12_gradient_boosting_machines/04_preparing_the_model_data.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Preparing Alpha Factors and Features to predict Stock Returns"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Imports & Settings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:53:33.508385Z",
     "start_time": "2020-06-19T23:53:33.506026Z"
    }
   },
   "outputs": [],
   "source": [
    "import warnings\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:53:33.968355Z",
     "start_time": "2020-06-19T23:53:33.510143Z"
    }
   },
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import talib\n",
    "from talib import RSI, BBANDS, MACD, ATR"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:53:33.971290Z",
     "start_time": "2020-06-19T23:53:33.969435Z"
    }
   },
   "outputs": [],
   "source": [
    "MONTH = 21\n",
    "YEAR = 12 * MONTH"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:53:33.987810Z",
     "start_time": "2020-06-19T23:53:33.972463Z"
    }
   },
   "outputs": [],
   "source": [
    "START = '2010-01-01'\n",
    "END = '2017-12-31'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:53:33.998338Z",
     "start_time": "2020-06-19T23:53:33.991832Z"
    }
   },
   "outputs": [],
   "source": [
    "sns.set_style('darkgrid')\n",
    "idx = pd.IndexSlice"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:53:34.008711Z",
     "start_time": "2020-06-19T23:53:34.002005Z"
    }
   },
   "outputs": [],
   "source": [
    "percentiles = [.001, .01, .02, .03, .04, .05]\n",
    "percentiles += [1-p for p in percentiles[::-1]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:53:34.018715Z",
     "start_time": "2020-06-19T23:53:34.011125Z"
    }
   },
   "outputs": [],
   "source": [
    "T = [1, 5, 10, 21, 42, 63]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Loading Quandl Wiki Stock Prices & Meta Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:53:37.728424Z",
     "start_time": "2020-06-19T23:53:34.021080Z"
    }
   },
   "outputs": [],
   "source": [
    "DATA_STORE = '../data/assets.h5'\n",
    "ohlcv = ['adj_open', 'adj_close', 'adj_low', 'adj_high', 'adj_volume']\n",
    "with pd.HDFStore(DATA_STORE) as store:\n",
    "    prices = (store['quandl/wiki/prices']\n",
    "              .loc[idx[START:END, :], ohlcv]\n",
    "              .rename(columns=lambda x: x.replace('adj_', ''))\n",
    "              .swaplevel()\n",
    "              .sort_index())\n",
    "    metadata = (store['us_equities/stocks'].loc[:, ['marketcap', 'sector']])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:53:37.763381Z",
     "start_time": "2020-06-19T23:53:37.729411Z"
    }
   },
   "outputs": [],
   "source": [
    "prices.volume /= 1e3\n",
    "prices.index.names = ['symbol', 'date']\n",
    "metadata.index.name = 'symbol'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Remove stocks with insufficient observations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:54:30.270909Z",
     "start_time": "2020-06-19T23:53:37.764472Z"
    }
   },
   "outputs": [],
   "source": [
    "min_obs = 7 * YEAR\n",
    "nobs = prices.groupby(level='symbol').size()\n",
    "keep = nobs[nobs > min_obs].index\n",
    "prices = prices.loc[idx[keep, :], :]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Align price and meta data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:54:30.280818Z",
     "start_time": "2020-06-19T23:54:30.271836Z"
    }
   },
   "outputs": [],
   "source": [
    "metadata = metadata[~metadata.index.duplicated() & metadata.sector.notnull()]\n",
    "metadata.sector = metadata.sector.str.lower().str.replace(' ', '_')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:55:06.551172Z",
     "start_time": "2020-06-19T23:54:30.281795Z"
    }
   },
   "outputs": [],
   "source": [
    "shared = (prices.index.get_level_values('symbol').unique()\n",
    "          .intersection(metadata.index))\n",
    "metadata = metadata.loc[shared, :]\n",
    "prices = prices.loc[idx[shared, :], :]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Limit universe to 1,000 stocks with highest market cap"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:55:17.253761Z",
     "start_time": "2020-06-19T23:55:06.552815Z"
    }
   },
   "outputs": [],
   "source": [
    "universe = metadata.marketcap.nlargest(1000).index\n",
    "prices = prices.loc[idx[universe, :], :]\n",
    "metadata = metadata.loc[universe]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:55:17.261899Z",
