SP1500stockPicker.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# S&P1500 STOCK PICKER\n",
    "#### AUTHOR : THOMAS ROCHEFORT-BEAUDOIN\n",
    "##### This project is licensed under the GNU General Public License v3.0 - see the LICENSE.md file for details\n",
    "---\n",
    "\n",
    "### Goal:\n",
    "The goal of this python script will be to select stocks from the S&P1500 constituents in order to beat the S&P500 index on a monthly basis. To do so we will construct a machine learning algorithm which will train on the financial ratios of the companies in the S&P1500.\n",
    "\n",
    "### Data used:\n",
    "4 datasets are used:\n",
    "- \" ratios_1990_2019.csv \" which consists of all the financial ratios of the Standard & Poor database from 1990 to 2019. The data is presented with the publication date of the various financial documents from which they are derived.\n",
    "\n",
    "- \" yield_1962_2019.csv \" which consists of all the monthly yields of the companies in the Standard & Poor database from 1962 to 2019.\n",
    "\n",
    "- \" SP1500constituents.csv \" which lists all the comapnies who are/were in the SP1500 index.\n",
    "\n",
    "- \" ^GSPC.csv \" which holds the monthly closing price of the SP500 index.\n",
    " \n",
    " \n",
    " ### Methodology:\n",
    " 1. The various datasets will first be merged into one main dataframe before the \"Data Cleaning\" steps:\n",
    "     1. The monthly yields are presented with a date corresponding to the last day of the trading month. The ratios are presented with their respective publication date. We therefore need to match the most recent publication date that was available at the end of the trading month for every company in the monthly yields file.\n",
    "     2. We need to generate the monthly return of the SP500 from the adjusted closing price and match with the monthly returns of the companies. We can then generate the column \" WIN \" which will hold a value of 0 if the monthly return of the stock is smaller than the monthly return of the SP500\n",
    "     3. We need to get rid of the companies which were not in the SP1500 index at every month.\n",
    "     4. We will save the generated dataframe as .pkl\n",
    "     \n",
    "2.  Data cleaning\n",
    "3.  Feature Selection\n",
    "4.  Training the Machine Learning algorithm and backtesting it\n",
    "\n",
    "### Feature Selection:\n",
    "\n",
    "The feature selection will be accomplished using the excellent feature_selector package from Will Koehrsen which you can find information at https://github.com/WillKoehrsen/feature-selector\n",
    "\n",
    " ### Machine Learning Algorithm:\n",
    " \n",
    " We will use a Random Forest regressor from SKlearn to try to predict the binary value of the \" WIN \" column. In other words:  We want to predict every month if a stock will beat the SP500 or not. The algorithm will train itself on a 7 year roling window period and then predict the following month. A rolling window keeps the algorithm dynamic (meaning its thinking will evolve over time) and the 7 year period corresponds to the historical average economic cycle.\n",
    " "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [],
   "source": [
    "## First we import the required packages:\n",
    "import numpy as np\n",
    "from datetime import timedelta\n",
    "from IPython.display import clear_output\n",
    "import time\n",
    "from feature_selector import FeatureSelector\n",
    "import pandas as pd\n",
    "import datetime\n",
    "import pickle\n",
    "import progressbar\n",
    "from sklearn.impute import SimpleImputer\n",
    "import dateutil.relativedelta\n",
    "import multiprocessing as mp"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.7/site-packages/ipykernel_launcher.py:3: FutureWarning: convert_objects is deprecated.  To re-infer data dtypes for object columns, use DataFrame.infer_objects()\n",
      "For all other conversions use the data-type specific converters pd.to_datetime, pd.to_timedelta and pd.to_numeric.\n",
      "  This is separate from the ipykernel package so we can avoid doing imports until\n",
      "/usr/local/lib/python3.7/site-packages/ipykernel_launcher.py:7: FutureWarning: convert_objects is deprecated.  To re-infer data dtypes for object columns, use DataFrame.infer_objects()\n",
      "For all other conversions use the data-type specific converters pd.to_datetime, pd.to_timedelta and pd.to_numeric.\n",
      "  import sys\n"
     ]
    }
   ],
   "source": [
    "## We then load the required yield and financial ratios database:\n",
    "ratio= pd.read_csv('ratios_1990_2019.csv')\n",
    "ratio = ratio.convert_objects(convert_numeric=True)\n",
    "yields=pd.read_csv('yield_1962_2019.csv')\n",
    "\n",
    "ticker=yields.tic ## Save for later since convert_numeric will break the strings\n",
    "yields = yields.convert_objects(convert_numeric=True)\n",
    "yields.tic=ticker\n",
    "\n",
    "# We wanna use the datetime format to facilitate the next steps so let's change it right now:\n",
    "ratio.public_date=pd.to_datetime(ratio.public_date)\n",
    "yields.datadate=pd.to_datetime(yields.datadate)\n",
    "\n",
    "#remove non-common shares from database\n",
    "yields = yields[yields.tpci == 0]  \n",
    "\n",
    "## We will want to use only the constituents from the S&P1500\n",
    "## The selection will be in the Data Cleaning section below\n",
    "SP1500constituents=pd.read_csv('SP1500constituents.csv')\n",
    "SP1500constituents['from_date']=pd.to_datetime(SP1500constituents['from_date'])\n",
    "SP1500constituents['thru_date']=pd.to_datetime(SP1500constituents['thru_date'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "## To alleviate the memory needed we can keep only the data from January 1995:\n",
    "d = datetime.datetime.strptime(\"1995-01-01\", \"%Y-%m-%d\")\n",
    "d2= datetime.datetime.strptime(\"2019-08-31\", \"%Y-%m-%d\")\n",
    "ratio= ratio[ratio[\"public_date\"].isin(pd.date_range(d, d2))]\n",
    "yields= yields[yields[\"datadate\"].isin(pd.date_range(d, d2))]\n",
    "\n",
    "# Let's reset the index:\n",
    "ratio=ratio.reset_index(drop=True)\n",
    "yields=yields.reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Combining the datasets:\n",
    "Now we want to combine both datasets on the columns \"gvkey\" and the date column.\n",
    "For the financial ratios, the column \"public_date\" represents the date of publication of the financial report.\n",
    "For the yields, the column ('datadate') represents the last day of the month where the return in the column \"trt1m\" is observed. We will use multiprocessing to iterate across the database and index the right ratios to the right date."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "dict_new_entries = {}    # We are using a dictionary to append the list of rows                                                                                                                                                       \n",
    "main=pd.DataFrame()      # Resulkting dataFrame converted from  dict_new_entries                                                                                                                                                      \n",
    "\n",
