LightGBM example.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 348,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import lightgbm as lgb\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn.metrics import accuracy_score\n",
    "from math import sqrt\n",
    "import seaborn as sns\n",
    "from core import *\n",
    "import quandl\n",
    "token='QWe8iSbyAFzRuod2aroM'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [],
   "source": [
    "sp = get_sp()\n",
    "rets=sp.pct_change()\n",
    "df=pd.DataFrame()\n",
    "df['X1']=rets\n",
    "df['X2']=get_bond_time_series()\n",
    "df['Y']=rets.shift(1)\n",
    "df['Ybar']=df.Y/df.Y.abs()\n",
    "df=df.dropna()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [],
   "source": [
    "s=[]\n",
    "for i in df.Y:\n",
    "    s.append(1 if i>0 else 0)\n",
    "df['Y2']=pd.Series(s,index=df.index)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 269,
   "metadata": {},
   "outputs": [],
   "source": [
    "x=df[['X1','X2']].values\n",
    "y=df.Y.values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 281,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train, x_test, y_train, y_test = train_test_split(x, y, test_size = 0.2,random_state=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 278,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_train_flat= flatten(y_train)\n",
    "y_test_flat = flatten(y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 279,
   "metadata": {},
   "outputs": [],
   "source": [
    "sc = StandardScaler()\n",
    "x_train = sc.fit_transform(x_train)\n",
    "x_test = sc.transform(x_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 295,
   "metadata": {},
   "outputs": [],
   "source": [
    "d_train = lgb.Dataset(x_train, label=y_train_flat)\n",
    "params = {}\n",
    "params['learning_rate'] = 0.003\n",
    "params['boosting_type'] = 'gbdt'\n",
    "params['objective'] = 'binary'\n",
    "params['metric'] = 'binary_logloss'\n",
    "params['sub_feature'] = 0.5\n",
    "params['num_leaves'] = 20\n",
    "params['min_data'] = 50\n",
    "params['max_depth'] = 20\n",
    "clf = lgb.train(params, d_train, 100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 296,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_pred=clf.predict(x_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 297,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x125338eb8>"
      ]
     },
     "execution_count": 297,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "pd.Series(y_pred).plot(kind='hist')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 298,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_pred_bar=[]\n",
    "for i in range(0,y_pred.size):\n",
    "    if y_pred[i]>=.6:       # setting threshold to .5\n",
    "       y_pred_bar.append(1)\n",
    "    else:  \n",
    "       y_pred_bar.append(0)\n",
    "y_pred_bar=np.array(y_pred_bar)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 300,
   "metadata": {},
   "outputs": [],
   "source": [
    "accuracy=accuracy_score(y_pred_bar,y_test_flat)\n",
    "cm = confusion_matrix(y_test_flat, y_pred_bar)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 301,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[26, 24],\n",
       "       [30, 59]])"
      ]
     },
     "execution_count": 301,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 302,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.6115107913669064"
      ]
     },
     "execution_count": 302,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 324,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.182620735353893"
      ]
     },
     "execution_count": 324,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "annual_ret/stds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 328,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.605072463768116"
      ]
     },
     "execution_count": 328,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_pred_in=clf.predict(x_train)\n",
    "accuracy_score(flatten(y_pred_in),flatten(y_train))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Experiment comparing in-sample verus out of sample IR\n",
    "First experiment is just to look at different slicing of the data using the same hyper parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 353,
