dataFiltering.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# CS7641 Machine Learning\n",
    "### <i>Application of Machine Learning in Pairs Trading</i>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data Filtering\n",
    "Our dataset contains two datasets: daily stock price data and monthly firm-level financial ratios data. We preselect stocks from Russel 3000 which can idealy represent 98% of total market shares of the U.S. Equity Market. Then we perform basic data screening and stock filtering to eliminate stocks that would affect our results.\n",
    "\n",
    "- Training set sample period: 2010/01/01 to 2015/12/31, use for clustering and training trading model\n",
    "- Testing period: 2016/01/01 to 2019/12/31, use for testing trading model\n",
    "\n",
    "### 1. Daily Stock Price Dataset\n",
    " - Delete stocks that are delisted during sample period (since we cannot trade them anymore)\n",
    " - Delete stocks that have negative stock price during sample period (will be problematic when trading)\n",
    " - Delete stocks that have long-time low trading volumn (avoid illiquidity trap)\n",
    " - Delete stocks that have more than half missing data (avoid funtionality error for PCA)\n",
    "\n",
    "This left us with 2060 stocks in total.\n",
    "\n",
    "### 2. Monthly Financial Ratios Dataset\n",
    "We use 20 financial ratios, including \n",
    " - Asset Turnover\n",
    " - BM (Book/Market)\n",
    " - Capitalization Ratio\n",
    " - Cash Flow/Total Debt\n",
    " - Cash Flow Margin\n",
    " - Current Ratio\n",
    " - DE ratio (Total Debt/Equity)\n",
    " - Total Debt/EBITDA\n",
    " - Free Cash Flow/Operating Cash Flow\n",
    " - Gross Profit/Total Assets\n",
    " - Interest Coverage Ratio\n",
    " - Long-term Debt/Total Liabilities\n",
    " - Net Profit Margin\n",
    " - P/E (Diluted, Excluding El)\n",
    " - Pre-tax Return on Total Earning Assets\n",
    " - Return on Assets\n",
    " - Return on Equity\n",
    " - Price/Book\n",
    " - Quick Ratio\n",
    " - Total Debt/Invested Capital\n",
    "\n",
    "We may exclude some ratios if their are too many missing data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import os\n",
    "import datetime\n",
    "import missingno as msno"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1. Stock Price"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/Sharonvy/opt/anaconda3/lib/python3.7/site-packages/IPython/core/interactiveshell.py:3058: DtypeWarning: Columns (2,7,9) have mixed types. Specify dtype option on import or set low_memory=False.\n",
      "  interactivity=interactivity, compiler=compiler, result=result)\n"
     ]
    }
   ],
   "source": [
    "#import stock price dataset\n",
    "price = pd.read_csv(\"~/Downloads/russel3000PriceDaily.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>PERMNO</th>\n",
       "      <th>date</th>\n",
       "      <th>SICCD</th>\n",
       "      <th>TICKER</th>\n",
       "      <th>DLSTCD</th>\n",
       "      <th>PRC</th>\n",
       "      <th>VOL</th>\n",
       "      <th>RET</th>\n",
       "      <th>SHROUT</th>\n",
       "      <th>RETX</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>10025</td>\n",
       "      <td>20070103</td>\n",
       "      <td>3070</td>\n",
       "      <td>AEPI</td>\n",
       "      <td>NaN</td>\n",
       "      <td>49.06</td>\n",
       "      <td>112525.0</td>\n",
       "      <td>-0.079722</td>\n",
       "      <td>7886.0</td>\n",
       "      <td>-0.079722</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>10025</td>\n",
       "      <td>20070104</td>\n",
       "      <td>3070</td>\n",
       "      <td>AEPI</td>\n",
       "      <td>NaN</td>\n",
       "      <td>50.34</td>\n",
       "      <td>87697.0</td>\n",
       "      <td>0.026090</td>\n",
       "      <td>7886.0</td>\n",
       "      <td>0.026090</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>10025</td>\n",
       "      <td>20070105</td>\n",
       "      <td>3070</td>\n",
       "      <td>AEPI</td>\n",
       "      <td>NaN</td>\n",
       "      <td>49.63</td>\n",
       "      <td>43162.0</td>\n",
       "      <td>-0.014104</td>\n",
       "      <td>7886.0</td>\n",
       "      <td>-0.014104</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>10025</td>\n",
       "      <td>20070108</td>\n",
       "      <td>3070</td>\n",
       "      <td>AEPI</td>\n",
       "      <td>NaN</td>\n",
       "      <td>49.50</td>\n",
       "      <td>37778.0</td>\n",
       "      <td>-0.002619</td>\n",
       "      <td>7886.0</td>\n",
       "      <td>-0.002619</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>10025</td>\n",
       "      <td>20070109</td>\n",
       "      <td>3070</td>\n",
       "      <td>AEPI</td>\n",
       "      <td>NaN</td>\n",
       "      <td>50.62</td>\n",
       "      <td>40393.0</td>\n",
       "      <td>0.022626</td>\n",
       "      <td>7886.0</td>\n",