     "start_time": "2020-06-19T23:55:17.254921Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "consumer_services        187\n",
       "finance                  168\n",
       "technology               116\n",
       "health_care              103\n",
       "capital_goods             94\n",
       "basic_industries          67\n",
       "public_utilities          66\n",
       "consumer_non-durables     61\n",
       "energy                    51\n",
       "consumer_durables         36\n",
       "miscellaneous             28\n",
       "transportation            23\n",
       "Name: sector, dtype: int64"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "metadata.sector.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:55:17.321209Z",
     "start_time": "2020-06-19T23:55:17.263046Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "MultiIndex: 2004775 entries, ('A', Timestamp('2010-01-04 00:00:00')) to ('ZION', Timestamp('2017-12-29 00:00:00'))\n",
      "Data columns (total 5 columns):\n",
      " #   Column  Non-Null Count    Dtype  \n",
      "---  ------  --------------    -----  \n",
      " 0   open    2004775 non-null  float64\n",
      " 1   close   2004775 non-null  float64\n",
      " 2   low     2004775 non-null  float64\n",
      " 3   high    2004775 non-null  float64\n",
      " 4   volume  2004775 non-null  float64\n",
      "dtypes: float64(5)\n",
      "memory usage: 84.3+ MB\n"
     ]
    }
   ],
   "source": [
    "prices.info(null_counts=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:55:17.326097Z",
     "start_time": "2020-06-19T23:55:17.322086Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Index: 1000 entries, AAPL to NTCT\n",
      "Data columns (total 2 columns):\n",
      " #   Column     Non-Null Count  Dtype  \n",
      "---  ------     --------------  -----  \n",
      " 0   marketcap  1000 non-null   float64\n",
      " 1   sector     1000 non-null   object \n",
      "dtypes: float64(1), object(1)\n",
      "memory usage: 23.4+ KB\n"
     ]
    }
   ],
   "source": [
    "metadata.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Compute Rolling Average Dollar Volume"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:55:18.220654Z",
     "start_time": "2020-06-19T23:55:17.327352Z"
    }
   },
   "outputs": [],
   "source": [
    "# compute dollar volume to determine universe\n",
    "prices['dollar_vol'] = prices.loc[:, 'close'].mul(prices.loc[:, 'volume'], axis=0).div(1e3)\n",
    "prices['dollar_vol'] = (prices\n",
    "                        .groupby('symbol',\n",
    "                                 group_keys=False,\n",
    "                                 as_index=False)\n",
    "                        .dollar_vol\n",
    "                        .rolling(window=21)\n",
    "                        .mean()\n",
    "                        .fillna(0)\n",
    "                        .reset_index(level=0, drop=True))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:55:18.658810Z",
     "start_time": "2020-06-19T23:55:18.221606Z"
    }
   },
   "outputs": [],
   "source": [
    "prices['dollar_vol_rank'] = (prices\n",
    "                             .groupby('date')\n",
    "                             .dollar_vol\n",
    "                             .rank(ascending=False))\n",
    "prices = prices.drop('dollar_vol', axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Add some Basic Factors"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Compute the Relative Strength Index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:55:19.357377Z",
     "start_time": "2020-06-19T23:55:18.659767Z"
    }
   },
   "outputs": [],
   "source": [
    "prices['rsi'] = prices.groupby(level='symbol').close.apply(RSI)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:55:20.147252Z",
     "start_time": "2020-06-19T23:55:19.358265Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax = sns.distplot(prices.rsi.dropna())\n",
    "ax.axvline(30, ls='--', lw=1, c='k')\n",
    "ax.axvline(70, ls='--', lw=1, c='k')\n",
    "ax.set_title('RSI Distribution with Signal Threshold')\n",
    "sns.despine()\n",
    "plt.tight_layout();"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Compute Bollinger Bands"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:55:20.150316Z",
     "start_time": "2020-06-19T23:55:20.148136Z"
    }
   },
   "outputs": [],
   "source": [
    "def compute_bb(close):\n",
    "    high, mid, low = BBANDS(close, timeperiod=20)\n",