   "metadata": {},
   "outputs": [],
   "source": [
    "res=[]\n",
    "for i in range(0,1000):\n",
    "    x_train, x_test, y_train, y_test = train_test_split(x, y, test_size = 0.25)\n",
    "    y_train_flat= flatten(y_train)\n",
    "    y_test_flat = flatten(y_test)\n",
    "    sc = StandardScaler()\n",
    "    x_train = sc.fit_transform(x_train)\n",
    "    x_test = sc.transform(x_test)\n",
    "    d_train = lgb.Dataset(x_train, label=y_train_flat)\n",
    "    clf = lgb.train(params, d_train, 100)\n",
    "\n",
    "    # Out of sample IR calculation\n",
    "    y_pred=clf.predict(x_test)\n",
    "    y_pred_bar=[]\n",
    "    for i in range(0,y_pred.size):\n",
    "        if y_pred[i]>=.6:       # setting threshold to .5\n",
    "           y_pred_bar.append(1)\n",
    "        else:  \n",
    "           y_pred_bar.append(0)\n",
    "    y_pred_bar=np.array(y_pred_bar)\n",
    "    IR=calc_IR(y_pred_bar,y_test)\n",
    "\n",
    "    # In sample \n",
    "    y_pred_in=clf.predict(x_train)\n",
    "    in_sample=accuracy_score(flatten(y_pred_in),flatten(y_train))\n",
    "    res.append([IR,in_sample])\n",
    "final_res=pd.DataFrame(res,columns=['IR','In Sample'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 378,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x125a55240>"
      ]
     },
     "execution_count": 378,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.scatterplot(y='IR',x='In Sample',data=final_res)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 381,
   "metadata": {},
   "outputs": [],
   "source": [
    "res=[]\n",
    "for i in range(0,2000):\n",
    "    x_train, x_test, y_train, y_test = train_test_split(x, y, test_size = 0.25)\n",
    "    y_train_flat= flatten(y_train)\n",
    "    y_test_flat = flatten(y_test)\n",
    "    sc = StandardScaler()\n",
    "    x_train = sc.fit_transform(x_train)\n",
    "    x_test = sc.transform(x_test)\n",
    "    d_train = lgb.Dataset(x_train, label=y_train_flat)\n",
    "    params = {}\n",
    "    params['learning_rate'] = np.random.random()/7\n",
    "    params['boosting_type'] = 'gbdt'\n",
    "    params['objective'] = 'binary'\n",
    "    params['metric'] = 'binary_logloss'\n",
    "    params['sub_feature'] = 0.5\n",
    "    params['num_leaves'] = np.random.randint(5,30)\n",
    "    params['min_data'] = 50\n",
    "    params['max_depth'] = np.random.randint(5,10)\n",
    "    clf = lgb.train(params, d_train, 100)\n",
    "\n",
    "    # Out of sample IR calculation\n",
    "    y_pred=clf.predict(x_test)\n",
    "    y_pred_bar=[]\n",
    "    for i in range(0,y_pred.size):\n",
    "        if y_pred[i]>=.6:       # setting threshold to .5\n",
    "           y_pred_bar.append(1)\n",
    "        else:  \n",
    "           y_pred_bar.append(0)\n",
    "    y_pred_bar=np.array(y_pred_bar)\n",
    "    IR=calc_IR(y_pred_bar,y_test)\n",
    "\n",
    "    # In sample \n",
    "    y_pred_in=clf.predict(x_train)\n",
    "    in_sample=accuracy_score(flatten(y_pred_in),flatten(y_train))\n",
    "    res.append([IR,in_sample])\n",
    "final_res_hp=pd.DataFrame(res,columns=['IR','In Sample'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 382,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x128c31550>"
      ]
     },
     "execution_count": 382,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.scatterplot(y='IR',x='In Sample',data=final_res_hp)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Additional futures market"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 385,
   "metadata": {},
   "outputs": [],
   "source": [
    "mkts={'SP 500':'CHRIS/CME_SP1',\n",
    "      #'US 10Y':'CHRIS/CME_TY1',\n",
    "      'Crude Oil':'CHRIS/CME_CL1',\n",
    "      'Dollar Index':'CHRIS/ICE_DX1',\n",
    "      #'TED Spread':'FRED/TEDRATE',\n",
    "      'Wheat':'CHRIS/CME_W1',\n",
    "      'Euro':'CHRIS/CME_EC1',\n",
    "      'GBP':'CHRIS/CME_BP1',\n",
    "      'Gold':'CHRIS/CME_EC1'\n",
    "      }\n",
    "\n",
    "data_index=pd.DataFrame()\n",
    "for m in mkts.keys():\n",
    "\ttry:\n",
    "\t\tdata_index[m]=quandl.get(mkts[m],authtoken=token).Last\n",
    "\texcept:\n",
    "\t\ttry:\n",
