fix: 温区级普适改近似计算(温区汇总数据估算,不逐笔遍历)——解决API超时,regime_winrates直接算温区月份占比

server.py

@@ -536,6 +536,24 @@ def get_tracking():
 @app.route("/api/research/strategies")
 def api_research_strategies():
     """策略版本列表（含回测结果摘要，支持 period_tag 区间过滤）"""
+    def _approx_regime_universality(strategy, regime, trades):
+        """温区级普适近似：温区内信号月份≈trades/温区月均笔数，温区总月份占比（避免逐笔遍历性能问题）"""
+        try:
+            import sqlite3 as _sq6
+            _c6 = _sq6.connect(str(DATA_DIR / "mofin.db"), timeout=10)
+            # 温区总月份
+            _rm = {r[0]: r[1] for r in _c6.execute(
+                "SELECT regime, COUNT(DISTINCT substr(date,1,7)) FROM market_regime WHERE market='a' GROUP BY regime").fetchall()}
+            _c6.close()
+            regime_total = _rm.get(regime, 0)
+            if not trades or regime_total == 0:
+                return {"months": 0, "score": 0, "regime_total_months": regime_total}
+            # 近似：温区内信号月份 ≈ trades / (温区月均笔数≈3)，月份占比 = 信号月份/温区总月份
+            est_months = max(1, min(trades // 3, regime_total))
+            score = round(min(est_months / regime_total * 100, 100))
+            return {"months": est_months, "score": score, "regime_total_months": regime_total}
+        except Exception:
+            return {"months": 0, "score": 0, "regime_total_months": 0}
     try:
         from strategy_lab import list_strategies, STRATEGY_DESCRIPTIONS
         pt = request.args.get('period_tag')
@@ -563,13 +581,13 @@ def api_research_strategies():
                     "cagr_pct": _r[7], "max_dd_pct": _r[8], "capital_final": _r[9],
                     "positions_taken": _r[10], "sharpe_ratio": _r[11],
                     "profit_factor": _r[12],
-                    # 2026-08-15 温区级组合级指标（温区行也要显示组合级列）
+                    # 温区级组合级指标（温区行也要显示组合级列）
                     "portfolio": {"cagr_pct": _r[7], "total_return_pct": _r[6],
                                   "portfolio_max_dd_pct": _r[8], "capital_final": _r[9],
                                   "positions_taken": _r[10], "sharpe_ratio": _r[11],
                                   "profit_factor": _r[12]},
-                    # 温区级 universality（该温区 trades 的月份分散度，对齐整体行普适）
-                    "universality": {"months": _r[2] and min(_r[2] // 4, 12) or 0},
+                    # 温区级 universality（近似：温区内信号月份≈trades/温区月均，温区总月份；避免逐笔遍历性能问题）
+                    "universality": _approx_regime_universality(_r[0], _r[1], _r[2]),
                 }
             _c.close()
         except Exception:
@@ -584,7 +602,7 @@ def api_research_strategies():
             _c5.close()
         except Exception:
             pass
-        # 2026-08-15 温区级普适：策略在适应温区内发出信号的月份分散度 + 温区总月份占比
+        # 温区级普适（近似）：温区内信号月份 ≈ trades/温区月均笔数×温区总月份，避免逐笔遍历（性能）
         # 温区总月份数（A股: choppy35.4%/trend_down30.3%/trend_up34.3%；按市场 regime 时长统计）
         try:
             import sqlite3 as _sq3
@@ -597,42 +615,7 @@ def api_research_strategies():
         except Exception:
             _total_months = 0
             _regime_months = {}
-        _regime_universality = {}  # (version, regime) -> {'months': N, 'score': 分散度}
-        try:
-            import sqlite3 as _sq4, json as _json4, collections as _coll
-            _c4 = _sq4.connect(str(DATA_DIR / "mofin.db"), timeout=10)
-            for _r in _c4.execute(
-                "SELECT version, results_json FROM strategy_research WHERE results_json IS NOT NULL"):
-                try:
-                    _res = _json4.loads(_r[1])
-                    _trades = _res.get("trades", [])
-                    if not _trades:
-                        continue
-                    # 按温区过滤该策略 trades（regime_winrates 已有温区归类）
-                    for _rg in ("trend_up", "choppy", "trend_down"):
-                        _rg_months = _coll.Counter()
-                        for _t in _trades:
-                            _ed = _t.get("entry_date", "")
-                            if _ed and _regime_map.get(_ed) == _rg:
-                                _rg_months[_ed[:7]] += 1
-                        if _rg_months:
-                            _n = len(_rg_months)
-                            _tot = sum(_rg_months.values())
-                            _score = 0.0
-                            if _n > 1:
-                                for _c in _rg_months.values():
-                                    _p = _c / _tot
-                                    _score -= _p * (0.0 if _p == 0 else __import__('math').log(_p))
-                                _score = _score / __import__('math').log(_n) * 100
-                            _regime_universality.setdefault(_r[0], {})[_rg] = {
-                                "months": _n, "score": round(_score, 0),
-                                "regime_total_months": _regime_months.get(_rg, 0),
-                            }
-                except Exception:
-                    continue
-            _c4.close()
-        except Exception:
-            _regime_universality = {}
+        # 温区级普适已用近似计算（_approx_regime_universality），不逐笔遍历（性能）
         _deprecated = {}
         try:
             import sqlite3 as _sq2