Update src/streamlit_app.py

src/streamlit_app.py

@@ -1,40 +1,246 @@
-import altair as alt
-import numpy as np
-import pandas as pd
 import streamlit as st
+import yfinance as yf
+import pandas as pd
+import numpy as np
+from hmmlearn.hmm import GaussianHMM
+from sklearn.svm import SVR
+from sklearn.preprocessing import StandardScaler
+import plotly.graph_objects as go
+import plotly.express as px
+from datetime import datetime, timedelta
+
+# --- Config ---
+st.set_page_config(page_title="Hybrid HMM-SVR Strategy Backtester", layout="wide")
+
+# --- Helper Functions ---
+
+@st.cache_data
+def fetch_data(ticker, start_date, end_date):
+    df = yf.download(ticker, start=start_date, end=end_date)
+    if isinstance(df.columns, pd.MultiIndex):
+        df.columns = df.columns.get_level_values(0)
+    return df
+
+def calculate_metrics(df, strategy_col='Strategy_Value', benchmark_col='Buy_Hold_Value'):
+    """Calculates CAG, Sharpe, Drawdown, etc."""
+    stats = {}
+    
+    for col, name in [(strategy_col, 'Hybrid Strategy'), (benchmark_col, 'Buy & Hold')]:
+        # Returns
+        initial = df[col].iloc[0]
+        final = df[col].iloc[-1]
+        total_return = (final - initial) / initial
+        
+        # Daily Returns
+        daily_ret = df[col].pct_change().dropna()
+        
+        # Sharpe (Annualized, assuming 365 trading days for crypto)
+        sharpe = (daily_ret.mean() / daily_ret.std()) * np.sqrt(365) if daily_ret.std() != 0 else 0
+        
+        # Max Drawdown
+        rolling_max = df[col].cummax()
+        drawdown = (df[col] - rolling_max) / rolling_max
+        max_drawdown = drawdown.min()
+        
+        stats[name] = {
+            "Total Return": f"{total_return:.2%}",
+            "Sharpe Ratio": f"{sharpe:.2f}",
+            "Max Drawdown": f"{max_drawdown:.2%}"
+        }
+    
+    return pd.DataFrame(stats)
+
+def train_hmm_model(train_df, n_states):
+    """Trains HMM on historical data (In-Sample)."""
+    # Features: Log Returns and Volatility
+    X_train = train_df[['Log_Returns', 'Volatility']].values * 100
+    
+    model = GaussianHMM(n_components=n_states, covariance_type="full", n_iter=100, random_state=42)
+    model.fit(X_train)
+    
+    # Sort states by Volatility (State 0 = Lowest Risk)
+    hidden_states = model.predict(X_train)
+    state_vol = []
+    for i in range(n_states):
+        avg_vol = X_train[hidden_states == i, 1].mean()
+        state_vol.append((i, avg_vol))
+    state_vol.sort(key=lambda x: x[1])
+    
+    # Create mapping: {Random_ID: Sorted_ID}
+    mapping = {old: new for new, (old, _) in enumerate(state_vol)}
+    
+    return model, mapping
+
+def train_svr_model(train_df):
+    """Trains SVR to predict next day's volatility."""
+    # Features for SVR: Returns, Current Vol, Downside Vol, Regime
+    feature_cols = ['Log_Returns', 'Volatility', 'Downside_Vol', 'Regime']
+    target_col = 'Target_Next_Vol'
+    
+    X = train_df[feature_cols].values
+    y = train_df[target_col].values
+    
+    # Scale features
+    scaler = StandardScaler()
+    X_scaled = scaler.fit_transform(X)
+    
+    # SVR with RBF kernel
+    model = SVR(kernel='rbf', C=100, gamma=0.1, epsilon=0.01)
+    model.fit(X_scaled, y)
+    
+    return model, scaler
+
+# --- Main Logic ---
+
+st.title("🧠 Saad Rizvi Gand phad strategy")
+st.markdown("""
+**The Hybrid Strategy:**
+1.  **Driver:** EMA Crossover (Fast > Slow = Bullish).
+2.  **Filter (HMM):** If Regime is "High Vol/Crash", **Block Trade** (Size = 0).
+3.  **Sizing (SVR):** If Regime is Safe, adjust size based on predicted risk. 
