Harden trading, training, and monitoring

crypto_spot_bot/time_series.py

@@ -4,6 +4,7 @@ import json
 import math
 from bisect import bisect_right
 from dataclasses import asdict, dataclass, field
+from datetime import UTC, datetime
 from typing import Any
 
 from crypto_spot_bot.config import Settings
@@ -156,6 +157,9 @@ class TimeSeriesForecast:
     candidates: list[dict[str, Any]] = field(default_factory=list)
     quality_gate_passed: bool | None = None
     quality_gate: dict[str, Any] = field(default_factory=dict)
+    model_created_at: str = ""
+    model_age_hours: float | None = None
+    model_fresh: bool = False
 
     def as_dict(self) -> dict[str, Any]:
         return asdict(self)
@@ -188,6 +192,10 @@ class TimeSeriesForecaster:
             return _empty_forecast(True, "not enough returns for PyTorch forecast")
 
         artifact = self._load_lstm_artifact()
+        model_created_at, model_age_hours, model_fresh = _model_freshness(
+            artifact,
+            self.settings.time_series_model_max_age_hours,
+        )
         quality_gate = self._load_quality_gate()
         quality_gate_passed = _quality_gate_passed(quality_gate)
         entry = _torch_recurrent_entry(symbol, artifact)
@@ -288,6 +296,9 @@ class TimeSeriesForecaster:
                 candidates=[{"model": model, "mae_percent": round(model_mae * 100, 4)}],
                 quality_gate_passed=quality_gate_passed,
                 quality_gate=quality_gate,
+                model_created_at=model_created_at,
+                model_age_hours=model_age_hours,
+                model_fresh=model_fresh,
             )
 
         direct_horizon = _is_direct_horizon(entry)
@@ -350,6 +361,9 @@ class TimeSeriesForecaster:
             candidates=[{"model": model, "mae_percent": round(model_mae * 100, 4)}],
             quality_gate_passed=quality_gate_passed,
             quality_gate=quality_gate,
+            model_created_at=model_created_at,
+            model_age_hours=model_age_hours,
+            model_fresh=model_fresh,
         )
 
     def _load_lstm_artifact(self) -> dict[str, Any]:
@@ -420,6 +434,9 @@ def _empty_forecast(enabled: bool, reason: str) -> TimeSeriesForecast:
         candidates=[],
         quality_gate_passed=None,
         quality_gate={},
+        model_created_at="",
+        model_age_hours=None,
+        model_fresh=False,
     )
 
 
@@ -436,6 +453,20 @@ def _quality_gate_passed(quality_gate: dict[str, Any]) -> bool | None:
     return None
 
 
+def _model_freshness(artifact: dict[str, Any], max_age_hours: float) -> tuple[str, float | None, bool]:
+    raw = str(artifact.get("created_at", "")).strip() if isinstance(artifact, dict) else ""
+    if not raw:
+        return "", None, False
+    try:
+        created_at = datetime.fromisoformat(raw.replace("Z", "+00:00"))
+    except ValueError:
+        return raw, None, False
+    if created_at.tzinfo is None:
+        created_at = created_at.replace(tzinfo=UTC)
+    age_hours = max(0.0, (datetime.now(UTC) - created_at.astimezone(UTC)).total_seconds() / 3600)
+    return raw, round(age_hours, 4), age_hours <= max(0.1, max_age_hours)
+
+
 def _log_returns(closes: list[float]) -> list[float]:
     return [math.log(closes[index] / closes[index - 1]) for index in range(1, len(closes))]