    "\t\t\tdata_index[m]=quandl.get(mkts[m],authtoken=token).Settle\n",
    "\t\texcept:\n",
    "\t\t\ttry:\n",
    "\t\t\t\tdata_index[m]=quandl.get(mkts[m],authtoken=token).Value\n",
    "\t\t\texcept:\n",
    "\t\t\t\ttry:\n",
    "\t\t\t\t\tdata_index[m]=quandl.get(mkts[m],authtoken=token).value\n",
    "\t\t\t\texcept:\n",
    "\t\t\t\t\ttry:\n",
    "\t\t\t\t\t\tdata_index[m]=quandl.get(mkts[m],authtoken=token).Rate\n",
    "\t\t\t\t\texcept:\n",
    "\t\t\t\t\t\tprint(m)\n",
    "data_pct=data_index.pct_change()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 439,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_pct=data_index[['SP 500', 'Crude Oil', 'Dollar Index',  'Wheat',#'TED Spread',\n",
    "       'Euro', 'GBP', 'Gold']]['2007':].dropna().pct_change().dropna()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 442,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_pct['Y']=data_pct['SP 500'].shift(1)\n",
    "data_pct['Ybar']=data_pct.Y/data_pct.Y.abs()\n",
    "data_pct=data_pct.dropna()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 443,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/mrefermat/anaconda3/lib/python3.7/site-packages/ipykernel_launcher.py:4: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  after removing the cwd from sys.path.\n"
     ]
    }
   ],
   "source": [
    "s=[]\n",
    "for i in data_pct.Y:\n",
    "    s.append(1 if i>0 else 0)\n",
    "data_pct['Y2']=pd.Series(s,index=data_pct.index)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 457,
   "metadata": {},
   "outputs": [],
   "source": [
    "x=data_pct[['SP 500', 'Crude Oil', 'Dollar Index', 'Wheat', 'Euro', 'GBP', 'Gold']].values\n",
    "y=data_pct.Y.values\n",
    "sc = StandardScaler()\n",
    "x_train = sc.fit_transform(x_train)\n",
    "x_test = sc.transform(x_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 479,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/mrefermat/anaconda3/lib/python3.7/site-packages/ipykernel_launcher.py:5: RuntimeWarning: invalid value encountered in double_scalars\n",
      "  \"\"\"\n"
     ]
    }
   ],
   "source": [
    "res=[]\n",
    "for i in range(0,1000):\n",
    "    x_train, x_test, y_train, y_test = train_test_split(x, y, test_size = 0.25)\n",
    "    y_train_flat= flatten(y_train)\n",
    "    y_test_flat = flatten(y_test)\n",
    "    sc = StandardScaler()\n",
    "    x_train = sc.fit_transform(x_train)\n",
    "    x_test = sc.transform(x_test)\n",
    "    d_train = lgb.Dataset(x_train, label=y_train_flat)\n",
    "    params = {}\n",
    "    params['learning_rate'] = np.random.random()/7\n",
    "    params['boosting_type'] = 'gbdt'\n",
    "    params['objective'] = 'binary'\n",
    "    params['metric'] = 'binary_logloss'\n",
    "    params['sub_feature'] = 0.5\n",
    "    params['num_leaves'] = np.random.randint(5,30)\n",
    "    params['min_data'] = 50\n",
    "    params['max_depth'] = 20 #np.random.randint(5,10)\n",
    "    clf = lgb.train(params, d_train, 100)\n",
    "\n",
    "    # Out of sample IR calculation\n",
    "    y_pred=clf.predict(x_test)\n",
    "    y_pred_bar=[]\n",
    "    for i in range(0,y_pred.size):\n",
    "        if y_pred[i]>=.54:       # setting threshold to .5\n",
    "           y_pred_bar.append(1)\n",
    "        else:  \n",
    "           y_pred_bar.append(0)\n",
    "    y_pred_bar=np.array(y_pred_bar)\n",
    "    IR=calc_IR(y_pred_bar,y_test,255)\n",
    "\n",
    "    # In sample \n",
    "    y_pred_in=clf.predict(x_train)\n",
    "    in_sample=accuracy_score(flatten(y_pred_in),flatten(y_train))\n",
    "    res.append([IR,in_sample])\n",
    "final_res_hp=pd.DataFrame(res,columns=['IR','In Sample'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 480,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x129756390>"
      ]
     },
     "execution_count": 480,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.scatterplot(y='IR',x='In Sample',data=final_res_hp)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 481,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1289e7668>"
      ]
     },
     "execution_count": 481,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "pd.Series(y_pred).plot(kind='hist')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}