       "      <td>0.022626</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   PERMNO      date SICCD TICKER  DLSTCD    PRC       VOL        RET  SHROUT  \\\n",
       "0   10025  20070103  3070   AEPI     NaN  49.06  112525.0  -0.079722  7886.0   \n",
       "1   10025  20070104  3070   AEPI     NaN  50.34   87697.0   0.026090  7886.0   \n",
       "2   10025  20070105  3070   AEPI     NaN  49.63   43162.0  -0.014104  7886.0   \n",
       "3   10025  20070108  3070   AEPI     NaN  49.50   37778.0  -0.002619  7886.0   \n",
       "4   10025  20070109  3070   AEPI     NaN  50.62   40393.0   0.022626  7886.0   \n",
       "\n",
       "        RETX  \n",
       "0  -0.079722  \n",
       "1   0.026090  \n",
       "2  -0.014104  \n",
       "3  -0.002619  \n",
       "4   0.022626  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#take a glance of original price dataset\n",
    "price.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Original dataset size:  8497689\n",
      "Original number of stocks:  3292\n"
     ]
    }
   ],
   "source": [
    "print(\"Original dataset size: \", len(price))\n",
    "print(\"Original number of stocks: \", len(set(price['PERMNO'])))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "#preprocessing\n",
    "price['_date'] = pd.to_datetime(price['date'], format = '%Y%m%d')\n",
    "price['turnover'] = price['VOL']/price['SHROUT']\n",
    "\n",
    "#delisted stocks\n",
    "dlst = price[price['DLSTCD'] >= 300]        #those are companies that are liquidated, dropped or exhanged during sample period\n",
    "dlst_list = list(set(dlst['PERMNO']))       #and should be exclude in our training set (since we cannot trade them anymore)\n",
    "\n",
    "#negative price stocks\n",
    "neg_price = price[price['PRC'] <= 0]        #delete stocks with negative price (will be problematic when trading)\n",
    "neg_price_list = list(set(neg_price['PERMNO']))\n",
    "\n",
    "#low trading volume\n",
    "low_vol = price[price['turnover'] <= 0.1]   #if the daily trading volume is less than 0.1% of its total shares outstanding\n",
    "low_vol_list = list(set(low_vol['PERMNO']))\n",
    "for item in low_vol_list:\n",
    "    temp = low_vol.loc[low_vol['PERMNO'] == item]\n",
    "    if len(temp) < 100:                     #avoid deleting too many stocks (data collection error or temporary derail)\n",
    "        low_vol_list.remove(item)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "#stocks that we want to exclude\n",
    "final_list = list(set(dlst_list) | set(neg_price_list) | set(low_vol_list))\n",
    "#final dataset for price\n",
    "stock_price = price.loc[price['PERMNO'].isin(final_list) == False]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dataset size:  7255941\n",
      "Number of stocks:  2740\n"
     ]
    }
   ],
   "source": [
    "print(\"Dataset size: \", len(stock_price))\n",
    "print(\"Number of stocks: \", len(set(stock_price['PERMNO'])))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "#set training set sample period for clsutering analysis, from 2010/01/01 to 2015/12/31\n",
    "#and set testing dataset from 2016/01/01 to 2019/12/31\n",
    "price_train = stock_price.loc[stock_price['date'] < 20160101]\n",
    "price_test = stock_price.loc[stock_price['date'] >= 20160101]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "#get the pivot table, with stock permno as index and datetime as columns\n",
    "df_price = pd.pivot_table(price_train, values = 'PRC', index = 'PERMNO', columns = 'date')\n",
    "df_price = df_price.add_prefix('price_')\n",
    "#last step, delete stocks with more than half nan price for clustering\n",
    "df_price = df_price.dropna(thresh = len(df_price.columns)//2)\n",
    "df_price = df_price.fillna(method = 'ffill', axis = 1, limit = 5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>date</th>\n",
       "      <th>price_20070103</th>\n",
       "      <th>price_20070104</th>\n",
       "      <th>price_20070105</th>\n",
       "      <th>price_20070108</th>\n",
       "      <th>price_20070109</th>\n",
       "      <th>price_20070110</th>\n",
       "      <th>price_20070111</th>\n",
       "      <th>price_20070112</th>\n",
       "      <th>price_20070116</th>\n",
       "      <th>price_20070117</th>\n",
       "      <th>...</th>\n",
       "      <th>price_20151217</th>\n",
       "      <th>price_20151218</th>\n",
       "      <th>price_20151221</th>\n",
       "      <th>price_20151222</th>\n",
       "      <th>price_20151223</th>\n",
       "      <th>price_20151224</th>\n",
       "      <th>price_20151228</th>\n",
       "      <th>price_20151229</th>\n",
       "      <th>price_20151230</th>\n",
       "      <th>price_20151231</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>PERMNO</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>10025</td>\n",
       "      <td>49.06</td>\n",
       "      <td>50.34</td>\n",
       "      <td>49.63</td>\n",
       "      <td>49.50</td>\n",