    "yields_date = yields['datadate'] # Extract the columns we need. Much quicker for search/iterating                                                                                                                                       \n",
    "yields_gvkey  = yields['gvkey']\n",
    "ratio_gvkey     = ratio['gvkey']\n",
    "\n",
    "number_columns = ratio.shape[1]           # Create empty row of the proper size. Will generate a row of NaNs                                                                                                                             \n",
    "empty_row = [None] * number_columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "def mergedb(key):\n",
    "    a = yields_date[key]\n",
    "    b = yields_gvkey[key]\n",
    "    rows_gvkey = ratio.loc[ratio['gvkey'] == b]                                                             ## Extract rows where 'stkcd' == b                                                                                           \n",
    "    rows_date = rows_gvkey.loc[rows_gvkey['public_date'] < a].sort_values(by='public_date', ascending=True )  ## Extract subrows where public_date < a. Ascending sorting                                                                          \n",
    "\n",
    "    if rows_date.size > 0:\n",
    "        row_to_add = rows_date.tail(1)              ## Keep most recent entry, located at the end of the sorted list rows_annodt                                                                                                    \n",
    "        dict_new_entries[key] = row_to_add.values[0]    ## Memorize this entry in dict                                                                                                                                                  \n",
    "\n",
    "    else:\n",
    "        dict_new_entries[key] = empty_row               ## Empty row. Will generate NaNs in every columns   \n",
    "    return dict_new_entries[key]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100% (3990930 of 3990930) |##############| Elapsed Time: 0:23:21 Time:  0:23:21\n"
     ]
    }
   ],
   "source": [
    "## Multiprocessing of the function mergedb\n",
    "nbrAvailCores = mp.cpu_count()\n",
    "pool = mp.Pool(processes=nbrAvailCores)\n",
    "resultsHandle = [pool.apply_async(mergedb, args=(z,)) for z in range(0, len(yields))]\n",
    "results = [r.get() for r in progressbar.progressbar(resultsHandle)]\n",
    "pool.close()\n",
    "\n",
    "main = pd.DataFrame(results, columns=ratio.columns)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "#We concatenate main and yields together\n",
    "complete=pd.concat([yields,main],axis=1)\n",
    "complete = complete[pd.notnull(complete['public_date'])]  ## Get rid of the NaN rows"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "complete.to_pickle(\"./database.pkl\")   ## We save the intermediary dataframe"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "complete = pd.read_pickle(\"./database.pkl\") ## This can serve as an intermediary step when running the notebook"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "SP500= pd.read_csv('^GSPC.csv',usecols=[\"Date\", \"Adj Close\"])\n",
    "SP500['sprtrn']= SP500['Adj Close'].pct_change()*100     ## Creates the column sprtrn with the monhtly SP500 return\n",
    "SP500= SP500[np.isfinite(SP500['sprtrn'])]\n",
    "SP500.Date=pd.to_datetime(SP500.Date)\n",
    "complete = pd.merge(complete.assign(grouper=complete['datadate'].dt.to_period('M')),\n",
    "               SP500.assign(grouper=SP500['Date'].dt.to_period('M')),\n",
    "               how='left', on='grouper')\n",
    "\n",
    "complete=complete.drop(columns=['grouper', 'Date','Adj Close'])\n",
    "complete['win'] = np.where((complete.trt1m > complete.sprtrn) \n",
    "                     , 1, 0)\n",
    "complete['decision_date']=complete.datadate+datetime.timedelta(1)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Cleaning the dataset\n",
    "---"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "complete = complete.loc[:,~complete.columns.duplicated()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "db = complete\n",
    "SP1500=SP1500constituents\n",
    "number_rows = complete.shape[0]\n",
    "empty_row = [0] * number_rows\n",
    "gvkeySP1500=SP1500['gvkey']\n",
    "gvkey=db['gvkey']\n",
    "date=db['datadate']\n",
    "\n",
    "## Defining a function to keep only the constituent of the SP1500 each months. Will mark a 1 next to securities part of the index.\n",
    "def findSP1500(key):\n",
    "    if np.sum(gvkeySP1500==gvkey[key])>0:\n",
    "        rows=SP1500.loc[SP1500['gvkey'] == gvkey[key]]\n",
    "        if (rows.from_date.iloc[0]<=date[key]<rows.thru_date.iloc[0]) & len(rows):\n",
    "            empty_row[key]=1     \n",
    "        else:\n",
    "            empty_row[key]=0\n",
    "    else:\n",
    "        empty_row[key]=0\n",
    "    return (empty_row[key])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100% (1829047 of 1829047) |##############| Elapsed Time: 0:01:35 Time:  0:01:35\n"
     ]
    }
   ],
   "source": [
    "# Multiprocessing of the function findSP1500\n",
    "nbrAvailCores = mp.cpu_count()\n",
    "pool = mp.Pool(processes=nbrAvailCores)\n",
    "resultsHandle = [pool.apply_async(findSP1500, args=(z,)) for z in range(0, len(complete))]\n",
    "results = [r.get() for r in progressbar.progressbar(resultsHandle)]\n",
    "pool.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "complete['SP1500']=results\n",
    "complete=complete[complete.SP1500==1] #remove non-SP1500 securities"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "gvkey                   0\n",
       "iid                  1359\n",
       "datadate                0\n",
       "tic                     0\n",
       "conm                    0\n",
       "trt1m                5141\n",
       "exchg                   0\n",
       "tpci                    0\n",
       "adate                 441\n",
       "qdate                   0\n",
       "public_date             0\n",
       "CAPEI                2332\n",
       "bm                   4764\n",
       "evm                  1942\n",
       "pe_op_basic          9984\n",
       "pe_op_dil           74009\n",
       "pe_exi              10594\n",
       "pe_inc              10261\n",
       "ps                    475\n",
       "pcf                   940\n",
       "dpr                 43310\n",
       "npm                   475\n",
       "opmbd                 475\n",
       "opmad                 482\n",
       "gpm                   490\n",
       "ptpm                  475\n",
       "cfm                  2909\n",
       "roa                   967\n",
       "roe                  5294\n",
       "roce                 3336\n",
       "                    ...  \n",
       "dltt_be              5698\n",
       "debt_assets           532\n",
       "debt_capital         5328\n",
       "de_ratio              535\n",
       "intcov              63530\n",
       "intcov_ratio        63542\n",
       "cash_ratio          56652\n",
       "quick_ratio         56634\n",
       "curr_ratio          56635\n",
       "cash_conversion     77394\n",
       "inv_turn            69869\n",
       "at_turn              1082\n",
       "rect_turn            6965\n",
       "pay_turn            12908\n",
       "sale_invcap           656\n",
       "sale_equity          5105\n",
       "sale_nwc           103932\n",
