Estratégia AlphaTrader Long Biased 2

alpha_trader/notebooks/explore_4.ipynb

@@ -9,7 +9,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 298,
+   "execution_count": 4,
    "metadata": {},
    "outputs": [
     {
@@ -18,7 +18,7 @@
        "True"
       ]
      },
-     "execution_count": 298,
+     "execution_count": 4,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -69,7 +69,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 304,
+   "execution_count": 5,
    "metadata": {},
    "outputs": [],
    "source": [

alpha_trader/src/dataManager.py

@@ -295,6 +295,48 @@ def determine_context(row):
             return np.nan
 
     def compute_context(self, start, end):
+        start = datetime.strptime(start, '%Y-%m-%d')
+        end = datetime.strptime(end, '%Y-%m-%d')
+
+        # Retrieve context data
+        context_data = self.get_context_data(pd.to_datetime('2011-08-1'), end)
+
+        # Market condition calculations
+        gradient_numerator = (context_data['price_usd_close'].diff(28) - context_data['price_realized_usd'].diff(28))
+        gradient_mean = gradient_numerator.expanding().mean()
+        gradient_std = gradient_numerator.expanding().std()
+        context_data['28d_mkt_gradient'] = (gradient_numerator - gradient_mean) / gradient_std
+
+        context_data['mayer_multiple'] = context_data['price_usd_close'] / context_data['price_usd_close'].rolling(200).mean()
+        context_data['price_profit_corr'] = context_data['price_usd_close'].rolling(7).corr(context_data['profit_relative'])
+
+        # Simple linear scaling function
+        def linear_scale(value, low, high):
+            # Values below low map to 0, above high map to 1, in between scale linearly
+            return np.clip((value - low) / (high - low), 0, 1)
+
+        # Normalize metrics to 0-1 based on chosen thresholds:
+        # Thresholds chosen from top detection conditions and reasoning about extremes.
+        # Adjust if necessary after backtesting.
+        mvrv_norm = linear_scale(context_data['mvrv_z_score'], 0.0, 3.8)  # 0 to 3.8 range
+        mayer_norm = linear_scale(context_data['mayer_multiple'], 1.0, 1.3)  # 1.0 to 1.3 range
+        nupl_norm = linear_scale(context_data['net_unrealized_profit_loss_account_based'], 0.0, 0.6)  # 0.0 to 0.6 range
+        gradient_norm = linear_scale(context_data['28d_mkt_gradient'], 0.0, 7.0)  # 0.0 to 7.0 range
+
+        # Combine into one continuous context measure:
+        # Average all four normalized values
+        context_data['context'] = (mvrv_norm + mayer_norm + nupl_norm + gradient_norm) / 4.0
+
+        # Drop unnecessary columns
+        context_data.drop(columns=['price_usd_close'], inplace=True)
+
+        # Filter the data to the requested start date
+        context_data = context_data[context_data.index >= start]
+        context_data = context_data[context_data.index <= end]
+
+        return context_data
+
+    def compute_context_boolean(self, start, end):
 
         start = datetime.strptime(start, '%Y-%m-%d')
         end = datetime.strptime(end, '%Y-%m-%d')
@@ -331,11 +373,15 @@ def compute_context(self, start, end):
 
         return context_data
 
-    def get_data(self, start, end, contextualize=True):
+    def get_data(self, start, end, contextualize=True, boolean=False):
 
         trigger_data = self.compute_triggers(start=start, end=end)
-        context_data = self.compute_context(start=start,end=end)
-
+
+        if boolean:
+            context_data = self.compute_context_boolean(start=start,end=end)
+        else:
+            context_data = self.compute_context(start=start,end=end)
+
         trigger_data.reset_index(inplace=True)
         context_data.reset_index(inplace=True)
         context_data['t'] = context_data['t'].dt.tz_localize(None)  # Ensure no timezone differences

