Document local Aurum-1D inference

README.md

@@ -50,7 +50,7 @@ Gold outputs a return estimate.
 - `config.json`: data, feature, split and transaction-cost configuration.
 - `metrics.json`: validation, locked-test, foundation-model and robustness results.
 
-The payload contains no raw Yahoo Finance/FRED data, Hugging Face token or intermediate checkpoint. The estimator was fitted with code revision `d5f5f5f`.
+The payload contains no raw Yahoo Finance/FRED data, Hugging Face token or intermediate checkpoint. The estimator was fitted with code revision `2983c13`.
 
 ## Evaluation under the repository protocol
 
@@ -94,25 +94,40 @@ DM compares forecast loss on the same locked dates; `Holm p` corrects the four f
 
 The candidate-aware White Reality Check for the full local-plus-foundation universe is `0.106` for Gold. This is a conditional bootstrap p-value, not a probability that the model will be profitable.
 
-## Reproduce inference
+## Reproduce inference locally
 
-The payload intentionally does not include raw market data. Construct the latest causal feature row with the source repository, preserve the schema order, then load the estimator:
+This is local inference only: no API, endpoint or Space is required. The payload intentionally does not include raw market data. Install the public source repository, download the current cache, reconstruct the latest causal feature row, preserve the schema order, then load the estimator:
 
 ```python
 import json
+import math
 from pathlib import Path
 
 import joblib
-
-ARTIFACT_DIR = Path(".")
-estimator = joblib.load(ARTIFACT_DIR / "model.joblib")
-schema = json.loads((ARTIFACT_DIR / "feature_schema.json").read_text())
-
-# Build this one-row DataFrame with the causal feature builder in the source
-# repository. It must contain exactly the columns listed in schema.
-X_next = build_causal_feature_row(raw_history)[schema["feature_columns"]]
-next_day_log_return = float(estimator.predict(X_next)[0])
-print(next_day_log_return)
+from huggingface_hub import hf_hub_download
+
+from gold_silver.config import load_config
+from gold_silver.data import load_cached_market_data
+from gold_silver.features import build_features
+
+# Run these commands once from the public source repository first:
+#   pip install -e .
+#   python scripts/download_data.py
+model = joblib.load(hf_hub_download("AurelPx/Aurum-1D", "model.joblib"))
+schema = json.loads(Path(
+    hf_hub_download("AurelPx/Aurum-1D", "feature_schema.json")
+).read_text())
+
+config = load_config("configs/default.yaml")
+history = load_cached_market_data(config)
+features = build_features(history, config)
+X_next = features[schema["feature_columns"]].tail(1)
+next_day_log_return = float(model.predict(X_next)[0])
+last_close = float(history["gold_close"].dropna().iloc[-1])
+print({
+    "next_day_log_return": next_day_log_return,
+    "implied_next_close": last_close * math.exp(next_day_log_return),
+})
 ```
 
 ## Data and reproducibility

estimator.joblib

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:06d7bf71c8426e7857a92d14d8d4f6319333cb5ba429a58a0786c8131179c943
+oid sha256:2381635b9bec90c8c26d2e2728a8dc549e06ee040e9021749b642faf10cb4611
 size 843665

metrics.json

@@ -251,5 +251,5 @@
       "btc": "BTC-USD"
     }
   },
-  "source_code_revision": "d5f5f5f"
+  "source_code_revision": "2983c13"
 }

model.joblib

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:06d7bf71c8426e7857a92d14d8d4f6319333cb5ba429a58a0786c8131179c943
+oid sha256:2381635b9bec90c8c26d2e2728a8dc549e06ee040e9021749b642faf10cb4611
 size 843665

requirements.txt

@@ -1,3 +1,4 @@
+huggingface_hub>=0.30
 joblib>=1.3
 numpy>=1.26
 pandas>=2.2