    "    return pd.DataFrame({'bb_high': high, 'bb_low': low}, index=close.index)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:55:21.501185Z",
     "start_time": "2020-06-19T23:55:20.151236Z"
    }
   },
   "outputs": [],
   "source": [
    "prices = (prices.join(prices\n",
    "                      .groupby(level='symbol')\n",
    "                      .close\n",
    "                      .apply(compute_bb)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:55:21.544427Z",
     "start_time": "2020-06-19T23:55:21.502044Z"
    }
   },
   "outputs": [],
   "source": [
    "prices['bb_high'] = prices.bb_high.sub(prices.close).div(prices.bb_high).apply(np.log1p)\n",
    "prices['bb_low'] = prices.close.sub(prices.bb_low).div(prices.close).apply(np.log1p)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:55:22.212231Z",
     "start_time": "2020-06-19T23:55:21.545314Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axes = plt.subplots(ncols=2, figsize=(15, 5))\n",
    "sns.distplot(prices.loc[prices.dollar_vol_rank<100, 'bb_low'].dropna(), ax=axes[0])\n",
    "sns.distplot(prices.loc[prices.dollar_vol_rank<100, 'bb_high'].dropna(), ax=axes[1])\n",
    "sns.despine()\n",
    "plt.tight_layout();"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Compute Average True Range"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:55:22.989155Z",
     "start_time": "2020-06-19T23:55:22.213161Z"
    }
   },
   "outputs": [],
   "source": [
    "prices['NATR'] = prices.groupby(level='symbol', \n",
    "                                group_keys=False).apply(lambda x: \n",
    "                                                        talib.NATR(x.high, x.low, x.close))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:55:22.992306Z",
     "start_time": "2020-06-19T23:55:22.990019Z"
    }
   },
   "outputs": [],
   "source": [
    "def compute_atr(stock_data):\n",
    "    df = ATR(stock_data.high, stock_data.low, \n",
    "             stock_data.close, timeperiod=14)\n",
    "    return df.sub(df.mean()).div(df.std())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:55:24.233346Z",
     "start_time": "2020-06-19T23:55:22.993235Z"
    }
   },
   "outputs": [],
   "source": [
    "prices['ATR'] = (prices.groupby('symbol', group_keys=False)\n",
    "                 .apply(compute_atr))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Compute Moving Average Convergence/Divergence"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:55:24.956498Z",
     "start_time": "2020-06-19T23:55:24.234232Z"
    }
   },
   "outputs": [],
   "source": [
    "prices['PPO'] = prices.groupby(level='symbol').close.apply(talib.PPO)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:55:24.959596Z",
     "start_time": "2020-06-19T23:55:24.957424Z"
    }
   },
   "outputs": [],
   "source": [
    "def compute_macd(close):\n",
    "    macd = MACD(close)[0]\n",
    "    return (macd - np.mean(macd))/np.std(macd)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:55:26.214881Z",
     "start_time": "2020-06-19T23:55:24.960708Z"
    }
   },
   "outputs": [],
   "source": [
    "prices['MACD'] = (prices\n",
    "                  .groupby('symbol', group_keys=False)\n",
    "                  .close\n",
    "                  .apply(compute_macd))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Combine Price and Meta Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:55:26.269940Z",
     "start_time": "2020-06-19T23:55:26.215809Z"
    }
   },
   "outputs": [],
   "source": [
    "metadata.sector = pd.factorize(metadata.sector)[0].astype(int)\n",
    "prices = prices.join(metadata[['sector']])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Compute Returns"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Historical Returns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:55:26.748222Z",
     "start_time": "2020-06-19T23:55:26.270784Z"
    }
   },
   "outputs": [],
   "source": [
    "by_sym = prices.groupby(level='symbol').close\n",
    "for t in T:\n",
    "    prices[f'r{t:02}'] = by_sym.pct_change(t)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Daily historical return quantiles"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:55:51.852433Z",
     "start_time": "2020-06-19T23:55:26.749171Z"
    }
   },
   "outputs": [],
   "source": [
    "for t in T:\n",
    "    prices[f'r{t:02}dec'] = (prices[f'r{t:02}'].groupby(level='date')\n",
