national

ai-engine/utils/national_leagues.py

@@ -0,0 +1,67 @@
+"""
+National-Team League Loader + Competition-Type Classifier
+=========================================================
+Loads the A-milli (senior men's) football league IDs from
+data/national_leagues.json and classifies a league name into a
+competition type. Powers the betting_brain national-match gate.
+
+Why this exists:
+    Backtest (2300 national matches) showed national matches behave very
+    differently from clubs — only the MS market carries edge, and only in
+    the 4.0–7.0 odds band for Hazırlık/Eleme fixtures (tournaments behave
+    inversely). Calibration is fine; the issue is *which* bets to allow.
+    See mds/national-team-strategy.md.
+
+Usage:
+    from utils.national_leagues import load_national_leagues, classify_competition
+    natl = load_national_leagues()         # set[str] of league_ids
+    ctype = classify_competition(name)     # "HAZIRLIK" | "ELEME" | "TURNUVA"
+"""
+
+from __future__ import annotations
+
+import json
+import os
+from typing import Set
+
+
+_DATA_FILE = os.path.join(
+    os.path.dirname(os.path.abspath(__file__)),
+    "..",
+    "data",
+    "national_leagues.json",
+)
+
+
+def load_national_leagues() -> Set[str]:
+    """Return the set of A-milli football league IDs (empty on any failure)."""
+    if not os.path.isfile(_DATA_FILE):
+        print(
+            f"⚠️  national_leagues.json not found at {_DATA_FILE}. "
+            "National-match gate disabled (no league treated as national)."
+        )
+        return set()
+    try:
+        with open(_DATA_FILE, "r", encoding="utf-8") as f:
+            data = json.load(f)
+        ids = set(str(x) for x in (data.get("league_ids") or []))
+        print(f"✅ Loaded {len(ids)} national-team league IDs")
+        return ids
+    except (json.JSONDecodeError, KeyError, TypeError) as exc:
+        print(f"⚠️  Failed to parse national_leagues.json: {exc}")
+        return set()
+
+
+def classify_competition(league_name: str) -> str:
+    """Map a league name to a competition type.
+
+    HAZIRLIK = friendlies, ELEME = qualifiers/play-offs, TURNUVA = finals/cups.
+    The backtest edge lives in HAZIRLIK+ELEME (MS, odds 4-7); TURNUVA is
+    handled conservatively (no bet) by the gate.
+    """
+    n = (league_name or "").lower()
+    if "hazırlık" in n or "hazirlik" in n or "friendl" in n:
+        return "HAZIRLIK"
+    if "eleme" in n or "play-off" in n or "playoff" in n or "qualif" in n:
+        return "ELEME"
+    return "TURNUVA"