       "      <td>50.62</td>\n",
       "      <td>49.68</td>\n",
       "      <td>49.20</td>\n",
       "      <td>48.90</td>\n",
       "      <td>46.83</td>\n",
       "      <td>47.05</td>\n",
       "      <td>...</td>\n",
       "      <td>80.80</td>\n",
       "      <td>79.01</td>\n",
       "      <td>76.40</td>\n",
       "      <td>77.93</td>\n",
       "      <td>78.98</td>\n",
       "      <td>80.55</td>\n",
       "      <td>78.97</td>\n",
       "      <td>77.15</td>\n",
       "      <td>77.22</td>\n",
       "      <td>77.15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10026</td>\n",
       "      <td>41.02</td>\n",
       "      <td>41.89</td>\n",
       "      <td>39.66</td>\n",
       "      <td>40.42</td>\n",
       "      <td>40.54</td>\n",
       "      <td>41.54</td>\n",
       "      <td>41.41</td>\n",
       "      <td>42.23</td>\n",
       "      <td>41.71</td>\n",
       "      <td>41.28</td>\n",
       "      <td>...</td>\n",
       "      <td>114.37</td>\n",
       "      <td>112.69</td>\n",
       "      <td>112.92</td>\n",
       "      <td>114.53</td>\n",
       "      <td>115.90</td>\n",
       "      <td>118.21</td>\n",
       "      <td>117.64</td>\n",
       "      <td>119.37</td>\n",
       "      <td>119.25</td>\n",
       "      <td>116.67</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10032</td>\n",
       "      <td>24.12</td>\n",
       "      <td>24.30</td>\n",
       "      <td>23.89</td>\n",
       "      <td>23.70</td>\n",
       "      <td>23.13</td>\n",
       "      <td>20.90</td>\n",
       "      <td>20.84</td>\n",
       "      <td>21.58</td>\n",
       "      <td>20.79</td>\n",
       "      <td>20.58</td>\n",
       "      <td>...</td>\n",
       "      <td>35.26</td>\n",
       "      <td>34.99</td>\n",
       "      <td>35.22</td>\n",
       "      <td>35.08</td>\n",
       "      <td>35.95</td>\n",
       "      <td>36.05</td>\n",
       "      <td>35.67</td>\n",
       "      <td>35.92</td>\n",
       "      <td>35.46</td>\n",
       "      <td>34.92</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10051</td>\n",
       "      <td>7.40</td>\n",
       "      <td>7.33</td>\n",
       "      <td>7.52</td>\n",
       "      <td>7.44</td>\n",
       "      <td>7.51</td>\n",
       "      <td>7.52</td>\n",
       "      <td>7.45</td>\n",
       "      <td>7.60</td>\n",
       "      <td>7.83</td>\n",
       "      <td>7.82</td>\n",
       "      <td>...</td>\n",
       "      <td>17.53</td>\n",
       "      <td>16.77</td>\n",
       "      <td>16.52</td>\n",
       "      <td>17.08</td>\n",
       "      <td>17.29</td>\n",
       "      <td>17.38</td>\n",
       "      <td>16.96</td>\n",
       "      <td>16.95</td>\n",
       "      <td>16.85</td>\n",
       "      <td>16.45</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10104</td>\n",
       "      <td>17.51</td>\n",
       "      <td>17.68</td>\n",
       "      <td>17.64</td>\n",
       "      <td>17.86</td>\n",
       "      <td>17.82</td>\n",
       "      <td>17.77</td>\n",
       "      <td>17.39</td>\n",
       "      <td>17.50</td>\n",
       "      <td>17.30</td>\n",
       "      <td>17.52</td>\n",
       "      <td>...</td>\n",
       "      <td>36.93</td>\n",
       "      <td>36.35</td>\n",
       "      <td>36.42</td>\n",
       "      <td>36.59</td>\n",
       "      <td>36.93</td>\n",
       "      <td>37.01</td>\n",
       "      <td>36.79</td>\n",
       "      <td>37.04</td>\n",
       "      <td>36.92</td>\n",
       "      <td>36.53</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 2266 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "date    price_20070103  price_20070104  price_20070105  price_20070108  \\\n",
       "PERMNO                                                                   \n",
       "10025            49.06           50.34           49.63           49.50   \n",
       "10026            41.02           41.89           39.66           40.42   \n",
       "10032            24.12           24.30           23.89           23.70   \n",
       "10051             7.40            7.33            7.52            7.44   \n",
       "10104            17.51           17.68           17.64           17.86   \n",
       "\n",
       "date    price_20070109  price_20070110  price_20070111  price_20070112  \\\n",
       "PERMNO                                                                   \n",
       "10025            50.62           49.68           49.20           48.90   \n",
       "10026            40.54           41.54           41.41           42.23   \n",
       "10032            23.13           20.90           20.84           21.58   \n",
       "10051             7.51            7.52            7.45            7.60   \n",
       "10104            17.82           17.77           17.39           17.50   \n",
       "\n",
       "date    price_20070116  price_20070117  ...  price_20151217  price_20151218  \\\n",