    "             .apply(lambda x: pd.qcut(x, q=10, labels=False, duplicates='drop')))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Daily sector return quantiles"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:57:51.523015Z",
     "start_time": "2020-06-19T23:55:51.853449Z"
    },
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "for t in T:\n",
    "    prices[f'r{t:02}q_sector'] = (prices.groupby(\n",
    "        ['date', 'sector'])[f'r{t:02}'].transform(\n",
    "            lambda x: pd.qcut(x, q=5, labels=False, duplicates='drop')))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Compute Forward Returns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:57:52.298280Z",
     "start_time": "2020-06-19T23:57:51.523875Z"
    }
   },
   "outputs": [],
   "source": [
    "for t in [1, 5, 21]:\n",
    "    prices[f'r{t:02}_fwd'] = prices.groupby(level='symbol')[f'r{t:02}'].shift(-t)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Remove outliers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:57:52.861428Z",
     "start_time": "2020-06-19T23:57:52.299200Z"
    }
   },
   "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>r01</th>\n",
       "      <th>r05</th>\n",
       "      <th>r10</th>\n",
       "      <th>r21</th>\n",
       "      <th>r42</th>\n",
       "      <th>r63</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>2.003775e+06</td>\n",
       "      <td>1.999775e+06</td>\n",
       "      <td>1.994775e+06</td>\n",
       "      <td>1.983775e+06</td>\n",
       "      <td>1.962775e+06</td>\n",
       "      <td>1.941775e+06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>7.519751e-04</td>\n",
       "      <td>3.726962e-03</td>\n",
       "      <td>7.353932e-03</td>\n",
       "      <td>1.555927e-02</td>\n",
       "      <td>3.113691e-02</td>\n",
       "      <td>4.619119e-02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>2.166262e-02</td>\n",
       "      <td>4.791746e-02</td>\n",
       "      <td>6.579895e-02</td>\n",
       "      <td>9.467552e-02</td>\n",
       "      <td>1.325751e-01</td>\n",
       "      <td>1.618423e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>-8.757416e-01</td>\n",
       "      <td>-8.768476e-01</td>\n",
       "      <td>-8.778415e-01</td>\n",
       "      <td>-8.802285e-01</td>\n",
       "      <td>-8.867366e-01</td>\n",
       "      <td>-8.863481e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>-8.088407e-03</td>\n",
       "      <td>-1.721664e-02</td>\n",
       "      <td>-2.291896e-02</td>\n",
       "      <td>-3.045918e-02</td>\n",
       "      <td>-3.531712e-02</td>\n",
       "      <td>-3.696833e-02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>6.561680e-04</td>\n",
       "      <td>3.702235e-03</td>\n",
       "      <td>7.173181e-03</td>\n",
       "      <td>1.503253e-02</td>\n",
       "      <td>2.899023e-02</td>\n",
       "      <td>4.217809e-02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>9.509191e-03</td>\n",
       "      <td>2.440601e-02</td>\n",
       "      <td>3.707177e-02</td>\n",
       "      <td>5.927618e-02</td>\n",
       "      <td>9.305628e-02</td>\n",
       "      <td>1.219666e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1.216425e+01</td>\n",
       "      <td>1.252657e+01</td>\n",
       "      <td>1.252657e+01</td>\n",
       "      <td>1.252657e+01</td>\n",
       "      <td>1.181643e+01</td>\n",
       "      <td>1.166968e+01</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                r01           r05           r10           r21           r42  \\\n",
       "count  2.003775e+06  1.999775e+06  1.994775e+06  1.983775e+06  1.962775e+06   \n",
       "mean   7.519751e-04  3.726962e-03  7.353932e-03  1.555927e-02  3.113691e-02   \n",
       "std    2.166262e-02  4.791746e-02  6.579895e-02  9.467552e-02  1.325751e-01   \n",
       "min   -8.757416e-01 -8.768476e-01 -8.778415e-01 -8.802285e-01 -8.867366e-01   \n",
       "25%   -8.088407e-03 -1.721664e-02 -2.291896e-02 -3.045918e-02 -3.531712e-02   \n",
       "50%    6.561680e-04  3.702235e-03  7.173181e-03  1.503253e-02  2.899023e-02   \n",
       "75%    9.509191e-03  2.440601e-02  3.707177e-02  5.927618e-02  9.305628e-02   \n",
       "max    1.216425e+01  1.252657e+01  1.252657e+01  1.252657e+01  1.181643e+01   \n",
       "\n",
       "                r63  \n",
       "count  1.941775e+06  \n",
       "mean   4.619119e-02  \n",
       "std    1.618423e-01  \n",
       "min   -8.863481e-01  \n",
       "25%   -3.696833e-02  \n",
       "50%    4.217809e-02  \n",