+""")
+
+# Sidebar Inputs
+with st.sidebar:
+    st.header("Settings")
+    ticker = st.text_input("Ticker", "BTC-USD")
+    
+    # Modified Date Logic: User selects Trading Period
+    backtest_start = st.date_input("Backtest Start Date", datetime.now() - timedelta(days=365))
+    backtest_end = st.date_input("Backtest End Date", datetime.now())
+    
+    st.caption("Note: Models will automatically train on the **4 years** of data prior to your selected Start Date.")
+    
+    st.divider()
+    short_window = st.number_input("Fast EMA", 12)
+    long_window = st.number_input("Slow EMA", 26)
+    n_states = st.slider("HMM States", 2, 4, 3)
 
-"""
-# Welcome to Streamlit!
-
-Edit `/streamlit_app.py` to customize this app to your heart's desire :heart:.
-If you have any questions, checkout our [documentation](https://docs.streamlit.io) and [community
-forums](https://discuss.streamlit.io).
-
-In the meantime, below is an example of what you can do with just a few lines of code:
-"""
-
-num_points = st.slider("Number of points in spiral", 1, 10000, 1100)
-num_turns = st.slider("Number of turns in spiral", 1, 300, 31)
-
-indices = np.linspace(0, 1, num_points)
-theta = 2 * np.pi * num_turns * indices
-radius = indices
-
-x = radius * np.cos(theta)
-y = radius * np.sin(theta)
-
-df = pd.DataFrame({
-    "x": x,
-    "y": y,
-    "idx": indices,
-    "rand": np.random.randn(num_points),
-})
-
-st.altair_chart(alt.Chart(df, height=700, width=700)
-    .mark_point(filled=True)
-    .encode(
-        x=alt.X("x", axis=None),
-        y=alt.Y("y", axis=None),
-        color=alt.Color("idx", legend=None, scale=alt.Scale()),
-        size=alt.Size("rand", legend=None, scale=alt.Scale(range=[1, 150])),
-    ))
+if st.button("Run Hybrid Backtest"):
+    # Calculate the Training Start Date (4 Years before Backtest Start)
+    train_start_date = pd.Timestamp(backtest_start) - pd.DateOffset(years=4)
+    
+    # Fetch ALL data (Training Period + Backtest Period)
+    df = fetch_data(ticker, train_start_date, backtest_end)
+    
+    if df is None or len(df) < 200:
+        st.error("Not enough data to backtest. Ensure the ticker existed 4 years prior to your start date.")
+    else:
+        # 1. Feature Engineering
+        df['Log_Returns'] = np.log(df['Close'] / df['Close'].shift(1))
+        df['Volatility'] = df['Log_Returns'].rolling(window=10).std()
+        
+        # Downside Volatility (Leverage Effect Feature)
+        df['Downside_Returns'] = df['Log_Returns'].apply(lambda x: x if x < 0 else 0)
+        df['Downside_Vol'] = df['Downside_Returns'].rolling(window=10).std()
+        
+        # Strategy Indicators
+        df['EMA_Short'] = df['Close'].ewm(span=short_window, adjust=False).mean()
+        df['EMA_Long'] = df['Close'].ewm(span=long_window, adjust=False).mean()
+        
+        # Target for SVR (Next Day Volatility)
+        df['Target_Next_Vol'] = df['Volatility'].shift(-1)
+        
+        df = df.dropna()
+        
+        # 2. Split Data based on Dates
+        train_df = df[df.index < pd.Timestamp(backtest_start)].copy()
+        test_df = df[df.index >= pd.Timestamp(backtest_start)].copy()
+        
+        if len(train_df) < 365:
+            st.warning(f"Warning: Only {len(train_df)} days found for training. HMM performs best with >2 years of data.")
+        
+        if len(test_df) < 10:
+             st.error("Not enough data for backtesting range.")
+        else:
+            st.info(f"Training on {len(train_df)} days ({train_df.index[0].date()} to {train_df.index[-1].date()}). Backtesting on {len(test_df)} days.")
