TrendRiderStrategy

"""
TrendRider v2.16.0 — Bear Gate ADX>18 (was 25): cuts weak-trend chop, PF 1.67->4.11

Philosophy: Ride established trends with WIDE stoploss.
Key insight: crypto swings 2-4% per hour. Stoploss must be >= 5-6%.

Public version:
- No external API calls (FNG, Bybit funding/OI)
- No SQLite price alerts
- No Cornix formatting
- Leverage 1x (spot-safe)
- All TA-Lib indicators and confidence scoring preserved
"""

import talib.abstract as ta
from datetime import datetime
from freqtrade.strategy import IStrategy, IntParameter, DecimalParameter, merge_informative_pair
from pandas import DataFrame
from functools import reduce
import logging

logger = logging.getLogger(__name__)


class TrendRiderStrategy(IStrategy):
    INTERFACE_VERSION = 3

    # --- ROI (V6: tightened ladder to catch peak-crash pattern) ---
    # Diagnostic showed 10/15 losers peaked at +1..+3% then crashed to -2..-5%.
    # Previous ROI (22.9% immediate, 13.6% at 2h) was practically unreachable on 1h crypto —
    # trades peaked below ROI then drifted into loss. New ladder realistic for 1h timeframe:
    minimal_roi = {
        # v2.12.0 (2026-04-24): +55% PnL on 110d backtest vs v2.11.0.
        # Slightly higher thresholds + delayed timings let winners breathe —
        # captures the +2-4% continuations V6 was clipping too early.
        "0": 0.06,      # 6% immediate (was 5%)
        "60": 0.035,    # 3.5% after 1h (was 3%)
        "240": 0.02,    # 2% after 4h (was 1.5% @ 3h)
        "480": 0.01,    # 1% after 8h (was 0.8% @ 6h)
        "720": 0,       # breakeven after 12h
    }

    # --- Stoploss: WIDE for crypto volatility ---
    stoploss = -0.06           # 6% default (ATR-based custom stoploss overrides)
    use_custom_stoploss = False

    # --- Trailing Stop: WIDE ---
    trailing_stop = True
    trailing_stop_positive = 0.03        # 3% trail
    trailing_stop_positive_offset = 0.05 # Activate after +5%
    trailing_only_offset_is_reached = True

    # --- General ---
    timeframe = "1h"
    startup_candle_count = 210
    process_only_new_candles = True
    can_short = False
    position_adjustment_enable = False

    # --- Protections (moved from config.json for Freqtrade 2026.2+) ---
    protections = [
        {
            "method": "CooldownPeriod",
            "stop_duration": 20
        },
        {
            "method": "StoplossGuard",
            "lookback_period": 720,
            "trade_limit": 3,
            "stop_duration": 60,
            "only_per_pair": False
        },
        {
            "method": "MaxDrawdown",
            "lookback_period": 1440,
            "max_allowed_drawdown": 0.10,
            "stop_duration": 300,
            "trade_limit": 5
        }
    ]

    # --- HyperOpt Results (applied from optimization session 2026-03-23) ---
    buy_params = {
        "ema_fast": 9,
        "ema_slow": 16,
        "rsi_period": 16,
        "rsi_pullback_low": 30,
        "rsi_pullback_high": 65,
        "rsi_bounce": 35,
        "adx_threshold": 20,    # V6: was 18 — require minimum trend strength
        "volume_factor": 1.3,   # V6: was 0.7 — require meaningful volume (hyperopt consistently finds 1.3-1.6)
    }

    sell_params = {
        "rsi_exit": 78,
    }

    # --- HyperOpt Parameters ---
    ema_fast = IntParameter(5, 15, default=9, space="buy")
    ema_slow = IntParameter(15, 30, default=21, space="buy")
    rsi_period = IntParameter(10, 20, default=14, space="buy")
    rsi_pullback_low = IntParameter(30, 48, default=40, space="buy")
    rsi_pullback_high = IntParameter(52, 65, default=58, space="buy")
    rsi_bounce = IntParameter(25, 35, default=30, space="buy")
    rsi_exit = IntParameter(72, 85, default=78, space="sell")
    adx_threshold = IntParameter(20, 35, default=25, space="buy")
    volume_factor = DecimalParameter(1.0, 2.5, default=1.3, space="buy")