       "PERMNO                                  ...                                   \n",
       "10025            46.83           47.05  ...           80.80           79.01   \n",
       "10026            41.71           41.28  ...          114.37          112.69   \n",
       "10032            20.79           20.58  ...           35.26           34.99   \n",
       "10051             7.83            7.82  ...           17.53           16.77   \n",
       "10104            17.30           17.52  ...           36.93           36.35   \n",
       "\n",
       "date    price_20151221  price_20151222  price_20151223  price_20151224  \\\n",
       "PERMNO                                                                   \n",
       "10025            76.40           77.93           78.98           80.55   \n",
       "10026           112.92          114.53          115.90          118.21   \n",
       "10032            35.22           35.08           35.95           36.05   \n",
       "10051            16.52           17.08           17.29           17.38   \n",
       "10104            36.42           36.59           36.93           37.01   \n",
       "\n",
       "date    price_20151228  price_20151229  price_20151230  price_20151231  \n",
       "PERMNO                                                                  \n",
       "10025            78.97           77.15           77.22           77.15  \n",
       "10026           117.64          119.37          119.25          116.67  \n",
       "10032            35.67           35.92           35.46           34.92  \n",
       "10051            16.96           16.95           16.85           16.45  \n",
       "10104            36.79           37.04           36.92           36.53  \n",
       "\n",
       "[5 rows x 2266 columns]"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_price.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of training stocks:  2105\n"
     ]
    }
   ],
   "source": [
    "print(\"Number of training stocks: \", len(df_price))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "#get the list permno for further data extraction\n",
    "permno = list(df_price.index)\n",
    "with open('permno_list.txt', 'w') as filehandle:    #write to file\n",
    "    for listitem in permno:\n",
    "        filehandle.write('%i\\n' % listitem)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_price.to_csv('stock_price_table.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2. Financial Ratios"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>permno</th>\n",
       "      <th>adate</th>\n",
       "      <th>qdate</th>\n",
       "      <th>public_date</th>\n",
       "      <th>bm</th>\n",
       "      <th>pe_exi</th>\n",
       "      <th>npm</th>\n",
       "      <th>cfm</th>\n",
       "      <th>roa</th>\n",
       "      <th>roe</th>\n",
       "      <th>GProf</th>\n",
       "      <th>totdebt_invcap</th>\n",
       "      <th>capital_ratio</th>\n",
       "      <th>debt_ebitda</th>\n",
       "      <th>lt_debt</th>\n",
       "      <th>cash_debt</th>\n",
       "      <th>fcf_ocf</th>\n",
       "      <th>de_ratio</th>\n",
       "      <th>at_turn</th>\n",
       "      <th>ptb</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>10025</td>\n",
       "      <td>20091031.0</td>\n",
       "      <td>20091031</td>\n",
       "      <td>20100131</td>\n",
       "      <td>0.365</td>\n",
       "      <td>7.845</td>\n",
       "      <td>0.041</td>\n",
       "      <td>0.068</td>\n",
       "      <td>0.212</td>\n",
       "      <td>0.473</td>\n",
       "      <td>0.498</td>\n",
       "      <td>0.704</td>\n",
       "      <td>0.698</td>\n",
       "      <td>2.221</td>\n",
       "      <td>0.615</td>\n",
       "      <td>0.31</td>\n",
       "      <td>0.729</td>\n",
       "      <td>3.750</td>\n",
       "      <td>1.984</td>\n",
       "      <td>2.740</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>10025</td>\n",
       "      <td>20091031.0</td>\n",
       "      <td>20091031</td>\n",
       "      <td>20100228</td>\n",
       "      <td>0.365</td>\n",
       "      <td>7.843</td>\n",
       "      <td>0.041</td>\n",
       "      <td>0.068</td>\n",
       "      <td>0.212</td>\n",
       "      <td>0.473</td>\n",
       "      <td>0.498</td>\n",
       "      <td>0.704</td>\n",
       "      <td>0.698</td>\n",
       "      <td>2.221</td>\n",
       "      <td>0.615</td>\n",
       "      <td>0.31</td>\n",
       "      <td>0.729</td>\n",
       "      <td>3.750</td>\n",
       "      <td>1.984</td>\n",
       "      <td>2.740</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>10025</td>\n",
       "      <td>20091031.0</td>\n",
       "      <td>20100131</td>\n",
       "      <td>20100331</td>\n",
       "      <td>0.338</td>\n",
       "      <td>13.208</td>\n",
       "      <td>0.019</td>\n",
       "      <td>0.047</td>\n",
       "      <td>0.148</td>\n",
       "      <td>0.183</td>\n",
