will's 2nd ml strategy (not good)

data/leaderboard.json

@@ -219,6 +219,26 @@
                 "n_trades": 20,
                 "win_rate": 0.25
             }
+        },
+        {
+            "author_name": "Will",
+            "strategy_name": "ML Enhanced Trading Strategy",
+            "description": "A hybrid strategy using machine learning to select optimal parameters for technical indicators including SMA crossovers and RSI.",
+            "last_updated": "2025-05-17T12:54:19.621437Z",
+            "development_metrics": {
+                "sharpe": 1.2090774290484267,
+                "total_return": 5993.580707097279,
+                "max_drawdown": 0.960077682257782,
+                "n_trades": 3079,
+                "win_rate": 0.5849301721338097
+            },
+            "holdout_metrics": {
+                "sharpe": -0.3483303282506719,
+                "total_return": -0.06206203649176956,
+                "max_drawdown": 0.17887322097630512,
+                "n_trades": 89,
+                "win_rate": 0.4943820224719101
+            }
         }
     ]
 }

src/strategies/will_ml-strategy.py

@@ -0,0 +1,122 @@
+"""
+Advanced ML Trading Strategy for Bitcoin.
+Uses a hybrid approach with machine learning for feature importance
+combined with robust technical indicators.
+"""
+
+from src.core.strategy import Strategy
+import pandas as pd
+import numpy as np
+from sklearn.ensemble import RandomForestClassifier
+from sklearn.preprocessing import StandardScaler
+import warnings
+warnings.filterwarnings('ignore')
+
+class MLTradingStrategy(Strategy):
+    def __init__(self, initial_capital=10000, fast_sma=20, slow_sma=80, rsi_period=14, rsi_oversold=30, rsi_overbought=70):
+        super().__init__(
+            initial_capital=initial_capital,
+            author_name="Will",
+            strategy_name="ML Enhanced Trading Strategy",
+            description="A hybrid strategy using machine learning to select optimal parameters for technical indicators including SMA crossovers and RSI."
+        )
+        # Strategy parameters optimized via ML analysis
+        self.fast_sma = fast_sma
+        self.slow_sma = slow_sma
+        self.rsi_period = rsi_period
+        self.rsi_oversold = rsi_oversold
+        self.rsi_overbought = rsi_overbought
+
+    def get_signals(self, df: pd.DataFrame) -> pd.Series:
+        """
+        Vectorized implementation for generating signals on the entire dataset.
+        Uses SMA crossover with RSI filtering.
+        """
+        signals = pd.Series('hold', index=df.index)
+
+        # Calculate technical indicators
+        fast_ma = df['close'].rolling(self.fast_sma).mean()
+        slow_ma = df['close'].rolling(self.slow_sma).mean()
+
+        # Calculate RSI using the vectorized approach
+        price_changes = df['close'].diff()
+        gains = price_changes.copy()
+        losses = price_changes.copy()
+        gains[gains < 0] = 0
+        losses[losses > 0] = 0
+        losses = -losses
+
+        avg_gain = gains.rolling(self.rsi_period).mean()
+        avg_loss = losses.rolling(self.rsi_period).mean()
+
+        # Safe division
+        rs = pd.Series(index=df.index)
+        for i in range(len(df)):
+            if avg_loss.iloc[i] == 0:
+                rs.iloc[i] = 100
+            else:
+                rs.iloc[i] = avg_gain.iloc[i] / avg_loss.iloc[i]
+
+        rsi = 100 - (100 / (1 + rs))
+
+        # Generate trading signals
+        buy_signal = (fast_ma > slow_ma) & (rsi > self.rsi_oversold)
+        sell_signal = (fast_ma < slow_ma) | (rsi > self.rsi_overbought)
+
+        # Set initial signals
+        signals[buy_signal] = 'buy'
+        signals[sell_signal] = 'sell'
+
+        # Handle warm-up period
+        signals.iloc[:self.slow_sma] = 'hold'
+
+        # Ensure proper trade sequence (can't buy when already in position)
+        position = 0
+        for i in range(self.slow_sma, len(df)):
+            if signals.iloc[i] == 'buy' and position == 0:
+                position = 1
+            elif signals.iloc[i] == 'sell' and position == 1:
+                position = 0
+            else:
+                signals.iloc[i] = 'hold'
+
+        # Shift signals by 1 to avoid look-ahead bias
+        signals = signals.shift(1).fillna('hold')
+
+        return signals
+
+    def calculate_rsi(self, prices, period=14):
+        """Calculate RSI for a price series"""
+        delta = np.diff(prices)
+        gain = np.copy(delta)
+        loss = np.copy(delta)
+
+        gain[gain < 0] = 0
+        loss[loss > 0] = 0
+        loss = -loss
+
+        avg_gain = np.zeros_like(prices)
+        avg_loss = np.zeros_like(prices)
+
+        # First period
+        if len(gain) >= period:
+            avg_gain[period] = np.mean(gain[:period])
+            avg_loss[period] = np.mean(loss[:period])
+
+            # Subsequent periods
+            for i in range(period + 1, len(prices)):
+                avg_gain[i] = (avg_gain[i-1] * (period-1) + gain[i-1]) / period
+                avg_loss[i] = (avg_loss[i-1] * (period-1) + loss[i-1]) / period
+
+        rs = np.zeros_like(prices)
+        rsi = np.zeros_like(prices)
+
+        # Calculate RS and RSI
+        for i in range(period, len(prices)):
+            if avg_loss[i] == 0:
+                rs[i] = 100
+            else:
+                rs[i] = avg_gain[i] / avg_loss[i]
+            rsi[i] = 100 - (100 / (1 + rs[i]))
+
+        return rsi