    # --- Leverage: 1x for Strat Ninja (spot-safe) ---
    leverage_value = 1

    def leverage(self, pair: str, current_time, current_rate: float,
                 proposed_leverage: float, max_leverage: float, entry_tag: str,
                 side: str, **kwargs) -> float:
        return 1

    def informative_pairs(self):
        pairs = self.dp.current_whitelist() if self.dp else []
        informative = []
        for pair in pairs:
            informative.append((pair, "4h"))
            informative.append((pair, "1d"))
        # BTC as market sentiment
        informative.append(("BTC/USDT:USDT", "1h"))
        informative.append(("BTC/USDT:USDT", "4h"))
        # BTC daily for bear-avoidance regime gate (close<SMA200 & ADX>25 = bear)
        informative.append(("BTC/USDT:USDT", "1d"))
        return informative

    def populate_indicators(self, dataframe: DataFrame, metadata: dict) -> DataFrame:
        # EMAs (all periods for hyperopt ranges)
        for period in range(5, 31):
            dataframe[f"ema_{period}"] = ta.EMA(dataframe, timeperiod=period)
        dataframe["ema_50"] = ta.EMA(dataframe, timeperiod=50)
        dataframe["ema_200"] = ta.EMA(dataframe, timeperiod=200)

        # RSI (all periods for hyperopt range 10-20)
        for period in range(10, 21):
            dataframe[f"rsi_{period}"] = ta.RSI(dataframe, timeperiod=period)

        # ADX
        dataframe["adx"] = ta.ADX(dataframe, timeperiod=14)
        dataframe["plus_di"] = ta.PLUS_DI(dataframe, timeperiod=14)
        dataframe["minus_di"] = ta.MINUS_DI(dataframe, timeperiod=14)

        # MACD
        macd = ta.MACD(dataframe, fastperiod=12, slowperiod=26, signalperiod=9)
        dataframe["macd"] = macd["macd"]
        dataframe["macdsignal"] = macd["macdsignal"]
        dataframe["macdhist"] = macd["macdhist"]
        dataframe["macdhist_prev"] = macd["macdhist"].shift(1)

        # Bollinger Bands
        bb = ta.BBANDS(dataframe, timeperiod=20, nbdevup=2.0, nbdevdn=2.0)
        dataframe["bb_upper"] = bb["upperband"]
        dataframe["bb_middle"] = bb["middleband"]
        dataframe["bb_lower"] = bb["lowerband"]
        # BB width for volatility regime
        dataframe["bb_width"] = (dataframe["bb_upper"] - dataframe["bb_lower"]) / (dataframe["bb_middle"] + 1e-10)
        dataframe["bb_width_sma"] = ta.SMA(dataframe["bb_width"], timeperiod=50)

        # Volume (fix #4: epsilon guard against division by zero)
        dataframe["volume_ema"] = ta.EMA(dataframe["volume"], timeperiod=20)
        dataframe["volume_ratio"] = dataframe["volume"] / (dataframe["volume_ema"] + 1e-10)

        # OBV
        dataframe["obv"] = ta.OBV(dataframe)
        dataframe["obv_ema"] = ta.EMA(dataframe["obv"], timeperiod=20)

        # ATR for dynamic stoploss
        dataframe["atr"] = ta.ATR(dataframe, timeperiod=14)

        # Regime
        dataframe["is_bull"] = (
            (dataframe["close"] > dataframe["ema_200"]) &
            (dataframe["ema_50"] > dataframe["ema_200"])
        ).astype(int)

        dataframe["is_bear"] = (
            (dataframe["close"] < dataframe["ema_200"]) &
            (dataframe["ema_50"] < dataframe["ema_200"])
        ).astype(int)

        # --- LONG pullback detection ---
        ema_slow_key = f"ema_{self.ema_slow.value}"
        if ema_slow_key in dataframe.columns:
            dataframe["pullback_to_ema"] = (
                (dataframe["low"] <= dataframe[ema_slow_key] * 1.02) &
                (dataframe["close"] > dataframe[ema_slow_key]) &
                (dataframe["close"] > dataframe["open"])  # Bullish candle
            ).astype(int)
        else:
            dataframe["pullback_to_ema"] = 0

        # EMA50 support bounce (LONG)
        dataframe["ema50_bounce"] = (
            (dataframe["low"] <= dataframe["ema_50"] * 1.01) &
            (dataframe["close"] > dataframe["ema_50"]) &
            (dataframe["close"] > dataframe["open"])
        ).astype(int)

