diff --git a/strategies/kdj_cross_trailing.kh b/strategies/kdj_cross_trailing.kh
new file mode 100644
index 0000000..2203523
--- /dev/null
+++ b/strategies/kdj_cross_trailing.kh
@@ -0,0 +1,70 @@
+{
+    "system": {
+        "userdata_path": "I:/QMT/userdata_mini"
+    },
+    "run_mode": "backtest",
+    "account": {
+        "account_id": "88888888",
+        "account_type": "STOCK"
+    },
+    "strategy_file": "C:/Users/Administrator/Documents/GitHub/OSkhQuant/strategies/kdj_cross_trailing.py",
+    "data_mode": "custom",
+    "backtest": {
+        "start_time": "20250101",
+        "end_time": "20251101",
+        "init_capital": 1000000.0,
+        "min_volume": 100,
+        "benchmark": "sh.000300",
+        "trade_cost": {
+            "min_commission": 5.0,
+            "commission_rate": 0.0001,
+            "stamp_tax_rate": 0.0005,
+            "flow_fee": 0.0,
+            "slippage": {
+                "type": "ratio",
+                "tick_size": 0.01,
+                "tick_count": 2,
+                "ratio": 0.01
+            }
+        },
+        "trigger": {
+            "type": "1d",
+            "custom_times": [
+                "09:30:00"
+            ],
+            "start_time": "09:30:00",
+            "end_time": "15:00:00",
+            "interval": 300
+        }
+    },
+    "data": {
+        "kline_period": "1d",
+        "dividend_type": "front",
+        "fields": [
+            "open",
+            "high",
+            "low",
+            "close",
+            "volume",
+            "amount",
+            "settelementPrice",
+            "openInterest",
+            "preClose",
+            "suspendFlag"
+        ],
+        "stock_list": [
+            "000001.SZ"
+        ]
+    },
+    "market_callback": {
+        "pre_market_enabled": false,
+        "pre_market_time": "08:30:00",
+        "post_market_enabled": false,
+        "post_market_time": "15:30:00"
+    },
+    "risk": {
+        "position_limit": 0.95,
+        "order_limit": 100,
+        "loss_limit": 0.1
+    }
+}
diff --git a/strategies/kdj_cross_trailing.py b/strategies/kdj_cross_trailing.py
new file mode 100644
index 0000000..6079088
--- /dev/null
+++ b/strategies/kdj_cross_trailing.py
@@ -0,0 +1,185 @@
+# coding: utf-8
+"""MACD trend strategy with MA filter and layered loss control."""
+import logging
+from typing import Dict, List
+
+from khQuantImport import *  # noqa: F401,F403
+from MyTT import MA, ATR
+
+logger = logging.getLogger("macd_trend_strategy")
+
+MIN_HISTORY_BARS = 60
+SHORT_EMA = 12
+LONG_EMA = 26
+SIGNAL = 9
+BUY_RATIO = 0.95
+MAX_POSITION_RATIO = 0.05
+ATR_PERIOD = 14
+TRAIL_ATR_MULT = 1.0
+TRAIL_DROP_PCT = 0.03
+PARTIAL_STOP_RATIO = 0.5
+STATE_ATTR = "_macd_trend_state"
+
+
+def _calc_buy_ratio(data: Dict, price: float) -> float:
+    cash = khGet(data, "cash") or 0.0
+    total = khGet(data, "total_asset") or 0.0
+    if cash <= 0 or total <= 0 or price <= 0:
+        return 0.0
+    intended = cash * BUY_RATIO
+    per_stock_cap = total * MAX_POSITION_RATIO
+    order_cash = min(intended, per_stock_cap)
+    if order_cash <= 0:
+        return 0.0
+    return max(0.0, min(order_cash / cash, 1.0))
+
+
+def init(stocks=None, data=None):  # pragma: no cover
+    pass
+
+
+def khHandlebar(data: Dict) -> List[Dict]:
+    signals: List[Dict] = []
+    stock_list = khGet(data, "stocks") or data.get("stocks") or []
+    date_num = khGet(data, "date_num") or data.get("date_num")
+    if not stock_list or not date_num:
+        return signals
+
+    positions = data.get("__positions__", {}) or {}
+    position_count = sum(
+        1
+        for info in positions.values()
+        if info and info.get("volume", 0) > 0
+    )
+    logger.info(
+        "positions on %s: %s",
+        khGet(data, "date") or khGet(data, "date_str") or date_num,
+        position_count,
+    )
+    state = _ensure_state(data)
+    position_states = state.setdefault("positions", {})
+    _cleanup_state(position_states, positions)