       "      <td>0.427</td>\n",
       "      <td>0.727</td>\n",
       "      <td>0.722</td>\n",
       "      <td>3.620</td>\n",
       "      <td>0.657</td>\n",
       "      <td>0.31</td>\n",
       "      <td>0.289</td>\n",
       "      <td>3.963</td>\n",
       "      <td>2.055</td>\n",
       "      <td>2.204</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>10025</td>\n",
       "      <td>20091031.0</td>\n",
       "      <td>20100131</td>\n",
       "      <td>20100430</td>\n",
       "      <td>0.338</td>\n",
       "      <td>14.025</td>\n",
       "      <td>0.019</td>\n",
       "      <td>0.047</td>\n",
       "      <td>0.148</td>\n",
       "      <td>0.183</td>\n",
       "      <td>0.427</td>\n",
       "      <td>0.727</td>\n",
       "      <td>0.722</td>\n",
       "      <td>3.620</td>\n",
       "      <td>0.657</td>\n",
       "      <td>0.31</td>\n",
       "      <td>0.289</td>\n",
       "      <td>3.963</td>\n",
       "      <td>2.055</td>\n",
       "      <td>2.340</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>10025</td>\n",
       "      <td>20091031.0</td>\n",
       "      <td>20100131</td>\n",
       "      <td>20100531</td>\n",
       "      <td>0.338</td>\n",
       "      <td>12.761</td>\n",
       "      <td>0.019</td>\n",
       "      <td>0.047</td>\n",
       "      <td>0.148</td>\n",
       "      <td>0.183</td>\n",
       "      <td>0.427</td>\n",
       "      <td>0.727</td>\n",
       "      <td>0.722</td>\n",
       "      <td>3.620</td>\n",
       "      <td>0.657</td>\n",
       "      <td>0.31</td>\n",
       "      <td>0.289</td>\n",
       "      <td>3.963</td>\n",
       "      <td>2.055</td>\n",
       "      <td>2.129</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   permno       adate     qdate  public_date     bm  pe_exi    npm    cfm  \\\n",
       "0   10025  20091031.0  20091031     20100131  0.365   7.845  0.041  0.068   \n",
       "1   10025  20091031.0  20091031     20100228  0.365   7.843  0.041  0.068   \n",
       "2   10025  20091031.0  20100131     20100331  0.338  13.208  0.019  0.047   \n",
       "3   10025  20091031.0  20100131     20100430  0.338  14.025  0.019  0.047   \n",
       "4   10025  20091031.0  20100131     20100531  0.338  12.761  0.019  0.047   \n",
       "\n",
       "     roa    roe  GProf  totdebt_invcap  capital_ratio  debt_ebitda  lt_debt  \\\n",
       "0  0.212  0.473  0.498           0.704          0.698        2.221    0.615   \n",
       "1  0.212  0.473  0.498           0.704          0.698        2.221    0.615   \n",
       "2  0.148  0.183  0.427           0.727          0.722        3.620    0.657   \n",
       "3  0.148  0.183  0.427           0.727          0.722        3.620    0.657   \n",
       "4  0.148  0.183  0.427           0.727          0.722        3.620    0.657   \n",
       "\n",
       "   cash_debt  fcf_ocf  de_ratio  at_turn    ptb  \n",
       "0       0.31    0.729     3.750    1.984  2.740  \n",
       "1       0.31    0.729     3.750    1.984  2.740  \n",
       "2       0.31    0.289     3.963    2.055  2.204  \n",
       "3       0.31    0.289     3.963    2.055  2.340  \n",
       "4       0.31    0.289     3.963    2.055  2.129  "
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#import financial ratios dataset\n",
    "ratios = pd.read_csv(\"~/Downloads/financialRatiosMonthly_new.csv\")\n",
    "ratios.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>permno</th>\n",
       "      <th>adate</th>\n",
       "      <th>qdate</th>\n",
       "      <th>public_date</th>\n",
       "      <th>bm</th>\n",
       "      <th>pe_exi</th>\n",
       "      <th>npm</th>\n",
       "      <th>cfm</th>\n",
       "      <th>roa</th>\n",
       "      <th>roe</th>\n",
       "      <th>...</th>\n",
       "      <th>totdebt_invcap</th>\n",
       "      <th>capital_ratio</th>\n",
       "      <th>debt_ebitda</th>\n",
       "      <th>lt_debt</th>\n",
       "      <th>cash_debt</th>\n",
       "      <th>fcf_ocf</th>\n",
       "      <th>de_ratio</th>\n",
       "      <th>at_turn</th>\n",
       "      <th>ptb</th>\n",
       "      <th>_date</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>210297</td>\n",
       "      <td>93423</td>\n",
       "      <td>20141231.0</td>\n",
       "      <td>20150930</td>\n",
       "      <td>20151231</td>\n",
       "      <td>0.095</td>\n",
       "      <td>45.783</td>\n",
       "      <td>0.096</td>\n",
       "      <td>0.217</td>\n",
       "      <td>0.159</td>\n",
       "      <td>0.292</td>\n",
       "      <td>...</td>\n",
       "      <td>0.718</td>\n",
       "      <td>0.900</td>\n",
       "      <td>3.557</td>\n",
       "      <td>0.726</td>\n",
       "      <td>0.209</td>\n",
       "      <td>0.768</td>\n",
       "      <td>12.406</td>\n",
       "      <td>0.472</td>\n",
       "      <td>12.501</td>\n",
       "      <td>2015-12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>210409</td>\n",