        # --- Multi-Timeframe data ---
        if self.dp:
            # 4h data for current pair
            df_4h = self.dp.get_pair_dataframe(pair=metadata['pair'], timeframe='4h')
            if len(df_4h) > 0:
                df_4h['ema_50'] = ta.EMA(df_4h, timeperiod=50)
                df_4h['ema_200'] = ta.EMA(df_4h, timeperiod=200)
                df_4h['rsi_14'] = ta.RSI(df_4h, timeperiod=14)
                df_4h['adx'] = ta.ADX(df_4h, timeperiod=14)
                df_4h['is_bull'] = (
                    (df_4h['close'] > df_4h['ema_200']) &
                    (df_4h['ema_50'] > df_4h['ema_200'])
                ).astype(int)
                dataframe = merge_informative_pair(
                    dataframe,
                    df_4h[['date', 'ema_50', 'ema_200', 'rsi_14', 'adx', 'is_bull']],
                    self.timeframe, '4h', ffill=True
                )
            else:
                dataframe['ema_50_4h'] = 0
                dataframe['ema_200_4h'] = 0
                dataframe['rsi_14_4h'] = 50
                dataframe['adx_4h'] = 0
                dataframe['is_bull_4h'] = 0

            # Daily data for macro trend
            df_1d = self.dp.get_pair_dataframe(pair=metadata['pair'], timeframe='1d')
            if len(df_1d) > 0:
                df_1d['ema_200'] = ta.EMA(df_1d, timeperiod=200)
                dataframe = merge_informative_pair(
                    dataframe,
                    df_1d[['date', 'ema_200']],
                    self.timeframe, '1d', ffill=True
                )
            else:
                dataframe['ema_200_1d'] = 0

            # BTC market sentiment
            df_btc = self.dp.get_pair_dataframe(pair='BTC/USDT:USDT', timeframe='1h')
            if len(df_btc) > 0:
                df_btc['btc_ema_200'] = ta.EMA(df_btc, timeperiod=200)
                df_btc['btc_ema_50'] = ta.EMA(df_btc, timeperiod=50)
                df_btc['btc_rsi'] = ta.RSI(df_btc, timeperiod=14)
                df_btc['btc_is_bull'] = (
                    (df_btc['close'] > df_btc['btc_ema_200']) &
                    (df_btc['btc_ema_50'] > df_btc['btc_ema_200'])
                ).astype(int)
                dataframe = merge_informative_pair(
                    dataframe,
                    df_btc[['date', 'btc_ema_200', 'btc_ema_50', 'btc_rsi', 'btc_is_bull']],
                    self.timeframe, '1h', ffill=True
                )
            else:
                dataframe['btc_is_bull_1h'] = 1
                dataframe['btc_rsi_1h'] = 50

            # --- BTC daily for bear-avoidance regime gate ---
            df_btc_1d = self.dp.get_pair_dataframe(pair='BTC/USDT:USDT', timeframe='1d')
            if len(df_btc_1d) > 0:
                df_btc_1d['btc_sma200'] = ta.SMA(df_btc_1d, timeperiod=200)
                df_btc_1d['btc_adx'] = ta.ADX(df_btc_1d, timeperiod=14)
                df_btc_1d['btc_is_bear'] = (
                    (df_btc_1d['close'] < df_btc_1d['btc_sma200'])
                    & (df_btc_1d['btc_adx'] > 18)
                ).astype(int)
                dataframe = merge_informative_pair(
                    dataframe,
                    df_btc_1d[['date', 'btc_sma200', 'btc_adx', 'btc_is_bear']],
                    self.timeframe, '1d', ffill=True
                )
            else:
                logger.warning('Bear gate: BTC/USDT:USDT 1d dataframe empty — gate inactive')
                dataframe['btc_is_bear_1d'] = 0
                dataframe['btc_adx_1d'] = 20
                dataframe['btc_sma200_1d'] = 0
        else:
            # Safety fallback when dp is not available
            dataframe['is_bull_4h'] = dataframe['is_bull']
            dataframe['rsi_14_4h'] = dataframe['rsi_14'] if 'rsi_14' in dataframe.columns else 50
            dataframe['adx_4h'] = dataframe['adx']
            dataframe['btc_is_bull_1h'] = 1
            dataframe['btc_rsi_1h'] = 50
            dataframe['ema_200_1d'] = 0
            dataframe['btc_is_bear_1d'] = 0
            dataframe['btc_adx_1d'] = 20
            dataframe['btc_sma200_1d'] = 0

