, "model_full": { "accuracy": 0.723404255319149, "confusion": [ [ 10, 0, 1 ], [ 2, 4, 3 ], [ 4, 3, 20 ] ], "directional_accuracy": 0.7894736842105263, "directional_hits": 30, "directional_n": 38, "hits": 34, "macro_f1": 0.6750181554103122, "mae": 0.2999148936170213, "n": 47, "sign_agreement": 0.8571428571428571, "sign_hits": 30, "sign_n": 35, "spearman": 0.5816073836387256 }, "model_head": { "accuracy": 0.7659574468085106, "confusion": [ [ 10, 0, 1 ], [ 2, 4, 3 ], [ 3, 2, 22 ] ], "directional_accuracy": 0.8421052631578947, "directional_hits": 32, "directional_n": 38, "hits": 36, "macro_f1": 0.7109175939364617, "mae": 0.302268085106383, "n": 47, "sign_agreement": 0.8888888888888888, "sign_hits": 32, "sign_n": 36, "spearman": 0.6912522895796331 }, "n": 47 }, "not in training & mentions gold": { "keyword": { "accuracy": 0.35135135135135137, "confusion": [ [ 3, 3, 0 ], [ 0, 8, 0 ], [ 0, 21, 2 ] ], "directional_accuracy": 0.1724137931034483, "directional_hits": 5, "directional_n": 29, "hits": 13, "macro_f1": 0.40888888888888886, "n": 37, "sign_agreement": 1.0, "sign_hits": 5, "sign_n": 5 }, "model_full": { "accuracy": 0.7027027027027027, "confusion": [ [ 6, 0, 0 ], [ 2, 4, 2 ], [ 4, 3, 16 ] ], "directional_accuracy": 0.7586206896551724, "directional_hits": 22, "directional_n": 29, "hits": 26, "macro_f1": 0.6601626016260163, "mae": 0.25864864864864867, "n": 37, "sign_agreement": 0.8461538461538461, "sign_hits": 22, "sign_n": 26, "spearman": 0.5732166454312866 }, "model_head": { "accuracy": 0.7567567567567568, "confusion": [ [ 6, 0, 0 ], [ 2, 4, 2 ], [ 3, 2, 18 ] ], "directional_accuracy": 0.8275862068965517, "directional_hits": 24, "directional_n": 29, "hits": 28, "macro_f1": 0.7048400755651097, "mae": 0.2563864864864865, "n": 37, "sign_agreement": 0.8888888888888888, "sign_hits": 24, "sign_n": 27, "spearman": 0.6927913852069925 }, "n": 37 }, "not in training set": { "keyword": { "accuracy": 0.3939393939393939, "confusion": [ [ 7, 14, 0 ], [ 0, 15, 0 ], [ 2, 24, 4 ] ], "directional_accuracy": 0.21568627450980393, "directional_hits": 11, "directional_n": 51, "hits": 26, "macro_f1": 0.3810457516339869, "n": 66, "sign_agreement": 0.8461538461538461, "sign_hits": 11, "sign_n": 13 }, "model_full": { "accuracy": 0.5757575757575758, "confusion": [ [ 12, 3, 6 ], [ 5, 6, 4 ], [ 7, 3, 20 ] ], "directional_accuracy": 0.6274509803921569, "directional_hits": 32, "directional_n": 51, "hits": 38, "macro_f1": 0.5481481481481482, "mae": 0.3142651515151515, "n": 66, "sign_agreement": 0.7111111111111111, "sign_hits": 32, "sign_n": 45, "spearman": 0.4561157459616647 }, "model_head": { "accuracy": 0.6666666666666666, "confusion": [ [ 16, 0, 5 ], [ 2, 6, 7 ], [ 6, 2, 22 ] ], "directional_accuracy": 0.7450980392156863, "directional_hits": 38, "directional_n": 51, "hits": 44, "macro_f1": 0.6401167471819647, "mae": 0.3205636363636364, "n": 66, "sign_agreement": 0.7755102040816326, "sign_hits": 38, "sign_n": 49, "spearman": 0.5415716725239331 }, "n": 66 } }, "threshold": 0.05, "usd_only_display_vs_xau": { "accuracy": 0.875, "confusion": [ [ 9, 0, 0 ], [ 0, 3, 0 ], [ 2, 0, 2 ] ], "directional_accuracy": 0.8461538461538461, "directional_hits": 11, "directional_n": 13, "hits": 14, "macro_f1": 0.8555555555555555, "n": 16, "sign_agreement": 0.8461538461538461, "sign_hits": 11, "sign_n": 13 } }