       "      <td>93426</td>\n",
       "      <td>20141231.0</td>\n",
       "      <td>20150930</td>\n",
       "      <td>20151231</td>\n",
       "      <td>1.229</td>\n",
       "      <td>-34.303</td>\n",
       "      <td>-0.019</td>\n",
       "      <td>0.030</td>\n",
       "      <td>0.077</td>\n",
       "      <td>-0.022</td>\n",
       "      <td>...</td>\n",
       "      <td>0.102</td>\n",
       "      <td>0.062</td>\n",
       "      <td>0.963</td>\n",
       "      <td>0.160</td>\n",
       "      <td>0.270</td>\n",
       "      <td>0.305</td>\n",
       "      <td>0.416</td>\n",
       "      <td>0.847</td>\n",
       "      <td>0.733</td>\n",
       "      <td>2015-12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>210522</td>\n",
       "      <td>93428</td>\n",
       "      <td>20141231.0</td>\n",
       "      <td>20150930</td>\n",
       "      <td>20151231</td>\n",
       "      <td>0.284</td>\n",
       "      <td>-392.890</td>\n",
       "      <td>-0.009</td>\n",
       "      <td>0.050</td>\n",
       "      <td>0.057</td>\n",
       "      <td>-0.010</td>\n",
       "      <td>...</td>\n",
       "      <td>0.342</td>\n",
       "      <td>0.342</td>\n",
       "      <td>4.661</td>\n",
       "      <td>0.466</td>\n",
       "      <td>0.243</td>\n",
       "      <td>0.702</td>\n",
       "      <td>1.117</td>\n",
       "      <td>0.521</td>\n",
       "      <td>4.192</td>\n",
       "      <td>2015-12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>210612</td>\n",
       "      <td>93429</td>\n",
       "      <td>20141231.0</td>\n",
       "      <td>20150930</td>\n",
       "      <td>20151231</td>\n",
       "      <td>0.049</td>\n",
       "      <td>26.708</td>\n",
       "      <td>0.316</td>\n",
       "      <td>0.386</td>\n",
       "      <td>1.010</td>\n",
       "      <td>0.799</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>1.963</td>\n",
       "      <td>0.864</td>\n",
       "      <td>0.549</td>\n",
       "      <td>1.698</td>\n",
       "      <td>19.610</td>\n",
       "      <td>2015-12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>210725</td>\n",
       "      <td>93436</td>\n",
       "      <td>20141231.0</td>\n",
       "      <td>20150930</td>\n",
       "      <td>20151231</td>\n",
       "      <td>0.040</td>\n",
       "      <td>-45.030</td>\n",
       "      <td>-0.178</td>\n",
       "      <td>-0.087</td>\n",
       "      <td>-0.031</td>\n",
       "      <td>-0.791</td>\n",
       "      <td>...</td>\n",
       "      <td>0.876</td>\n",
       "      <td>0.659</td>\n",
       "      <td>-13.879</td>\n",
       "      <td>0.350</td>\n",
       "      <td>-0.012</td>\n",
       "      <td>NaN</td>\n",
       "      <td>5.533</td>\n",
       "      <td>0.583</td>\n",
       "      <td>23.994</td>\n",
       "      <td>2015-12</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 21 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        permno       adate     qdate  public_date     bm   pe_exi    npm  \\\n",
       "210297   93423  20141231.0  20150930     20151231  0.095   45.783  0.096   \n",
       "210409   93426  20141231.0  20150930     20151231  1.229  -34.303 -0.019   \n",
       "210522   93428  20141231.0  20150930     20151231  0.284 -392.890 -0.009   \n",
       "210612   93429  20141231.0  20150930     20151231  0.049   26.708  0.316   \n",
       "210725   93436  20141231.0  20150930     20151231  0.040  -45.030 -0.178   \n",
       "\n",
       "          cfm    roa    roe  ...  totdebt_invcap  capital_ratio  debt_ebitda  \\\n",
       "210297  0.217  0.159  0.292  ...           0.718          0.900        3.557   \n",
       "210409  0.030  0.077 -0.022  ...           0.102          0.062        0.963   \n",
       "210522  0.050  0.057 -0.010  ...           0.342          0.342        4.661   \n",
       "210612  0.386  1.010  0.799  ...           0.000          0.000        0.000   \n",
       "210725 -0.087 -0.031 -0.791  ...           0.876          0.659      -13.879   \n",
       "\n",
       "        lt_debt  cash_debt  fcf_ocf  de_ratio  at_turn     ptb    _date  \n",
       "210297    0.726      0.209    0.768    12.406    0.472  12.501  2015-12  \n",
       "210409    0.160      0.270    0.305     0.416    0.847   0.733  2015-12  \n",
       "210522    0.466      0.243    0.702     1.117    0.521   4.192  2015-12  \n",
       "210612    0.000      1.963    0.864     0.549    1.698  19.610  2015-12  \n",
       "210725    0.350     -0.012      NaN     5.533    0.583  23.994  2015-12  \n",
       "\n",
       "[5 rows x 21 columns]"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#since the public date is already two month behind qdate (defined to be), we do not need to lag the data anymore\n",
    "ratios['_date'] = pd.to_datetime(ratios['public_date'], format = '%Y%m%d').dt.to_period('M')\n",
    "#select only training period\n",