        # Ensure columns exist (safety for backtesting edge cases)
        for col, default in [
            ('is_bull_4h', 1), ('rsi_14_4h', 50), ('adx_4h', 20),
            ('btc_is_bull_1h', 1), ('btc_rsi_1h', 50),
            ('btc_is_bear_1d', 0), ('btc_adx_1d', 20), ('btc_sma200_1d', 0),
            ('ema_200_1d', 0),
        ]:
            if col not in dataframe.columns:
                dataframe[col] = default

        # --- Fear & Greed Index: static neutral (no API) ---
        dataframe['fng_value'] = 50

        # --- On-chain: static defaults (no API) ---
        dataframe['funding_rate'] = 0.0
        dataframe['funding_extreme'] = 0
        dataframe['oi_change'] = 0.0

        return dataframe

    def populate_entry_trend(self, dataframe: DataFrame, metadata: dict) -> DataFrame:
        rsi = f"rsi_{self.rsi_period.value}"

        # ========== LONG ENTRIES ==========

        # === LONG 1: Trend Pullback to EMA ===
        conditions_pullback = [
            dataframe["is_bull"] == 1,
            dataframe["pullback_to_ema"] == 1,
            dataframe[rsi] > self.rsi_pullback_low.value,
            dataframe[rsi] < self.rsi_pullback_high.value,
            dataframe["adx"] > self.adx_threshold.value,
            dataframe["volume_ratio"] > self.volume_factor.value,
            dataframe["plus_di"] > dataframe["minus_di"],
            dataframe["obv"] > dataframe["obv_ema"],
            dataframe["volume"] > 0,
            dataframe["btc_rsi_1h"] > 35,
            dataframe["fng_value"] >= 25,      # Not extreme fear
            dataframe["fng_value"] <= 85,      # Not extreme greed
            dataframe[rsi] < 70,               # Not overbought
        ]
        # Daily EMA200 filter — helps filter bad entries
        if 'ema_200_1d' in dataframe.columns:
            conditions_pullback.append(dataframe["close"] > dataframe["ema_200_1d"])

        dataframe.loc[
            reduce(lambda x, y: x & y, conditions_pullback),
            ["enter_long", "enter_tag"]
        ] = (1, "trend_pullback")

        # === LONG 2: EMA50 Support Bounce ===
        conditions_ema50 = [
            dataframe["is_bull"] == 1,
            dataframe["ema50_bounce"] == 1,
            dataframe[rsi] > 30,
            dataframe[rsi] < 50,
            dataframe["adx"] > 20,
            dataframe["volume_ratio"] > 1.0,
            dataframe["macdhist"] > dataframe["macdhist"].shift(1),
            dataframe["volume"] > 0,
            dataframe["btc_rsi_1h"] > 35,
            dataframe["fng_value"] >= 25,
            dataframe["fng_value"] <= 85,
            dataframe[rsi] < 70,
        ]
        dataframe.loc[
            reduce(lambda x, y: x & y, conditions_ema50),
            ["enter_long", "enter_tag"]
        ] = (1, "ema50_bounce")

        # === LONG 3: RSI Oversold Bounce (V6: added ADX filter + unified volume threshold) ===
        conditions_rsi = [
            dataframe["close"] > dataframe["ema_200"],
            dataframe[rsi].shift(1) < self.rsi_bounce.value,
            dataframe[rsi] > self.rsi_bounce.value,
            dataframe["close"] > dataframe["bb_lower"],
            dataframe["close"] > dataframe["open"],
            dataframe["adx"] > self.adx_threshold.value,        # V6: trend strength gate
            dataframe["volume_ratio"] > self.volume_factor.value, # V6: was hardcoded 0.8
            dataframe["obv"] > dataframe["obv_ema"],
            dataframe["volume"] > 0,
            dataframe["btc_rsi_1h"] > 35,
            dataframe["fng_value"] >= 25,
            dataframe["fng_value"] <= 85,
        ]
        dataframe.loc[
            reduce(lambda x, y: x & y, conditions_rsi),
            ["enter_long", "enter_tag"]
        ] = (1, "rsi_bounce")