       "rd_sale               280\n",
       "adv_sale              631\n",
       "staff_sale            631\n",
       "accrual               936\n",
       "ptb                  4764\n",
       "PEG_trailing       170492\n",
       "DIVYIELD           164557\n",
       "PEG_1yrforward      54752\n",
       "PEG_ltgforward      89439\n",
       "sprtrn                  0\n",
       "win                     0\n",
       "decision_date           0\n",
       "SP1500                  0\n",
       "Length: 86, dtype: int64"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "complete.isna().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "## We cannot trade stocks with no monthly yields\n",
    "clean_db1 = complete[np.isfinite(complete['trt1m'])]\n",
    "## We can drop data that is not useful for us\n",
    "clean_db=clean_db1.drop(columns=['iid','adate','qdate','public_date','tic','SP1500'])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "## We do not want to use items where more than 50% of the rows are missing. \n",
    "clean_db = clean_db.loc[:, clean_db.isnull().mean() < .5]\n",
    "clean_db['DIVYIELD'] = clean_db['DIVYIELD'].str.rstrip('%').astype('float') / 100.0 # convert DIVYIELD"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "## For the rest of the ratios, we will impute the missing values with the mean of the total column\n",
    "imp=SimpleImputer(missing_values=np.nan, strategy=\"mean\" )\n",
    "#database1['ptb'] = imp.fit_transform(database1[['ptb']])\n",
    "impute_columns=clean_db.columns[clean_db.isna().any()].tolist()\n",
    "for i in impute_columns:\n",
    "    clean_db[i] = imp.fit_transform(clean_db[[i]])\n",
    "clean_db['tic']=clean_db1.tic\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "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>gvkey</th>\n",
       "      <th>datadate</th>\n",
       "      <th>conm</th>\n",
       "      <th>trt1m</th>\n",
       "      <th>exchg</th>\n",
       "      <th>tpci</th>\n",
       "      <th>CAPEI</th>\n",
       "      <th>bm</th>\n",
       "      <th>evm</th>\n",
       "      <th>pe_op_basic</th>\n",
       "      <th>...</th>\n",
       "      <th>accrual</th>\n",
       "      <th>ptb</th>\n",
       "      <th>PEG_trailing</th>\n",
       "      <th>DIVYIELD</th>\n",
       "      <th>PEG_1yrforward</th>\n",
       "      <th>PEG_ltgforward</th>\n",
       "      <th>sprtrn</th>\n",
       "      <th>win</th>\n",
       "      <th>decision_date</th>\n",
       "      <th>tic</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1004</td>\n",
       "      <td>1995-02-28</td>\n",
       "      <td>AAR CORP</td>\n",
       "      <td>1.8519</td>\n",
       "      <td>11.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>18.514</td>\n",
       "      <td>1.068</td>\n",
       "      <td>10.849</td>\n",
       "      <td>16.364</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.062</td>\n",
       "      <td>0.963</td>\n",
       "      <td>0.046</td>\n",
       "      <td>0.0356</td>\n",
       "      <td>1.396</td>\n",
       "      <td>1.591086</td>\n",
       "      <td>3.607415</td>\n",
       "      <td>0</td>\n",
       "      <td>1995-03-01</td>\n",
       "      <td>AIR</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1004</td>\n",
       "      <td>1995-03-31</td>\n",
       "      <td>AAR CORP</td>\n",
       "      <td>-11.8182</td>\n",
       "      <td>11.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>18.857</td>\n",
       "      <td>1.068</td>\n",
       "      <td>10.849</td>\n",
       "      <td>16.667</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.062</td>\n",
       "      <td>0.981</td>\n",
       "      <td>0.047</td>\n",
       "      <td>0.0349</td>\n",
       "      <td>1.421</td>\n",
       "      <td>2.273000</td>\n",
       "      <td>2.732919</td>\n",
       "      <td>0</td>\n",
       "      <td>1995-04-01</td>\n",
       "      <td>AIR</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1004</td>\n",
       "      <td>1995-04-30</td>\n",
       "      <td>AAR CORP</td>\n",
       "      <td>14.3918</td>\n",
       "      <td>11.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>16.628</td>\n",
       "      <td>1.068</td>\n",
       "      <td>10.849</td>\n",
       "      <td>14.697</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.062</td>\n",
       "      <td>0.865</td>\n",
       "      <td>0.041</td>\n",
       "      <td>0.0396</td>\n",
       "      <td>1.253</td>\n",
       "      <td>2.004000</td>\n",
       "      <td>2.796036</td>\n",
       "      <td>1</td>\n",
       "      <td>1995-05-01</td>\n",
       "      <td>AIR</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1004</td>\n",
       "      <td>1995-05-31</td>\n",
       "      <td>AAR CORP</td>\n",
       "      <td>10.9091</td>\n",
       "      <td>11.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>20.453</td>\n",
       "      <td>1.032</td>\n",
       "      <td>10.145</td>\n",
       "      <td>15.537</td>\n",
       "      <td>...</td>\n",
       "      <td>0.006</td>\n",
       "      <td>0.969</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.0349</td>\n",
       "      <td>1.325</td>\n",
       "      <td>2.119000</td>\n",
       "      <td>3.631171</td>\n",
       "      <td>1</td>\n",
       "      <td>1995-06-01</td>\n",
       "      <td>AIR</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1004</td>\n",
       "      <td>1995-06-30</td>\n",
       "      <td>AAR CORP</td>\n",
       "      <td>17.2131</td>\n",
       "      <td>11.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>22.684</td>\n",
       "      <td>1.032</td>\n",
       "      <td>10.145</td>\n",
       "      <td>17.232</td>\n",
       "      <td>...</td>\n",
       "      <td>0.006</td>\n",
       "      <td>1.074</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.0315</td>\n",
       "      <td>1.713</td>\n",
       "      <td>2.350000</td>\n",
       "      <td>2.127854</td>\n",
       "      <td>1</td>\n",
       "      <td>1995-07-01</td>\n",
       "      <td>AIR</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 81 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   gvkey   datadate      conm    trt1m  exchg  tpci   CAPEI     bm     evm  \\\n",
       "0   1004 1995-02-28  AAR CORP   1.8519   11.0   0.0  18.514  1.068  10.849   \n",
       "1   1004 1995-03-31  AAR CORP -11.8182   11.0   0.0  18.857  1.068  10.849   \n",
       "2   1004 1995-04-30  AAR CORP  14.3918   11.0   0.0  16.628  1.068  10.849   \n",
       "3   1004 1995-05-31  AAR CORP  10.9091   11.0   0.0  20.453  1.032  10.145   \n",
       "4   1004 1995-06-30  AAR CORP  17.2131   11.0   0.0  22.684  1.032  10.145   \n",
       "\n",
       "   pe_op_basic  ...  accrual    ptb  PEG_trailing  DIVYIELD  PEG_1yrforward  \\\n",
       "0       16.364  ...   -0.062  0.963         0.046    0.0356           1.396   \n",
       "1       16.667  ...   -0.062  0.981         0.047    0.0349           1.421   \n",
       "2       14.697  ...   -0.062  0.865         0.041    0.0396           1.253   \n",
       "3       15.537  ...    0.006  0.969         0.000    0.0349           1.325   \n",
       "4       17.232  ...    0.006  1.074         0.000    0.0315           1.713   \n",
       "\n",
       "   PEG_ltgforward    sprtrn  win  decision_date  tic  \n",