       "75%    1.219666e-01  \n",
       "max    1.166968e+01  "
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "prices[[f'r{t:02}' for t in T]].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:57:52.868952Z",
     "start_time": "2020-06-19T23:57:52.864279Z"
    }
   },
   "outputs": [],
   "source": [
    "outliers = prices[prices.r01>1].index.get_level_values('symbol').unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:57:59.365760Z",
     "start_time": "2020-06-19T23:57:52.870028Z"
    }
   },
   "outputs": [],
   "source": [
    "prices = prices.drop(outliers, level='symbol')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Create time and sector dummy variables"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:57:59.726663Z",
     "start_time": "2020-06-19T23:57:59.366967Z"
    }
   },
   "outputs": [],
   "source": [
    "prices['year'] = prices.index.get_level_values('date').year\n",
    "prices['month'] = prices.index.get_level_values('date').month\n",
    "prices['weekday'] = prices.index.get_level_values('date').weekday"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Store Model Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:57:59.884344Z",
     "start_time": "2020-06-19T23:57:59.727505Z"
    },
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "MultiIndex: 1994931 entries, ('A', Timestamp('2010-01-04 00:00:00')) to ('ZION', Timestamp('2017-12-29 00:00:00'))\n",
      "Data columns (total 38 columns):\n",
      " #   Column           Non-Null Count    Dtype  \n",
      "---  ------           --------------    -----  \n",
      " 0   open             1994931 non-null  float64\n",
      " 1   close            1994931 non-null  float64\n",
      " 2   low              1994931 non-null  float64\n",
      " 3   high             1994931 non-null  float64\n",
      " 4   volume           1994931 non-null  float64\n",
      " 5   dollar_vol_rank  1994931 non-null  float64\n",
      " 6   rsi              1981001 non-null  float64\n",
      " 7   bb_high          1976026 non-null  float64\n",
      " 8   bb_low           1976022 non-null  float64\n",
      " 9   NATR             1981001 non-null  float64\n",
      " 10  ATR              1981001 non-null  float64\n",
      " 11  PPO              1970056 non-null  float64\n",
      " 12  MACD             1962096 non-null  float64\n",
      " 13  sector           1994931 non-null  int64  \n",
      " 14  r01              1993936 non-null  float64\n",
      " 15  r05              1989956 non-null  float64\n",
      " 16  r10              1984981 non-null  float64\n",
      " 17  r21              1974036 non-null  float64\n",
      " 18  r42              1953141 non-null  float64\n",
      " 19  r63              1932246 non-null  float64\n",
      " 20  r01dec           1993933 non-null  float64\n",
      " 21  r05dec           1989956 non-null  float64\n",
      " 22  r10dec           1984981 non-null  float64\n",
      " 23  r21dec           1974036 non-null  float64\n",
      " 24  r42dec           1953141 non-null  float64\n",
      " 25  r63dec           1932246 non-null  float64\n",
      " 26  r01q_sector      1993933 non-null  float64\n",
      " 27  r05q_sector      1989956 non-null  float64\n",
      " 28  r10q_sector      1984981 non-null  float64\n",
      " 29  r21q_sector      1974036 non-null  float64\n",
      " 30  r42q_sector      1953141 non-null  float64\n",
      " 31  r63q_sector      1932246 non-null  float64\n",
      " 32  r01_fwd          1993936 non-null  float64\n",
      " 33  r05_fwd          1989956 non-null  float64\n",
      " 34  r21_fwd          1974036 non-null  float64\n",
      " 35  year             1994931 non-null  int64  \n",
      " 36  month            1994931 non-null  int64  \n",
      " 37  weekday          1994931 non-null  int64  \n",
      "dtypes: float64(34), int64(4)\n",
      "memory usage: 586.1+ MB\n"
     ]
    }
   ],
   "source": [
    "prices.info(null_counts=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-06-19T23:58:00.258966Z",
     "start_time": "2020-06-19T23:57:59.885847Z"
    }
   },
   "outputs": [],
   "source": [
    "prices.drop(['open', 'close', 'low', 'high', 'volume'], axis=1).to_hdf('data.h5', 'model_data')"
   ]
  }
 ],
 "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": "3.7.7"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": true,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": true
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}