        # === LONG 4: EMA Crossover (golden cross on fast EMAs) ===
        ema_fast_key = f"ema_{self.ema_fast.value}"
        ema_slow_key = f"ema_{self.ema_slow.value}"
        # V6: added ADX filter + unified volume threshold (was hardcoded 0.5 — too loose)
        conditions_ema_cross = [
            (dataframe[ema_fast_key] > dataframe[ema_slow_key]) &
            (dataframe[ema_fast_key].shift(1) <= dataframe[ema_slow_key].shift(1)),  # crossed above
            dataframe[rsi] > 40,
            dataframe[rsi] < 75,
            dataframe["close"] > dataframe["ema_200"],
            dataframe["adx"] > self.adx_threshold.value,         # V6: trend strength gate
            dataframe["volume_ratio"] > self.volume_factor.value, # V6: was hardcoded 0.5
            dataframe["volume"] > 0,
            dataframe["btc_rsi_1h"] > 35,
            dataframe["fng_value"] >= 25,
            dataframe["fng_value"] <= 85,
        ]
        dataframe.loc[
            reduce(lambda x, y: x & y, conditions_ema_cross),
            ["enter_long", "enter_tag"]
        ] = (1, "ema_crossover")

        # === LONG 5: Bollinger Band Bounce (V4: tightened vol 0.3→0.7, added ADX>18) ===
        conditions_bb = [
            dataframe["close"] <= dataframe["bb_lower"] * 1.005,           # close within 0.5% of BB lower
            dataframe["close"] > dataframe["open"],                         # bullish candle (bounce)
            dataframe[rsi] < 45,
            dataframe["volume_ratio"] > 0.7,                                # V4: was 0.3, filter weak bounces
            dataframe["adx"] > 18,                                          # V4: trend strength filter
            dataframe["volume"] > 0,
            dataframe["btc_rsi_1h"] > 35,
            dataframe["fng_value"] >= 25,
            dataframe["fng_value"] <= 85,
        ]
        dataframe.loc[
            reduce(lambda x, y: x & y, conditions_bb),
            ["enter_long", "enter_tag"]
        ] = (1, "bb_bounce")

        # === LONG 6: MACD Histogram Reversal (tightened: RSI 40-60, EMA200 filter, volume 0.8x) ===
        conditions_macd = [
            (dataframe["macdhist"] > 0) &
            (dataframe["macdhist"].shift(1) <= 0),  # histogram crossed above zero
            dataframe["close"] > dataframe["ema_50"],
            dataframe["close"] > dataframe["ema_200"],  # confirm uptrend
            dataframe[rsi] > 40,
            dataframe[rsi] < 60,
            dataframe["adx"] > 15,
            dataframe["volume_ratio"] > 0.8,            # volume confirmation
            dataframe["volume"] > 0,
            dataframe["btc_rsi_1h"] > 35,
            dataframe["fng_value"] >= 25,
            dataframe["fng_value"] <= 85,
        ]
        dataframe.loc[
            reduce(lambda x, y: x & y, conditions_macd),
            ["enter_long", "enter_tag"]
        ] = (1, "macd_reversal")

        # --- Bear-avoidance gate (V6A+gate) ---
        # Veto every long entry when BTC daily is in confirmed bear regime
        # (close<SMA200 & ADX>25). Validated on 480d backtest 2026-05-08:
        # those days carried 122 trades at -0.07% avg, -$4.22 net drag.
        if 'btc_is_bear_1d' in dataframe.columns:
            bear_mask = dataframe['btc_is_bear_1d'] == 1
            if len(dataframe) and bear_mask.iloc[-1] and dataframe['enter_long'].iloc[-1] == 1:
                logger.info(f"Bear gate VETO {metadata['pair']}: BTC<SMA200 & ADX>18 — long entry blocked")
            dataframe.loc[bear_mask, ['enter_long', 'enter_tag']] = (0, '')
        else:
            logger.warning('Bear gate: btc_is_bear_1d column missing — gate INACTIVE this cycle')

        return dataframe

    def populate_exit_trend(self, dataframe: DataFrame, metadata: dict) -> DataFrame:
        rsi = f"rsi_{self.rsi_period.value}"
        ema_fast = f"ema_{self.ema_fast.value}"
        ema_slow = f"ema_{self.ema_slow.value}"

        # ========== LONG EXITS ==========

        # EXIT 1: RSI very overbought
        dataframe.loc[
            (dataframe[rsi] > self.rsi_exit.value) &
            (dataframe["volume"] > 0),
            ["exit_long", "exit_tag"]
        ] = (1, "rsi_overbought")

        # EXIT 2: Bearish EMA cross with MACD confirmation
        dataframe.loc[
            (dataframe[ema_fast] < dataframe[ema_slow]) &
            (dataframe[ema_fast].shift(1) >= dataframe[ema_slow].shift(1)) &
            (dataframe["macdhist"] < 0) &
            (dataframe[rsi] > 50) &
            (dataframe["volume"] > 0),
            ["exit_long", "exit_tag"]
        ] = (1, "ema_bearish_cross")