       "0        1.591086  3.607415    0     1995-03-01  AIR  \n",
       "1        2.273000  2.732919    0     1995-04-01  AIR  \n",
       "2        2.004000  2.796036    1     1995-05-01  AIR  \n",
       "3        2.119000  3.631171    1     1995-06-01  AIR  \n",
       "4        2.350000  2.127854    1     1995-07-01  AIR  \n",
       "\n",
       "[5 rows x 81 columns]"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "## We can now visualize the cleaned DB:\n",
    "clean_db.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature Selection"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "train = clean_db.drop(columns = ['win','conm','datadate','gvkey','trt1m','decision_date','tic'])\n",
    "train_labels = clean_db['win']\n",
    "fs = FeatureSelector(data = train, labels = train_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5 features with a correlation magnitude greater than 0.97.\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fs.identify_collinear(correlation_threshold=0.975)\n",
    "fs.plot_collinear(plot_all=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training Gradient Boosting Model\n",
      "\n",
      "Training until validation scores don't improve for 100 rounds.\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[1000]\tvalid_0's auc: 0.643037\tvalid_0's l2: 0.234656\n",
      "Training until validation scores don't improve for 100 rounds.\n",
      "Early stopping, best iteration is:\n",
      "[887]\tvalid_0's auc: 0.635882\tvalid_0's l2: 0.236009\n",
      "Training until validation scores don't improve for 100 rounds.\n",
      "Early stopping, best iteration is:\n",
      "[882]\tvalid_0's auc: 0.643785\tvalid_0's l2: 0.234676\n",
      "Training until validation scores don't improve for 100 rounds.\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[996]\tvalid_0's auc: 0.643935\tvalid_0's l2: 0.234576\n",
      "Training until validation scores don't improve for 100 rounds.\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[999]\tvalid_0's auc: 0.63893\tvalid_0's l2: 0.235618\n",
      "Training until validation scores don't improve for 100 rounds.\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[998]\tvalid_0's auc: 0.640571\tvalid_0's l2: 0.23493\n",
      "Training until validation scores don't improve for 100 rounds.\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[992]\tvalid_0's auc: 0.643529\tvalid_0's l2: 0.234733\n",
      "Training until validation scores don't improve for 100 rounds.\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[972]\tvalid_0's auc: 0.642815\tvalid_0's l2: 0.234678\n",
      "Training until validation scores don't improve for 100 rounds.\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[1000]\tvalid_0's auc: 0.643294\tvalid_0's l2: 0.234586\n",
      "Training until validation scores don't improve for 100 rounds.\n",
      "Did not meet early stopping. Best iteration is:\n",
      "[998]\tvalid_0's auc: 0.635904\tvalid_0's l2: 0.235995\n",
      "\n",
      "1 features with zero importance after one-hot encoding.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "fs.identify_zero_importance(task = 'regression', eval_metric = 'auc', \n",
    "                            n_iterations = 10, early_stopping = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "70 features required for cumulative importance of 0.99 after one hot encoding.\n",
      "4 features do not contribute to cumulative importance of 0.99.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "fs.identify_low_importance(cumulative_importance = 0.99)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total of 8 features identified for removal\n",
      "['cfm', 'aftret_equity', 'debt_ebitda', 'ptpm', 'tpci', 'opmad', 'capital_ratio', 'debt_invcap']\n"
     ]
    }
   ],
   "source": [
    "all_to_remove = fs.check_removal()\n",
    "print(all_to_remove)\n",
    "clean_removed_db = clean_db.drop(columns = all_to_remove)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "clean_removed_db.drop_duplicates(keep=False,inplace=True) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### TRADING ALGORITHM\n",
    "---"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "def rfstockpicker_backtest(database,prediction_date,rolling_window,\n",
    "                           nb_months,n_estimators,n_jobs):\n",
    "    database = database\n",
    "    results={}\n",
    "    pd.options.mode.chained_assignment = None # Gets rid of a warning\n",
    "    for i in range(0,nb_months):\n",
    "        d = datetime.datetime.strptime(prediction_date, \"%Y-%m-%d\")\n",
    "\n",
    "        d2 = d - dateutil.relativedelta.relativedelta(months=i)\n",
    "        d3 = d - dateutil.relativedelta.relativedelta(months=i+1)\n",
    "        d4 = d3 - dateutil.relativedelta.relativedelta(years=rolling_window)    ##Creating a rolling window interval\n",
    "        \n",
    "        test_set = database[database['decision_date'].isin(pd.date_range(d2, d2))]\n",
    "        training_set=database[database['decision_date'].isin(pd.date_range(d4, d3))]\n",
    "\n",
    "        # Labels are the values we want to predict\n",
    "        train_labels = np.array(training_set['win'])\n",
    "        # Remove the labels from the features\n",
    "        columns_to_drop=['win','sprtrn','gvkey','datadate','conm','trt1m','decision_date','tic']\n",
    "        train_features= training_set.drop(columns_to_drop, axis = 1)\n",
    "        # Saving feature names for later use\n",
    "        train_feature_list = list(train_features.columns)\n",
    "        # Convert to numpy array\n",
    "        train_features = np.array(train_features)\n",
    "        \n",
    "        test_labels = np.array(test_set['win'])\n",
    "        # Remove the labels from the test features\n",
    "        test_features= test_set.drop(columns_to_drop, axis = 1)\n",
    "        # Convert to numpy array\n",
    "        test_features = np.array(test_features)\n",
    "\n",
    "        # Import the model we are using\n",
    "        from sklearn.ensemble import RandomForestRegressor\n",
    "        # Instantiate model \n",
    "        rf = RandomForestRegressor(n_estimators = n_estimators, random_state = 42,n_jobs=n_jobs)\n",
    "        # Train the model on training data\n",
    "        rf.fit(train_features, train_labels);\n",
    "\n",
    "        # Use the forest's predict method on the test data\n",
    "        stock_predictions = rf.predict(test_features)\n",
    "        result_set=test_set\n",
    "        result_set['predictions']=stock_predictions \n",
    "        results[i]=result_set\n",
    "\n",
    "       # Get numerical feature importances\n",
    "        importances = list(rf.feature_importances_)\n",
    "        # List of tuples with variable and importance\n",
    "        feature_importances = [(feature, round(importance, 2)) for feature, importance in zip(train_feature_list, importances)]\n",
    "        # Sort the feature importances by most important first\n",
    "        feature_importances = sorted(feature_importances, key = lambda x: x[1], reverse = True)\n",
    "        \n",
    "    return results,feature_importances,importances,train_feature_list\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "results,feature_imp,importances,train_feature_list=rfstockpicker_backtest(clean_removed_db,(\"2018-12-01\"),7,1,100,24)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "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>gvkey</th>\n",