+
+    try:
+        history = khHistory(
+            symbol_list=stock_list,
+            fields=["close", "high", "low"],
+            bar_count=MIN_HISTORY_BARS,
+            fre_step="1d",
+            current_time=date_num,
+            fq="pre",
+            force_download=False,
+        )
+    except Exception as err:  # pragma: no cover
+        logger.warning("failed to load history: %s", err)
+        return signals
+
+    for code in stock_list:
+        df = history.get(code)
+        if df is None or len(df) < MIN_HISTORY_BARS:
+            continue
+        try:
+            close = df["close"].values
+        except KeyError:
+            continue
+        highs = df.get("high").values if "high" in df else None
+        lows = df.get("low").values if "low" in df else None
+        if len(close) < 60:
+            continue
+        ma20 = MA(close, 20)[-1]
+        ma60 = MA(close, 60)[-1]
+        price = float(close[-1])
+        if price <= ma60 or ma20 <= ma60:
+            continue
+        atr_value = _calc_atr(close, highs, lows)
+        has_pos = khHas(data, code)
+        if has_pos:
+            risk_sells = _apply_loss_controls(
+                data,
+                code,
+                price,
+                atr_value,
+                positions.get(code) or {},
+                position_states,
+            )
+            if risk_sells:
+                signals.extend(risk_sells)
+                continue
+
+        if not has_pos:
+            ratio = _calc_buy_ratio(data, price)
+            if ratio <= 0:
+                continue
+            reason = f"{code[:6]} trend buy"
+            signals.extend(generate_signal(data, code, price, ratio, "buy", reason))
+    return signals
+
+
+def _ensure_state(data: Dict) -> Dict:
+    framework = data.get("__framework__")
+    if framework is not None:
+        state = getattr(framework, STATE_ATTR, None)
+        if state is None:
+            state = {}
+            setattr(framework, STATE_ATTR, state)
+        return state
+    return data.setdefault(STATE_ATTR, {})
+
+
+def _cleanup_state(position_states: Dict, positions: Dict) -> None:
+    for code in list(position_states.keys()):
+        if code not in positions:
+            position_states.pop(code, None)
+
+
+def _calc_atr(close, highs, lows) -> float:
+    if close is None or highs is None or lows is None:
+        return 0.0
+    try:
+        atr_series = ATR(close, highs, lows, ATR_PERIOD)
+        if atr_series is None or len(atr_series) == 0:
+            return 0.0
+        return float(atr_series[-1])
+    except Exception as err:  # pragma: no cover
+        logger.debug("ATR calc failed: %s", err)
+        return 0.0
+
+
+def _apply_loss_controls(
+    data: Dict,
+    code: str,
+    price: float,
+    atr_value: float,
+    position_info: Dict,
+    position_states: Dict,
+) -> List[Dict]:
+    entry_price = float(position_info.get("avg_price") or 0.0)
+    if entry_price <= 0:
+        return []
+    state = position_states.setdefault(
+        code, {"highest": entry_price, "reduced_once": False}
+    )
+    highest = max(float(state.get("highest", entry_price)), price)
+    state["highest"] = highest
+
+    reason = ""
+    trail_by_pct = highest * (1 - TRAIL_DROP_PCT)
+    trail_by_atr = highest - (TRAIL_ATR_MULT * atr_value if atr_value else 0.0)
+    trailing_price = max(trail_by_pct, trail_by_atr)
+    if price <= trailing_price:
+        reason = f"{code[:6]} trail stop {price:.2f}<={trailing_price:.2f}"
+    if not reason:
+        return []
+
+    reduced_once = state.get("reduced_once", False)
+    ratio = 1.0 if reduced_once else PARTIAL_STOP_RATIO
+    if ratio <= 0:
+        return []
+    sell_signals = generate_signal(data, code, price, ratio, "sell", reason)
+    if not sell_signals:
+        return []
+    if reduced_once or ratio >= 1.0:
+        position_states.pop(code, None)
+    else:
+        state["reduced_once"] = True
+    return sell_signals