    "ratios_train = ratios.loc[(ratios['public_date'] >= 20151201) & (ratios['public_date']<= 20151231)]\n",
    "ratios_train.tail()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1a30e83b10>"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1800x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#visualize missing data\n",
    "msno.matrix(ratios_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "ratio_cols = ['permno'] + list(ratios_train.columns[4:-1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>permno</th>\n",
       "      <th>bm</th>\n",
       "      <th>pe_exi</th>\n",
       "      <th>npm</th>\n",
       "      <th>cfm</th>\n",
       "      <th>roa</th>\n",
       "      <th>roe</th>\n",
       "      <th>GProf</th>\n",
       "      <th>totdebt_invcap</th>\n",
       "      <th>capital_ratio</th>\n",
       "      <th>debt_ebitda</th>\n",
       "      <th>lt_debt</th>\n",
       "      <th>cash_debt</th>\n",
       "      <th>fcf_ocf</th>\n",
       "      <th>de_ratio</th>\n",
       "      <th>at_turn</th>\n",
       "      <th>ptb</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>10025</td>\n",
       "      <td>0.273</td>\n",
       "      <td>13.679</td>\n",
       "      <td>0.025</td>\n",
       "      <td>0.054</td>\n",
       "      <td>0.210</td>\n",
       "      <td>0.308</td>\n",
       "      <td>0.484</td>\n",
       "      <td>0.710</td>\n",
       "      <td>0.702</td>\n",
       "      <td>2.296</td>\n",
       "      <td>0.602</td>\n",
       "      <td>0.201</td>\n",
       "      <td>0.827</td>\n",
       "      <td>3.910</td>\n",
       "      <td>2.573</td>\n",
       "      <td>3.534</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>10026</td>\n",
       "      <td>0.303</td>\n",
       "      <td>31.279</td>\n",
       "      <td>0.072</td>\n",
       "      <td>0.111</td>\n",
       "      <td>0.205</td>\n",
       "      <td>0.112</td>\n",
       "      <td>0.456</td>\n",
       "      <td>0.002</td>\n",
       "      <td>0.002</td>\n",
       "      <td>0.010</td>\n",
       "      <td>0.008</td>\n",
       "      <td>0.736</td>\n",
       "      <td>0.524</td>\n",
       "      <td>0.238</td>\n",
       "      <td>1.349</td>\n",
       "      <td>3.385</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>10032</td>\n",
       "      <td>0.659</td>\n",
       "      <td>12.745</td>\n",
       "      <td>0.036</td>\n",
       "      <td>0.054</td>\n",
       "      <td>0.100</td>\n",
       "      <td>0.115</td>\n",
       "      <td>0.169</td>\n",
       "      <td>0.239</td>\n",
       "      <td>0.235</td>\n",
       "      <td>1.588</td>\n",
       "      <td>0.301</td>\n",
       "      <td>0.089</td>\n",
       "      <td>0.542</td>\n",
       "      <td>1.021</td>\n",
       "      <td>1.603</td>\n",
       "      <td>1.364</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>10051</td>\n",
       "      <td>0.981</td>\n",
       "      <td>-548.330</td>\n",
       "      <td>-0.001</td>\n",
       "      <td>0.037</td>\n",
       "      <td>0.058</td>\n",
       "      <td>-0.002</td>\n",
       "      <td>0.058</td>\n",
       "      <td>0.553</td>\n",
       "      <td>0.524</td>\n",
       "      <td>7.474</td>\n",
       "      <td>0.658</td>\n",
       "      <td>0.061</td>\n",
       "      <td>0.373</td>\n",
       "      <td>1.675</td>\n",
       "      <td>0.834</td>\n",
       "      <td>1.216</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>10104</td>\n",
       "      <td>0.297</td>\n",
       "      <td>17.070</td>\n",
       "      <td>0.250</td>\n",
       "      <td>0.326</td>\n",
       "      <td>0.164</td>\n",
       "      <td>0.198</td>\n",
       "      <td>0.298</td>\n",
       "      <td>0.443</td>\n",
       "      <td>0.427</td>\n",
       "      <td>2.221</td>\n",
       "      <td>0.635</td>\n",
       "      <td>0.232</td>\n",
       "      <td>0.878</td>\n",
       "      <td>1.173</td>\n",
       "      <td>0.365</td>\n",
       "      <td>3.259</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   permno     bm   pe_exi    npm    cfm    roa    roe  GProf  totdebt_invcap  \\\n",
       "0   10025  0.273   13.679  0.025  0.054  0.210  0.308  0.484           0.710   \n",
       "1   10026  0.303   31.279  0.072  0.111  0.205  0.112  0.456           0.002   \n",
       "2   10032  0.659   12.745  0.036  0.054  0.100  0.115  0.169           0.239   \n",
       "3   10051  0.981 -548.330 -0.001  0.037  0.058 -0.002  0.058           0.553   \n",
       "4   10104  0.297   17.070  0.250  0.326  0.164  0.198  0.298           0.443   \n",
       "\n",
       "   capital_ratio  debt_ebitda  lt_debt  cash_debt  fcf_ocf  de_ratio  at_turn  \\\n",
       "0          0.702        2.296    0.602      0.201    0.827     3.910    2.573   \n",
       "1          0.002        0.010    0.008      0.736    0.524     0.238    1.349   \n",
       "2          0.235        1.588    0.301      0.089    0.542     1.021    1.603   \n",