       "      <th>datadate</th>\n",
       "      <th>conm</th>\n",
       "      <th>trt1m</th>\n",
       "      <th>exchg</th>\n",
       "      <th>CAPEI</th>\n",
       "      <th>bm</th>\n",
       "      <th>evm</th>\n",
       "      <th>pe_op_basic</th>\n",
       "      <th>pe_op_dil</th>\n",
       "      <th>...</th>\n",
       "      <th>ptb</th>\n",
       "      <th>PEG_trailing</th>\n",
       "      <th>DIVYIELD</th>\n",
       "      <th>PEG_1yrforward</th>\n",
       "      <th>PEG_ltgforward</th>\n",
       "      <th>sprtrn</th>\n",
       "      <th>win</th>\n",
       "      <th>decision_date</th>\n",
       "      <th>tic</th>\n",
       "      <th>predictions</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>570920</th>\n",
       "      <td>13714</td>\n",
       "      <td>2018-11-30</td>\n",
       "      <td>CBS CORP</td>\n",
       "      <td>-6.0785</td>\n",
       "      <td>11.0</td>\n",
       "      <td>12.998</td>\n",
       "      <td>0.124000</td>\n",
       "      <td>9.879</td>\n",
       "      <td>12.202</td>\n",
       "      <td>12.307</td>\n",
       "      <td>...</td>\n",
       "      <td>7.407000</td>\n",
       "      <td>1.048000</td>\n",
       "      <td>0.012600</td>\n",
       "      <td>0.486003</td>\n",
       "      <td>1.591086</td>\n",
       "      <td>1.785936</td>\n",
       "      <td>0</td>\n",
       "      <td>2018-12-01</td>\n",
       "      <td>CBS.A</td>\n",
       "      <td>0.98</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>571215</th>\n",
       "      <td>13714</td>\n",
       "      <td>2018-11-30</td>\n",
       "      <td>CBS CORP</td>\n",
       "      <td>-5.5275</td>\n",
       "      <td>11.0</td>\n",
       "      <td>12.998</td>\n",
       "      <td>0.124000</td>\n",
       "      <td>9.879</td>\n",
       "      <td>12.202</td>\n",
       "      <td>12.307</td>\n",
       "      <td>...</td>\n",
       "      <td>7.407000</td>\n",
       "      <td>1.048000</td>\n",
       "      <td>0.012600</td>\n",
       "      <td>0.486003</td>\n",
       "      <td>1.591086</td>\n",
       "      <td>1.785936</td>\n",
       "      <td>0</td>\n",
       "      <td>2018-12-01</td>\n",
       "      <td>CBS</td>\n",
       "      <td>0.98</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>697805</th>\n",
       "      <td>17934</td>\n",
       "      <td>2018-11-30</td>\n",
       "      <td>SPROUTS FARMERS MARKET</td>\n",
       "      <td>-14.3920</td>\n",
       "      <td>14.0</td>\n",
       "      <td>28.920</td>\n",
       "      <td>0.229000</td>\n",
       "      <td>10.180</td>\n",
       "      <td>21.862</td>\n",
       "      <td>22.223</td>\n",
       "      <td>...</td>\n",
       "      <td>5.313000</td>\n",
       "      <td>0.801000</td>\n",
       "      <td>0.022435</td>\n",
       "      <td>0.486003</td>\n",
       "      <td>1.591086</td>\n",
       "      <td>1.785936</td>\n",
       "      <td>0</td>\n",
       "      <td>2018-12-01</td>\n",
       "      <td>SFM</td>\n",
       "      <td>0.98</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1792940</th>\n",
       "      <td>180652</td>\n",
       "      <td>2018-11-30</td>\n",
       "      <td>VERISK ANALYTICS INC</td>\n",
       "      <td>2.9039</td>\n",
       "      <td>14.0</td>\n",
       "      <td>44.324</td>\n",
       "      <td>0.137000</td>\n",
       "      <td>17.132</td>\n",
       "      <td>37.333</td>\n",
       "      <td>38.166</td>\n",
       "      <td>...</td>\n",
       "      <td>8.141000</td>\n",
       "      <td>2.917000</td>\n",
       "      <td>0.022435</td>\n",
       "      <td>0.486003</td>\n",
       "      <td>1.591086</td>\n",
       "      <td>1.785936</td>\n",
       "      <td>1</td>\n",
       "      <td>2018-12-01</td>\n",
       "      <td>VRSK</td>\n",
       "      <td>0.98</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1155825</th>\n",
       "      <td>31843</td>\n",
       "      <td>2018-11-30</td>\n",
       "      <td>HARMONIC INC</td>\n",
       "      <td>1.9964</td>\n",
       "      <td>14.0</td>\n",
       "      <td>-12.288</td>\n",
       "      <td>0.608000</td>\n",
       "      <td>72.770</td>\n",
       "      <td>-11.479</td>\n",
       "      <td>-11.479</td>\n",
       "      <td>...</td>\n",
       "      <td>2.164000</td>\n",
       "      <td>-0.045000</td>\n",
       "      <td>0.022435</td>\n",
       "      <td>0.486003</td>\n",
       "      <td>1.591086</td>\n",
       "      <td>1.785936</td>\n",
       "      <td>1</td>\n",
       "      <td>2018-12-01</td>\n",
       "      <td>HLIT</td>\n",
       "      <td>0.97</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1735826</th>\n",
       "      <td>165993</td>\n",
       "      <td>2018-11-30</td>\n",
       "      <td>FIDELITY NATIONAL INFO SVCS</td>\n",
       "      <td>3.6984</td>\n",
       "      <td>11.0</td>\n",
       "      <td>48.746</td>\n",
       "      <td>0.339000</td>\n",
       "      <td>13.630</td>\n",
       "      <td>40.193</td>\n",
       "      <td>40.664</td>\n",
       "      <td>...</td>\n",
       "      <td>2.866000</td>\n",
       "      <td>0.756000</td>\n",
       "      <td>0.012300</td>\n",
       "      <td>0.486003</td>\n",
       "      <td>1.591086</td>\n",
       "      <td>1.785936</td>\n",
       "      <td>1</td>\n",
       "      <td>2018-12-01</td>\n",
       "      <td>FIS</td>\n",
       "      <td>0.97</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>156468</th>\n",
       "      <td>4503</td>\n",
       "      <td>2018-11-30</td>\n",
       "      <td>EXXON MOBIL CORP</td>\n",
       "      <td>0.8032</td>\n",
       "      <td>11.0</td>\n",
       "      <td>15.907</td>\n",
       "      <td>0.611000</td>\n",
       "      <td>10.946</td>\n",
       "      <td>22.701</td>\n",
       "      <td>22.701</td>\n",
       "      <td>...</td>\n",
       "      <td>1.576000</td>\n",
       "      <td>1.521000</td>\n",
       "      <td>0.041200</td>\n",
       "      <td>0.486003</td>\n",
       "      <td>1.591086</td>\n",
       "      <td>1.785936</td>\n",
       "      <td>0</td>\n",
       "      <td>2018-12-01</td>\n",
       "      <td>XOM</td>\n",
       "      <td>0.96</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>293188</th>\n",
       "      <td>7585</td>\n",
       "      <td>2018-11-30</td>\n",
       "      <td>MOTOROLA SOLUTIONS INC</td>\n",
       "      <td>7.0904</td>\n",
       "      <td>11.0</td>\n",
       "      <td>67.278</td>\n",
       "      <td>0.569087</td>\n",
       "      <td>12.121</td>\n",
       "      <td>21.925</td>\n",
       "      <td>23.798</td>\n",
       "      <td>...</td>\n",
       "      <td>3.131228</td>\n",
       "      <td>2.329234</td>\n",
       "      <td>0.017000</td>\n",
       "      <td>0.486003</td>\n",
       "      <td>1.591086</td>\n",
       "      <td>1.785936</td>\n",
       "      <td>1</td>\n",
       "      <td>2018-12-01</td>\n",
       "      <td>MSI</td>\n",
       "      <td>0.96</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>364541</th>\n",
       "      <td>9225</td>\n",
       "      <td>2018-11-30</td>\n",
       "      <td>ROLLINS INC</td>\n",