        # EXIT 3: Price drops below EMA200 by 1%+ (trend broken, softened to avoid premature exits)
        dataframe.loc[
            (dataframe["close"] < dataframe["ema_200"] * 0.99) &
            (dataframe["close"].shift(1) >= dataframe["ema_200"].shift(1)) &
            (dataframe["volume"] > 0),
            ["exit_long", "exit_tag"]
        ] = (1, "trend_broken")

        # EXIT 4 (V4): Trend early warning — RSI overbought reversal near EMA200
        # Catches trend exhaustion before price breaks support, saving avg -3% vs trend_broken
        dataframe.loc[
            (dataframe["close"] < dataframe["ema_200"] * 0.995) &  # within 0.5% of breaking
            (dataframe[rsi] > 72) &                                  # exhausted
            (dataframe["macdhist"] < dataframe["macdhist"].shift(1)) & # momentum dropping
            (dataframe["volume"] > 0),
            ["exit_long", "exit_tag"]
        ] = (1, "trend_early_warning")

        return dataframe


    # --- Improved Confidence Scoring (inline from trendrider_confidence) ---
    def _calc_confidence(self, last: dict) -> tuple:
        """Calculate signal confidence based on weighted indicator alignment.

        Max score ~17.5. Returns (level_str, bar_str, details_list, numeric_level).
        """
        score = 0.0
        details = []
        rsi_key = f"rsi_{self.rsi_period.value}"
        rsi_val = last.get(rsi_key, 50)

        # RSI in healthy zone (not overbought): +1.5
        if 35 < rsi_val < 60:
            score += 1.5
            details.append("RSI healthy")

        # Strong trend (ADX): +2.5 strong, +1.5 moderate
        adx_val = last.get('adx', 0)
        if adx_val > 30:
            score += 2.5
            details.append("Strong trend")
        elif adx_val > self.adx_threshold.value:
            score += 1.5
            details.append("Moderate trend")

        # Volume confirmation: +2.5 high, +1.5 normal
        vol_ratio = last.get('volume_ratio', 0)
        if vol_ratio > 1.5:
            score += 2.5
            details.append("High volume")
        elif vol_ratio > 1.0:
            score += 1.5
            details.append("Normal volume")

        # MACD positive histogram: +1.5, bonus +0.5 if rising
        macd_hist = last.get('macdhist', 0)
        macd_hist_prev = last.get('macdhist_prev', 0)
        if macd_hist > 0:
            score += 1.5
            if macd_hist > macd_hist_prev:
                score += 0.5
                details.append("MACD positive+rising")
            else:
                details.append("MACD positive")

        # OBV rising AND above EMA: +1.5
        if last.get('obv', 0) > last.get('obv_ema', 0):
            score += 1.5
            details.append("OBV rising")

        # BTC healthy (RSI 40-70): +1.5
        btc_rsi = last.get('btc_rsi_1h', 50)
        if 40 < btc_rsi < 70:
            score += 1.5
            details.append("BTC healthy")

        # 4h trend alignment AND ADX_4h > 20: +1.5
        if last.get('is_bull_4h', 0) == 1 and last.get('adx_4h', 0) > 20:
            score += 1.5
            details.append("4H trend aligned")

        # Bollinger Band position (close near lower = good for long): +1
        close = last.get('close', 0)
        bb_lower = last.get('bb_lower', 0)
        bb_upper = last.get('bb_upper', 0)
        bb_range = bb_upper - bb_lower if bb_upper > bb_lower else 1
        if bb_lower > 0 and close > 0:
            bb_position = (close - bb_lower) / bb_range
            if bb_position < 0.35:
                score += 1.0
                details.append("Near BB lower")

        # Plus_DI > Minus_DI spread > 10: +1
        plus_di = last.get('plus_di', 0)
        minus_di = last.get('minus_di', 0)
        if plus_di - minus_di > 10:
            score += 1.0
            details.append("Strong DI spread")

        # FNG bonus: neutral/healthy (40-60): +1
        fng_val = last.get('fng_value', 50)
        if 40 <= fng_val <= 60:
            score += 1.0
            details.append("FNG neutral")

        # On-chain: healthy funding rate: +1
        funding = last.get('funding_rate', 0)
        if abs(funding) < 0.0001:  # Normal funding
            score += 1
            details.append("Healthy funding")