       "3          0.524        7.474    0.658      0.061    0.373     1.675    0.834   \n",
       "4          0.427        2.221    0.635      0.232    0.878     1.173    0.365   \n",
       "\n",
       "     ptb  \n",
       "0  3.534  \n",
       "1  3.385  \n",
       "2  1.364  \n",
       "3  1.216  \n",
       "4  3.259  "
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#select only those columns\n",
    "ratios_data = ratios_train.loc[:,ratios_train.columns.isin(ratio_cols)].reset_index(drop = True)\n",
    "ratios_data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of stocks in ratios dataset:  1795\n"
     ]
    }
   ],
   "source": [
    "print(\"Number of stocks in ratios dataset: \", len(ratios_data))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [],
   "source": [
    "#save training set to file\n",
    "ratios_data.to_csv('ratios_data_table.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>PERMNO</th>\n",
       "      <th>date</th>\n",
       "      <th>SICCD</th>\n",
       "      <th>TICKER</th>\n",
       "      <th>DLSTCD</th>\n",
       "      <th>PRC</th>\n",
       "      <th>VOL</th>\n",
       "      <th>RET</th>\n",
       "      <th>SHROUT</th>\n",
       "      <th>RETX</th>\n",
       "      <th>_date</th>\n",
       "      <th>turnover</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>2266</td>\n",
       "      <td>10025</td>\n",
       "      <td>20160104</td>\n",
       "      <td>3081</td>\n",
       "      <td>AEPI</td>\n",
       "      <td>NaN</td>\n",
       "      <td>73.19</td>\n",
       "      <td>29246.0</td>\n",
       "      <td>-0.051329</td>\n",
       "      <td>5103.0</td>\n",
       "      <td>-0.051329</td>\n",
       "      <td>2016-01-04</td>\n",
       "      <td>5.731139</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2267</td>\n",
       "      <td>10025</td>\n",
       "      <td>20160105</td>\n",
       "      <td>3081</td>\n",
       "      <td>AEPI</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.40</td>\n",
       "      <td>56220.0</td>\n",
       "      <td>0.016532</td>\n",
       "      <td>5103.0</td>\n",
       "      <td>0.016532</td>\n",
       "      <td>2016-01-05</td>\n",
       "      <td>11.017049</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2268</td>\n",
       "      <td>10025</td>\n",
       "      <td>20160106</td>\n",
       "      <td>3081</td>\n",
       "      <td>AEPI</td>\n",
       "      <td>NaN</td>\n",
       "      <td>77.49</td>\n",
       "      <td>53924.0</td>\n",
       "      <td>0.041532</td>\n",
       "      <td>5103.0</td>\n",
       "      <td>0.041532</td>\n",
       "      <td>2016-01-06</td>\n",
       "      <td>10.567117</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2269</td>\n",
       "      <td>10025</td>\n",
       "      <td>20160107</td>\n",
       "      <td>3081</td>\n",
       "      <td>AEPI</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.50</td>\n",
       "      <td>20810.0</td>\n",
       "      <td>-0.038586</td>\n",
       "      <td>5103.0</td>\n",
       "      <td>-0.038586</td>\n",
       "      <td>2016-01-07</td>\n",
       "      <td>4.077993</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2270</td>\n",
       "      <td>10025</td>\n",
       "      <td>20160108</td>\n",
       "      <td>3081</td>\n",
       "      <td>AEPI</td>\n",
       "      <td>NaN</td>\n",
       "      <td>74.82</td>\n",
       "      <td>47478.0</td>\n",
       "      <td>0.004295</td>\n",
       "      <td>5103.0</td>\n",
       "      <td>0.004295</td>\n",
       "      <td>2016-01-08</td>\n",
       "      <td>9.303939</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      PERMNO      date SICCD TICKER  DLSTCD    PRC      VOL        RET  \\\n",
       "2266   10025  20160104  3081   AEPI     NaN  73.19  29246.0  -0.051329   \n",
       "2267   10025  20160105  3081   AEPI     NaN  74.40  56220.0   0.016532   \n",
       "2268   10025  20160106  3081   AEPI     NaN  77.49  53924.0   0.041532   \n",
       "2269   10025  20160107  3081   AEPI     NaN  74.50  20810.0  -0.038586   \n",
       "2270   10025  20160108  3081   AEPI     NaN  74.82  47478.0   0.004295   \n",
       "\n",
       "      SHROUT       RETX      _date   turnover  \n",
       "2266  5103.0  -0.051329 2016-01-04   5.731139  \n",
       "2267  5103.0   0.016532 2016-01-05  11.017049  \n",
       "2268  5103.0   0.041532 2016-01-06  10.567117  \n",
       "2269  5103.0  -0.038586 2016-01-07   4.077993  \n",
       "2270  5103.0   0.004295 2016-01-08   9.303939  "
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#filtering stocks for testing dataset\n",
    "testing_set = price_test.loc[price_test['PERMNO'].isin(list(set(ratios['permno'])))]\n",
    "testing_set.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [],
   "source": [
    "testing_set.to_csv('tetsing_data.csv')"
   ]
  },
  {
   "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}