       "      <td>7.8378</td>\n",
       "      <td>11.0</td>\n",
       "      <td>83.129</td>\n",
       "      <td>0.061000</td>\n",
       "      <td>29.510</td>\n",
       "      <td>40.687</td>\n",
       "      <td>40.687</td>\n",
       "      <td>...</td>\n",
       "      <td>18.566000</td>\n",
       "      <td>3.953000</td>\n",
       "      <td>0.011100</td>\n",
       "      <td>0.486003</td>\n",
       "      <td>1.591086</td>\n",
       "      <td>1.785936</td>\n",
       "      <td>1</td>\n",
       "      <td>2018-12-01</td>\n",
       "      <td>ROL</td>\n",
       "      <td>0.96</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1597682</th>\n",
       "      <td>133504</td>\n",
       "      <td>2018-11-30</td>\n",
       "      <td>LUMINEX CORP</td>\n",
       "      <td>2.0855</td>\n",
       "      <td>14.0</td>\n",
       "      <td>46.203</td>\n",
       "      <td>0.360000</td>\n",
       "      <td>15.882</td>\n",
       "      <td>34.250</td>\n",
       "      <td>34.663</td>\n",
       "      <td>...</td>\n",
       "      <td>2.709000</td>\n",
       "      <td>1.465000</td>\n",
       "      <td>0.008340</td>\n",
       "      <td>0.486003</td>\n",
       "      <td>1.591086</td>\n",
       "      <td>1.785936</td>\n",
       "      <td>1</td>\n",
       "      <td>2018-12-01</td>\n",
       "      <td>LMNX</td>\n",
       "      <td>0.96</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>10 rows × 74 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          gvkey   datadate                         conm    trt1m  exchg  \\\n",
       "570920    13714 2018-11-30                     CBS CORP  -6.0785   11.0   \n",
       "571215    13714 2018-11-30                     CBS CORP  -5.5275   11.0   \n",
       "697805    17934 2018-11-30       SPROUTS FARMERS MARKET -14.3920   14.0   \n",
       "1792940  180652 2018-11-30         VERISK ANALYTICS INC   2.9039   14.0   \n",
       "1155825   31843 2018-11-30                 HARMONIC INC   1.9964   14.0   \n",
       "1735826  165993 2018-11-30  FIDELITY NATIONAL INFO SVCS   3.6984   11.0   \n",
       "156468     4503 2018-11-30             EXXON MOBIL CORP   0.8032   11.0   \n",
       "293188     7585 2018-11-30       MOTOROLA SOLUTIONS INC   7.0904   11.0   \n",
       "364541     9225 2018-11-30                  ROLLINS INC   7.8378   11.0   \n",
       "1597682  133504 2018-11-30                 LUMINEX CORP   2.0855   14.0   \n",
       "\n",
       "          CAPEI        bm     evm  pe_op_basic  pe_op_dil  ...        ptb  \\\n",
       "570920   12.998  0.124000   9.879       12.202     12.307  ...   7.407000   \n",
       "571215   12.998  0.124000   9.879       12.202     12.307  ...   7.407000   \n",
       "697805   28.920  0.229000  10.180       21.862     22.223  ...   5.313000   \n",
       "1792940  44.324  0.137000  17.132       37.333     38.166  ...   8.141000   \n",
       "1155825 -12.288  0.608000  72.770      -11.479    -11.479  ...   2.164000   \n",
       "1735826  48.746  0.339000  13.630       40.193     40.664  ...   2.866000   \n",
       "156468   15.907  0.611000  10.946       22.701     22.701  ...   1.576000   \n",
       "293188   67.278  0.569087  12.121       21.925     23.798  ...   3.131228   \n",
       "364541   83.129  0.061000  29.510       40.687     40.687  ...  18.566000   \n",
       "1597682  46.203  0.360000  15.882       34.250     34.663  ...   2.709000   \n",
       "\n",
       "         PEG_trailing  DIVYIELD  PEG_1yrforward  PEG_ltgforward    sprtrn  \\\n",
       "570920       1.048000  0.012600        0.486003        1.591086  1.785936   \n",
       "571215       1.048000  0.012600        0.486003        1.591086  1.785936   \n",
       "697805       0.801000  0.022435        0.486003        1.591086  1.785936   \n",
       "1792940      2.917000  0.022435        0.486003        1.591086  1.785936   \n",
       "1155825     -0.045000  0.022435        0.486003        1.591086  1.785936   \n",
       "1735826      0.756000  0.012300        0.486003        1.591086  1.785936   \n",
       "156468       1.521000  0.041200        0.486003        1.591086  1.785936   \n",
       "293188       2.329234  0.017000        0.486003        1.591086  1.785936   \n",
       "364541       3.953000  0.011100        0.486003        1.591086  1.785936   \n",
       "1597682      1.465000  0.008340        0.486003        1.591086  1.785936   \n",
       "\n",
       "         win  decision_date    tic  predictions  \n",
       "570920     0     2018-12-01  CBS.A         0.98  \n",
       "571215     0     2018-12-01    CBS         0.98  \n",
       "697805     0     2018-12-01    SFM         0.98  \n",
       "1792940    1     2018-12-01   VRSK         0.98  \n",
       "1155825    1     2018-12-01   HLIT         0.97  \n",
       "1735826    1     2018-12-01    FIS         0.97  \n",
       "156468     0     2018-12-01    XOM         0.96  \n",
       "293188     1     2018-12-01    MSI         0.96  \n",
       "364541     1     2018-12-01    ROL         0.96  \n",
       "1597682    1     2018-12-01   LMNX         0.96  \n",
       "\n",
       "[10 rows x 74 columns]"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "portfolio=results[0].nlargest(10, 'predictions')\n",
    "portfolio"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The compounded return of the portfolio over the period is: -4.22 %\n",
      "The compounded return of the SP500 over the period is: 1.79 %\n"
     ]
    }
   ],
   "source": [
    "monthly_yields=[]\n",
    "SP500_monthly_yields=[]\n",
    "compounded_portfolio=1\n",
    "compounded_SP500=1\n",
    "for i in range(0,len(results)):\n",
    "    monthly_yields.append((results[i].nlargest(5, 'predictions')).trt1m.mean().round(2))\n",
    "    SP500_monthly_yields.append(results[i].sprtrn.mean().round(2))\n",
    "    compounded_portfolio=(compounded_portfolio*(1+(monthly_yields[i]/100)))\n",
    "    compounded_SP500=compounded_SP500*(1+(SP500_monthly_yields[i]/100))\n",
    "print('The compounded return of the portfolio over the period is:',round(((compounded_portfolio-1))*100,2),'%')\n",
    "print('The compounded return of the SP500 over the period is:',round((compounded_SP500-1)*100,2),'%')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Variable: CAPEI                Importance: 0.06\n",
      "Variable: pcf                  Importance: 0.06\n",
      "Variable: ptb                  Importance: 0.06\n",
      "Variable: PEG_1yrforward       Importance: 0.06\n",
      "Variable: ps                   Importance: 0.05\n",
      "Variable: PEG_ltgforward       Importance: 0.05\n",
      "Variable: pe_op_basic          Importance: 0.04\n",
      "Variable: pe_op_dil            Importance: 0.04\n",
      "Variable: pe_exi               Importance: 0.04\n",
      "Variable: pe_inc               Importance: 0.04\n",
      "Variable: PEG_trailing         Importance: 0.04\n",
      "Variable: DIVYIELD             Importance: 0.04\n",
      "Variable: bm                   Importance: 0.01\n",
      "Variable: evm                  Importance: 0.01\n",
      "Variable: dpr                  Importance: 0.01\n",
      "Variable: npm                  Importance: 0.01\n",
      "Variable: opmbd                Importance: 0.01\n",
      "Variable: gpm                  Importance: 0.01\n",