        # Smooth mapping to 1-10 (max score ~17.5)
        numeric = max(1, min(10, round(score * 10 / 17.5)))

        # Level label
        if numeric >= 8:
            level = "STRONG"
        elif numeric >= 6:
            level = "GOOD"
        elif numeric >= 4:
            level = "MEDIUM"
        else:
            level = "WEAK"

        # Dynamic bar
        bar = "|" * numeric + "-" * (10 - numeric) + f" {numeric}/10"

        return level, bar, details, numeric

    def _market_context(self, last: dict) -> str:
        """Generate market context string."""
        btc_rsi = last.get('btc_rsi_1h', 50)
        btc_bull = last.get('btc_is_bull_1h', 0)
        bull_4h = last.get('is_bull_4h', 0)

        if btc_bull and btc_rsi > 55:
            btc_status = "Bullish"
        elif btc_rsi > 40:
            btc_status = "Neutral"
        else:
            btc_status = "Bearish"

        tf_4h = "Uptrend" if bull_4h else "Downtrend"

        parts = [f"BTC: {btc_status} (RSI {btc_rsi:.0f})", f"4H: {tf_4h}"]

        return " | ".join(parts)

    def _get_market_regime(self, last: dict) -> str:
        """Detect market regime from ADX + EMA200 + BB width."""
        adx_val = last.get('adx', 0)
        ema_200 = last.get('ema_200', 0)
        close = last.get('close', 0)
        is_bull = last.get('is_bull', 0)
        bb_width = last.get('bb_width', 0)
        bb_width_sma = last.get('bb_width_sma', 0)

        high_vol = bb_width > bb_width_sma * 1.5 if bb_width_sma > 0 else False

        if adx_val < 20:
            return "Ranging (High Vol)" if high_vol else "Ranging"
        elif is_bull and close > ema_200:
            return "Trending Bull"
        else:
            return "Trending Bear (High Vol)" if high_vol else "Trending Bear"

    def custom_exit(self, pair: str, trade, current_time: datetime,
                    current_rate: float, current_profit: float, **kwargs):
        """V4 cascading early exit — stop bleeding before 24h timeout.

        Real dry-run data (51 trades): time_exit_24h cost -$13.01 across 9 trades,
        avg -2.85% loss after holding full 24h. Cascade catches losers earlier:
        - 2h: cut if -1.5% (already broken thesis)
        - 4h: cut if red (no recovery momentum)
        - 8h: cut if not at +0.5% (dead trade)
        - 16h: cut if not at +1% (final mercy)
        """
        duration_hours = (current_time - trade.open_date_utc).total_seconds() / 3600
        if duration_hours >= 2 and current_profit < -0.015:
            return "early_loss_cut_2h"
        if duration_hours >= 4 and current_profit < 0:
            return "early_loss_cut_4h"
        if duration_hours >= 8 and current_profit < 0.005:
            return "early_loss_cut_8h"
        if duration_hours >= 16 and current_profit < 0.01:
            return "early_loss_cut_16h"
        # FIX: catch trades drifting negative between 16h-24h
        # Data: 9 trades hit time_exit_24h at avg -2.92%, costing $13.01 (=total bot profit).
        if duration_hours >= 19 and current_profit < 0:
            return "early_loss_cut_19h"
        if duration_hours >= 22 and current_profit < -0.01:
            return "early_loss_cut_22h"
        if duration_hours >= 24:
            return "time_exit_24h"
        return None

    def confirm_trade_entry(self, pair: str, order_type: str, amount: float, rate: float,
                           time_in_force: str, current_time: datetime, entry_tag: str | None,
                           side: str, **kwargs) -> bool:
        # Calculate levels (LONG only, can_short = False)
        sl_price = rate * (1 + self.stoploss)
        tp2_price = rate * 1.05   # +5%

        leverage = self.leverage_value
        side_str = "LONG"

        # Risk/reward ratio
        risk = abs(rate - sl_price)
        reward = abs(tp2_price - rate)
        rr_ratio = reward / risk if risk > 0 else 0

        # Entry reason mapping
        reasons = {
            "trend_pullback": "Pullback to EMA in uptrend, bounce with volume confirmation",
            "ema50_bounce": "Deep pullback to EMA50, bounce with rising MACD",
            "rsi_bounce": "RSI oversold, bounce from lower Bollinger in bull market",
            "ema_crossover": "EMA9 crossed above EMA16, golden cross with trend confirmation",
            "bb_bounce": "Price bounced from lower Bollinger Band with oversold RSI",
            "macd_reversal": "MACD histogram turned positive, momentum shift above EMA50",
        }
        reason = reasons.get(entry_tag, entry_tag or "Signal")