+            
+            with st.spinner("Training HMM (Regime Detection)..."):
+                hmm_model, state_map = train_hmm_model(train_df, n_states)
+                
+                # Predict Train Regimes (Needed for SVR training input)
+                X_train_hmm = train_df[['Log_Returns', 'Volatility']].values * 100
+                train_raw_states = hmm_model.predict(X_train_hmm)
+                train_df['Regime'] = [state_map.get(s, s) for s in train_raw_states]
+                
+            with st.spinner("Training SVR (Volatility Forecasting)..."):
+                svr_model, svr_scaler = train_svr_model(train_df)
+                
+            with st.spinner("Running Backtest Loop..."):
+                # --- OUT OF SAMPLE BACKTEST ---
+                
+                # 1. Predict Regimes for Test Data
+                X_test_hmm = test_df[['Log_Returns', 'Volatility']].values * 100
+                test_raw_states = hmm_model.predict(X_test_hmm)
+                test_df['Regime'] = [state_map.get(s, s) for s in test_raw_states]
+                
+                # 2. Predict Volatility for Test Data (Using SVR)
+                X_test_svr = test_df[['Log_Returns', 'Volatility', 'Downside_Vol', 'Regime']].values
+                X_test_svr_scaled = svr_scaler.transform(X_test_svr)
+                test_df['Predicted_Vol'] = svr_model.predict(X_test_svr_scaled)
+                
+                # 3. Calculate Strategy Logic
+                high_vol_state = n_states - 1
+                
+                # Base Signal (EMA)
+                test_df['Signal'] = np.where(test_df['EMA_Short'] > test_df['EMA_Long'], 1, 0)
+                
+                # Calculate Baseline Risk (Average Volatility seen in Training)
+                avg_train_vol = train_df['Volatility'].mean()
+                
+                # Calculate Position Size (The "Dimmer Switch")
+                # Logic: Size = Average_Vol / Predicted_Vol
+                # If Predicted > Average, Size < 1.0 (Reduce Risk)
+                # If Predicted < Average, Size > 1.0 (Increase Risk) -> Capped at 1.0 for safety
+                test_df['Risk_Ratio'] = test_df['Predicted_Vol'] / avg_train_vol
+                test_df['Position_Size'] = (1.0 / test_df['Risk_Ratio']).clip(upper=1.0, lower=0.0)
+                
+                # Override: If HMM says CRASH, Size = 0
+                test_df['Position_Size'] = np.where(
+                    test_df['Regime'] == high_vol_state, 
+                    0.0, 
+                    test_df['Position_Size']
+                )
+                
+                # Final Position: Signal * Size
+                # We shift(1) because we calculate size today for tomorrow's return
+                test_df['Final_Position'] = (test_df['Signal'] * test_df['Position_Size']).shift(1)
+                
+                # 4. Returns
+                test_df['Strategy_Returns'] = test_df['Final_Position'] * test_df['Log_Returns']
+                test_df['Buy_Hold_Returns'] = test_df['Log_Returns']
+                
+                # Cumulative
+                test_df['Strategy_Value'] = (1 + test_df['Strategy_Returns']).cumprod()
+                test_df['Buy_Hold_Value'] = (1 + test_df['Buy_Hold_Returns']).cumprod()
+                test_df.dropna(inplace=True)
+                
+                # --- RESULTS ---
+                
+                metrics_df = calculate_metrics(test_df)
+                st.subheader("Performance Metrics")
+                st.table(metrics_df)
+                
+                # Charts
+                col1, col2 = st.columns([2, 1])
+                
+                with col1:
+                    st.subheader("Equity Curve")
+                    fig = go.Figure()
+                    fig.add_trace(go.Scatter(x=test_df.index, y=test_df['Buy_Hold_Value'], name='Buy & Hold', line=dict(color='gray', dash='dot')))
+                    fig.add_trace(go.Scatter(x=test_df.index, y=test_df['Strategy_Value'], name='Hybrid Strategy', line=dict(color='#00CC96', width=2)))
+                    st.plotly_chart(fig, use_container_width=True)
+                    
+                with col2:
+                    st.subheader("Position Sizing (SVR Effect)")
+                    st.caption("How SVR adjusted trade size over time (0.0 to 1.0)")
+                    fig_size = px.area(test_df, x=test_df.index, y='Position_Size', title="Dynamic Exposure")
+                    st.plotly_chart(fig_size, use_container_width=True)
+                
+                st.subheader("SVR Prediction Accuracy (Test Set)")
+                fig_svr = go.Figure()
+                # Show a slice to avoid clutter
+                slice_df = test_df.iloc[-100:] 
+                fig_svr.add_trace(go.Scatter(x=slice_df.index, y=slice_df['Target_Next_Vol'], name='Actual Volatility'))
+                fig_svr.add_trace(go.Scatter(x=slice_df.index, y=slice_df['Predicted_Vol'], name='SVR Prediction', line=dict(dash='dot')))
+                st.plotly_chart(fig_svr, use_container_width=True)