      "Variable: roa                  Importance: 0.01\n",
      "Variable: roe                  Importance: 0.01\n",
      "Variable: roce                 Importance: 0.01\n",
      "Variable: efftax               Importance: 0.01\n",
      "Variable: aftret_eq            Importance: 0.01\n",
      "Variable: aftret_invcapx       Importance: 0.01\n",
      "Variable: pretret_noa          Importance: 0.01\n",
      "Variable: pretret_earnat       Importance: 0.01\n",
      "Variable: GProf                Importance: 0.01\n",
      "Variable: equity_invcap        Importance: 0.01\n",
      "Variable: totdebt_invcap       Importance: 0.01\n",
      "Variable: int_debt             Importance: 0.01\n",
      "Variable: int_totdebt          Importance: 0.01\n",
      "Variable: cash_lt              Importance: 0.01\n",
      "Variable: invt_act             Importance: 0.01\n",
      "Variable: rect_act             Importance: 0.01\n",
      "Variable: debt_at              Importance: 0.01\n",
      "Variable: short_debt           Importance: 0.01\n",
      "Variable: curr_debt            Importance: 0.01\n",
      "Variable: lt_debt              Importance: 0.01\n",
      "Variable: profit_lct           Importance: 0.01\n",
      "Variable: ocf_lct              Importance: 0.01\n",
      "Variable: cash_debt            Importance: 0.01\n",
      "Variable: fcf_ocf              Importance: 0.01\n",
      "Variable: lt_ppent             Importance: 0.01\n",
      "Variable: dltt_be              Importance: 0.01\n",
      "Variable: debt_assets          Importance: 0.01\n",
      "Variable: debt_capital         Importance: 0.01\n",
      "Variable: de_ratio             Importance: 0.01\n",
      "Variable: intcov               Importance: 0.01\n",
      "Variable: intcov_ratio         Importance: 0.01\n",
      "Variable: cash_ratio           Importance: 0.01\n",
      "Variable: quick_ratio          Importance: 0.01\n",
      "Variable: curr_ratio           Importance: 0.01\n",
      "Variable: cash_conversion      Importance: 0.01\n",
      "Variable: inv_turn             Importance: 0.01\n",
      "Variable: at_turn              Importance: 0.01\n",
      "Variable: rect_turn            Importance: 0.01\n",
      "Variable: pay_turn             Importance: 0.01\n",
      "Variable: sale_invcap          Importance: 0.01\n",
      "Variable: sale_equity          Importance: 0.01\n",
      "Variable: sale_nwc             Importance: 0.01\n",
      "Variable: accrual              Importance: 0.01\n",
      "Variable: exchg                Importance: 0.0\n",
      "Variable: rd_sale              Importance: 0.0\n",
      "Variable: adv_sale             Importance: 0.0\n",
      "Variable: staff_sale           Importance: 0.0\n"
     ]
    }
   ],
   "source": [
    "[print('Variable: {:20} Importance: {}'.format(*pair)) for pair in feature_imp];"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "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": [
    "# Import matplotlib for plotting and use magic command for Jupyter Notebooks\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "# list of x locations for plotting\n",
    "x_values = list(range(len(importances)))\n",
    "# Make a bar chart\n",
    "plt.bar(x_values, importances, orientation = 'vertical')\n",
    "# Tick labels for x axis\n",
    "plt.xticks(x_values, train_feature_list, rotation='vertical')\n",
    "plt.grid(True)\n",
    "plt.box(True)\n",
    "plt.rcParams[\"figure.facecolor\"] = \"w\"\n",
    "# Axis labels and title\n",
    "plt.ylabel('Importance'); plt.xlabel('Feature'); plt.title('Feature Importance');\n",
    "#plt.savefig('Importance des variables_1.png', bbox_inches = 'tight')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "## We can see that a lot of labels have very small importance \n",
    "# Lets try removing features with importance value smaller than 0.02\n",
    "df_importance = pd.DataFrame(feature_imp, columns = ['Feature', 'Importance']) \n",
    "feature_to_drop=df_importance.Feature[df_importance['Importance']<0.02]\n",
    "new_db=clean_removed_db.drop(columns=feature_to_drop)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "## We can run the same backtest over one month with the simplified db:\n",
    "results2,feature_imp2,importances2,train_feature_list2=rfstockpicker_backtest(new_db,(\"2018-12-01\"),7,1,100,24)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "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>gvkey</th>\n",
       "      <th>datadate</th>\n",
       "      <th>conm</th>\n",
       "      <th>trt1m</th>\n",
       "      <th>CAPEI</th>\n",
       "      <th>pe_op_basic</th>\n",
       "      <th>pe_op_dil</th>\n",
       "      <th>pe_exi</th>\n",
       "      <th>pe_inc</th>\n",
       "      <th>ps</th>\n",
       "      <th>...</th>\n",
       "      <th>ptb</th>\n",
       "      <th>PEG_trailing</th>\n",
       "      <th>DIVYIELD</th>\n",
       "      <th>PEG_1yrforward</th>\n",
       "      <th>PEG_ltgforward</th>\n",
       "      <th>sprtrn</th>\n",
       "      <th>win</th>\n",
       "      <th>decision_date</th>\n",
       "      <th>tic</th>\n",
       "      <th>predictions</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>51919</th>\n",
       "      <td>2176</td>\n",
       "      <td>2018-11-30</td>\n",
       "      <td>BERKSHIRE HATHAWAY</td>\n",
       "      <td>5.9456</td>\n",
       "      <td>11.279</td>\n",
       "      <td>25.952</td>\n",
       "      <td>25.952</td>\n",
       "      <td>10.664</td>\n",
       "      <td>10.664</td>\n",
       "      <td>1.176</td>\n",
       "      <td>...</td>\n",
       "      <td>0.783000</td>\n",
       "      <td>0.220000</td>\n",
       "      <td>0.022435</td>\n",
       "      <td>0.486003</td>\n",
       "      <td>1.591086</td>\n",
       "      <td>1.785936</td>\n",
       "      <td>1</td>\n",
       "      <td>2018-12-01</td>\n",
       "      <td>BRK.A</td>\n",
       "      <td>0.935</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>52199</th>\n",
       "      <td>2176</td>\n",
       "      <td>2018-11-30</td>\n",
       "      <td>BERKSHIRE HATHAWAY</td>\n",
       "      <td>6.3133</td>\n",
       "      <td>11.279</td>\n",
       "      <td>25.952</td>\n",
       "      <td>25.952</td>\n",
       "      <td>10.664</td>\n",
       "      <td>10.664</td>\n",
       "      <td>1.176</td>\n",
       "      <td>...</td>\n",
       "      <td>0.783000</td>\n",
       "      <td>0.220000</td>\n",
       "      <td>0.022435</td>\n",
       "      <td>0.486003</td>\n",
       "      <td>1.591086</td>\n",
       "      <td>1.785936</td>\n",
       "      <td>1</td>\n",
       "      <td>2018-12-01</td>\n",
       "      <td>BRK.B</td>\n",
       "      <td>0.935</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1155825</th>\n",
       "      <td>31843</td>\n",
       "      <td>2018-11-30</td>\n",
       "      <td>HARMONIC INC</td>\n",
       "      <td>1.9964</td>\n",
       "      <td>-12.288</td>\n",
       "      <td>-11.479</td>\n",
       "      <td>-11.479</td>\n",
       "      <td>-10.396</td>\n",
       "      <td>-10.396</td>\n",
       "      <td>1.250</td>\n",
       "      <td>...</td>\n",
       "      <td>2.164000</td>\n",
       "      <td>-0.045000</td>\n",
       "      <td>0.022435</td>\n",
       "      <td>0.486003</td>\n",