        # Get current indicators for context
        dataframe, _ = self.dp.get_analyzed_dataframe(pair, self.timeframe)
        if len(dataframe) > 0:
            last = dataframe.iloc[-1]
            rsi_key = f"rsi_{self.rsi_period.value}"
            rsi_val = last.get(rsi_key, 0)
            adx_val = last.get("adx", 0)
            vol_ratio = last.get("volume_ratio", 0)
            macd_hist = last.get("macdhist", 0)
        else:
            rsi_val = adx_val = vol_ratio = macd_hist = 0
            last = {}

        # Confidence & market context
        conf_level, conf_bar, conf_details, conf_numeric = self._calc_confidence(last)
        market_ctx = self._market_context(last)
        regime = self._get_market_regime(last)

        # --- REJECT WEAK SIGNALS ---
        min_conf = 6 if "Bear" in regime else 5
        if conf_numeric < min_conf:
            logger.info(f"Rejecting signal for {pair}: confidence {conf_numeric}/10 < {min_conf} (regime: {regime})")
            return False

        # --- Main Telegram Signal ---
        msg = (
            f"*TRENDRIDER SIGNAL*\n"
            f"{'='*28}\n"
            f"*{pair}* | *{side_str}* | {leverage}x\n"
            f"{'='*28}\n\n"
            f"*Entry:* `{rate:.2f}` USDT\n"
            f"*Stop Loss:* `{sl_price:.2f}` ({self.stoploss*100:+.1f}%)\n"
            f"  R:R = 1:{rr_ratio:.1f}\n\n"
            f"*Confidence:* {conf_level}\n"
            f"  [{conf_bar}]\n"
            f"  {', '.join(conf_details)}\n\n"
            f"*Regime:* {regime}\n"
            f"*Indicators:*\n"
            f"  RSI: {rsi_val:.1f} | ADX: {adx_val:.1f}\n"
            f"  Volume: {vol_ratio:.2f}x | MACD: {'+'  if macd_hist > 0 else '-'}\n\n"
            f"*Market:* {market_ctx}\n\n"
            f"*Why:* {reason}\n"
            f"{'='*28}\n"
            f"_TrendRider AI_"
        )
        self.dp.send_msg(msg, always_send=True)

        return True

    def confirm_trade_exit(self, pair: str, trade, order_type: str, amount: float,
                          rate: float, time_in_force: str, exit_reason: str,
                          current_time: datetime, **kwargs) -> bool:
        # Calculate results (LONG only)
        profit_pct = ((rate - trade.open_rate) / trade.open_rate) * 100 * trade.leverage
        duration_hours = (current_time - trade.open_date_utc).total_seconds() / 3600

        # Exit reason mapping
        exit_reasons = {
            "roi": "ROI target reached",
            "stop_loss": "Stop Loss hit",
            "trailing_stop_loss": "Trailing Stop",
            "exit_signal": "Exit signal",
            "rsi_overbought": "RSI overbought (>81)",
            "ema_bearish_cross": "EMA bearish crossover",
            "trend_broken": "Trend broken (below EMA200)",
            "force_exit": "Force exit",
            "time_exit_24h": "Time exit (24h, low profit)",
        }
        reason_text = exit_reasons.get(exit_reason, exit_reason)

        # Result line
        if profit_pct > 0:
            result_line = f"+{profit_pct:.2f}%"
        else:
            result_line = f"{profit_pct:.2f}%"

        # Duration formatting
        if duration_hours < 1:
            dur_str = f"{int(duration_hours * 60)}m"
        elif duration_hours < 24:
            dur_str = f"{duration_hours:.1f}h"
        else:
            dur_str = f"{duration_hours/24:.1f}d"

        msg = (
            f"*TRADE CLOSED* {'WIN' if profit_pct > 0 else 'LOSS'}\n"
            f"{'='*25}\n"
            f"*{pair}* | LONG | {trade.leverage}x\n"
            f"{'='*25}\n\n"
            f"*Entry:* `{trade.open_rate:.2f}`\n"
            f"*Exit:* `{rate:.2f}`\n"
            f"*Result:* *{result_line}*\n"
            f"*Duration:* {dur_str}\n"
            f"*Reason:* {reason_text}\n"
            f"*Max price:* `{trade.max_rate:.2f}`\n"
            f"{'='*25}\n"
            f"_TrendRider AI_"
        )

        self.dp.send_msg(msg, always_send=True)
        return True