alpha_trader/src/tests.ipynb

@@ -0,0 +1,207 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from tradingStrategies import BuyNHold, AlphaTraderLongBiased, AlphaTraderLongBiased2, AFT01   "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "bnh = BuyNHold(1000, '2020-01-01', '2024-06-1')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "test = AlphaTraderLongBiased2(1000, '2020-01-01', '2024-12-17')\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "test.data"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import pandas as pd\n",
+    "import matplotlib.pyplot as plt\n",
+    "\n",
+    "# Assume self.data with a DatetimeIndex and 'position' column exists\n",
+    "data = test.data.copy()\n",
+    "\n",
+    "# Define masks for each type of position\n",
+    "is_cash = (data['position'] == 0)\n",
+    "is_long = (data['position'] > 0)\n",
+    "is_short = (data['position'] < 0)\n",
+    "\n",
+    "def get_durations(mask, index):\n",
+    "    \"\"\"\n",
+    "    Given a boolean mask (e.g., is_cash) and a DatetimeIndex,\n",
+    "    find consecutive stretches of True and return their durations in days.\n",
+    "    \"\"\"\n",
+    "    int_mask = mask.astype(int)\n",
+    "    # Start points where we go from False to True\n",
+    "    starts = (int_mask.diff() == 1)\n",
+    "    # End points where we go from True to False\n",
+    "    ends = (int_mask.diff() == -1)\n",
+    "    \n",
+    "    starts_idx = index[starts]\n",
+    "    ends_idx = index[ends]\n",
+    "    \n",
+    "    # Handle edge cases:\n",
+    "    # If we start already in True condition at the beginning\n",
+    "    if mask.iloc[0]:\n",
+    "        starts_idx = starts_idx.insert(0, index[0])\n",
+    "    \n",
+    "    # If we end still in True condition at the end\n",
+    "    if mask.iloc[-1]:\n",
+    "        ends_idx = ends_idx.append(pd.Index([index[-1]]))\n",
+    "        \n",
+    "    min_len = min(len(starts_idx), len(ends_idx))\n",
+    "    starts_idx = starts_idx[:min_len]\n",
+    "    ends_idx = ends_idx[:min_len]\n",
+    "    \n",
+    "    durations = []\n",
+    "    for s, e in zip(starts_idx, ends_idx):\n",
+    "        if e > s:\n",
+    "            duration_days = (e - s).total_seconds() / (3600 * 24)\n",
+    "            durations.append(duration_days)\n",
+    "    return durations\n",
+    "\n",
+    "# Get durations for each position type\n",
+    "cash_durations = get_durations(is_cash, data.index)\n",
+    "long_durations = get_durations(is_long, data.index)\n",
+    "short_durations = get_durations(is_short, data.index)\n",
+    "\n",
+    "# Plot histograms on the same figure with different colors\n",
+    "plt.figure(figsize=(10, 6))\n",
+    "plt.hist(cash_durations, bins=30, alpha=0.5, label='Cash Periods', color='blue', edgecolor='black')\n",
+    "#plt.hist(long_durations, bins=30, alpha=0.5, label='Long Periods', color='green', edgecolor='black')\n",
+    "#plt.hist(short_durations, bins=30, alpha=0.5, label='Short Periods', color='red', edgecolor='black')\n",
+    "\n",
+    "plt.title('Histogram of Durations by Position Type')\n",
+    "plt.xlabel('Duration (Days)')\n",
+    "plt.ylabel('Frequency')\n",
+    "plt.legend()\n",
+    "plt.grid(True)\n",
+    "plt.show()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import matplotlib.pyplot as plt\n",
+    "\n",
+    "fig, ax1 = plt.subplots(figsize=(10, 6))\n",
+    "\n",
+    "# Plot the first dataset on the primary y-axis\n",
+    "ax1.plot(test.data['net_worth'], color='blue', label='Context')\n",
+    "ax1.set_ylabel('Delta', color='blue')\n",
+    "ax1.tick_params(axis='y', labelcolor='blue')\n",
+    "\n",
+    "# Create a second y-axis\n",
+    "ax2 = ax1.twinx()  \n",
+    "# Plot the second dataset on the secondary y-axis\n",
+    "ax2.plot(test.data['context'], color='red', label='Close')\n",
+    "ax2.set_ylabel('Close', color='red')\n",
+    "ax2.tick_params(axis='y', labelcolor='red')\n",
+    "\n",
+    "# Optionally, set titles and grid\n",
+    "plt.title('Dual Axis Plot')\n",
+    "ax1.grid()\n",
+    "\n",
+    "# Show the plot\n",
+    "plt.show()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import matplotlib.pyplot as plt\n",
+    "\n",
+    "fig, ax1 = plt.subplots(figsize=(10, 6))\n",
+    "\n",
+    "# Plot the first dataset on the primary y-axis\n",
+    "ax1.plot(test.data['diff'], color='blue', label='Context')\n",
+    "ax1.set_ylabel('Delta', color='blue')\n",
+    "ax1.tick_params(axis='y', labelcolor='blue')\n",
+    "\n",
+    "# Create a second y-axis\n",
+    "ax2 = ax1.twinx()  \n",
+    "# Plot the second  on the secondary y-axis\n",
+    "ax2.plot(test.data['close'], color='red', label='Close')\n",
+    "ax2.set_ylabel('Close', color='red')\n",
+    "ax2.tick_params(axis='y', labelcolor='red')\n",
+    "\n",
+    "# Optionally, set titles and grid\n",
+    "plt.title('Dual Axis Plot')\n",
+    "ax1.grid()\n",
+    "\n",
+    "# Show the plot\n",
+    "plt.show()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "venv",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.11.5"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}