       "      <td>1.591086</td>\n",
       "      <td>1.785936</td>\n",
       "      <td>1</td>\n",
       "      <td>2018-12-01</td>\n",
       "      <td>HLIT</td>\n",
       "      <td>0.930</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1638810</th>\n",
       "      <td>142956</td>\n",
       "      <td>2018-11-30</td>\n",
       "      <td>OMNICELL INC</td>\n",
       "      <td>9.2362</td>\n",
       "      <td>127.616</td>\n",
       "      <td>81.264</td>\n",
       "      <td>83.176</td>\n",
       "      <td>57.951</td>\n",
       "      <td>57.951</td>\n",
       "      <td>3.714</td>\n",
       "      <td>...</td>\n",
       "      <td>4.494000</td>\n",
       "      <td>2.329234</td>\n",
       "      <td>0.022435</td>\n",
       "      <td>0.486003</td>\n",
       "      <td>1.591086</td>\n",
       "      <td>1.785936</td>\n",
       "      <td>1</td>\n",
       "      <td>2018-12-01</td>\n",
       "      <td>OMCL</td>\n",
       "      <td>0.930</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1653933</th>\n",
       "      <td>147204</td>\n",
       "      <td>2018-11-30</td>\n",
       "      <td>COGENT COMMUNICATIONS HLDGS</td>\n",
       "      <td>-5.6368</td>\n",
       "      <td>135.522</td>\n",
       "      <td>106.082</td>\n",
       "      <td>108.292</td>\n",
       "      <td>226.000</td>\n",
       "      <td>226.000</td>\n",
       "      <td>4.772</td>\n",
       "      <td>...</td>\n",
       "      <td>3.131228</td>\n",
       "      <td>2.329234</td>\n",
       "      <td>0.041600</td>\n",
       "      <td>0.486003</td>\n",
       "      <td>1.591086</td>\n",
       "      <td>1.785936</td>\n",
       "      <td>0</td>\n",
       "      <td>2018-12-01</td>\n",
       "      <td>CCOI</td>\n",
       "      <td>0.920</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1704668</th>\n",
       "      <td>162127</td>\n",
       "      <td>2018-11-30</td>\n",
       "      <td>CONSOLIDATED COMM HLDGS INC</td>\n",
       "      <td>9.9042</td>\n",
       "      <td>52.264</td>\n",
       "      <td>-25.551</td>\n",
       "      <td>-25.551</td>\n",
       "      <td>18.412</td>\n",
       "      <td>18.412</td>\n",
       "      <td>0.628</td>\n",
       "      <td>...</td>\n",
       "      <td>1.239000</td>\n",
       "      <td>0.009000</td>\n",
       "      <td>0.124000</td>\n",
       "      <td>0.486003</td>\n",
       "      <td>1.591086</td>\n",
       "      <td>1.785936</td>\n",
       "      <td>1</td>\n",
       "      <td>2018-12-01</td>\n",
       "      <td>CNSL</td>\n",
       "      <td>0.920</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1787717</th>\n",
       "      <td>179666</td>\n",
       "      <td>2018-11-30</td>\n",
       "      <td>LOGMEIN INC</td>\n",
       "      <td>7.4431</td>\n",
       "      <td>193.336</td>\n",
       "      <td>141.180</td>\n",
       "      <td>143.533</td>\n",
       "      <td>32.996</td>\n",
       "      <td>32.996</td>\n",
       "      <td>3.899</td>\n",
       "      <td>...</td>\n",
       "      <td>1.335000</td>\n",
       "      <td>2.329234</td>\n",
       "      <td>0.013900</td>\n",
       "      <td>0.486003</td>\n",
       "      <td>1.591086</td>\n",
       "      <td>1.785936</td>\n",
       "      <td>1</td>\n",
       "      <td>2018-12-01</td>\n",
       "      <td>LOGM</td>\n",
       "      <td>0.920</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>233585</th>\n",
       "      <td>6266</td>\n",
       "      <td>2018-11-30</td>\n",
       "      <td>JOHNSON &amp; JOHNSON</td>\n",
       "      <td>5.5790</td>\n",
       "      <td>27.940</td>\n",
       "      <td>21.504</td>\n",
       "      <td>22.185</td>\n",
       "      <td>305.000</td>\n",
       "      <td>321.000</td>\n",
       "      <td>4.653</td>\n",
       "      <td>...</td>\n",
       "      <td>5.321000</td>\n",
       "      <td>2.329234</td>\n",
       "      <td>0.025700</td>\n",
       "      <td>0.486003</td>\n",
       "      <td>1.591086</td>\n",
       "      <td>1.785936</td>\n",
       "      <td>1</td>\n",
       "      <td>2018-12-01</td>\n",
       "      <td>JNJ</td>\n",
       "      <td>0.910</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>293957</th>\n",
       "      <td>7620</td>\n",
       "      <td>2018-11-30</td>\n",
       "      <td>MURPHY OIL CORP</td>\n",
       "      <td>0.9102</td>\n",
       "      <td>-31.460</td>\n",
       "      <td>245.000</td>\n",
       "      <td>252.500</td>\n",
       "      <td>-39.825</td>\n",
       "      <td>-38.854</td>\n",
       "      <td>2.470</td>\n",
       "      <td>...</td>\n",
       "      <td>1.149000</td>\n",
       "      <td>2.329234</td>\n",
       "      <td>0.031400</td>\n",
       "      <td>0.486003</td>\n",
       "      <td>1.591086</td>\n",
       "      <td>1.785936</td>\n",
       "      <td>0</td>\n",
       "      <td>2018-12-01</td>\n",
       "      <td>MUR</td>\n",
       "      <td>0.900</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>571291</th>\n",
       "      <td>13721</td>\n",
       "      <td>2018-11-30</td>\n",
       "      <td>ZOETIS INC</td>\n",
       "      <td>4.2662</td>\n",
       "      <td>66.656</td>\n",
       "      <td>33.143</td>\n",
       "      <td>33.389</td>\n",
       "      <td>39.539</td>\n",
       "      <td>39.539</td>\n",
       "      <td>7.755</td>\n",
       "      <td>...</td>\n",
       "      <td>19.753000</td>\n",
       "      <td>0.978000</td>\n",
       "      <td>0.005590</td>\n",
       "      <td>0.486003</td>\n",
       "      <td>1.591086</td>\n",
       "      <td>1.785936</td>\n",
       "      <td>1</td>\n",
       "      <td>2018-12-01</td>\n",
       "      <td>ZTS</td>\n",
       "      <td>0.900</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>10 rows × 21 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          gvkey   datadate                         conm   trt1m    CAPEI  \\\n",
       "51919      2176 2018-11-30           BERKSHIRE HATHAWAY  5.9456   11.279   \n",
       "52199      2176 2018-11-30           BERKSHIRE HATHAWAY  6.3133   11.279   \n",
       "1155825   31843 2018-11-30                 HARMONIC INC  1.9964  -12.288   \n",
       "1638810  142956 2018-11-30                 OMNICELL INC  9.2362  127.616   \n",
       "1653933  147204 2018-11-30  COGENT COMMUNICATIONS HLDGS -5.6368  135.522   \n",
       "1704668  162127 2018-11-30  CONSOLIDATED COMM HLDGS INC  9.9042   52.264   \n",
       "1787717  179666 2018-11-30                  LOGMEIN INC  7.4431  193.336   \n",
       "233585     6266 2018-11-30            JOHNSON & JOHNSON  5.5790   27.940   \n",
       "293957     7620 2018-11-30              MURPHY OIL CORP  0.9102  -31.460   \n",
       "571291    13721 2018-11-30                   ZOETIS INC  4.2662   66.656   \n",
       "\n",
       "         pe_op_basic  pe_op_dil   pe_exi   pe_inc     ps  ...        ptb  \\\n",
       "51919         25.952     25.952   10.664   10.664  1.176  ...   0.783000   \n",
       "52199         25.952     25.952   10.664   10.664  1.176  ...   0.783000   \n",
       "1155825      -11.479    -11.479  -10.396  -10.396  1.250  ...   2.164000   \n",
       "1638810       81.264     83.176   57.951   57.951  3.714  ...   4.494000   \n",
       "1653933      106.082    108.292  226.000  226.000  4.772  ...   3.131228   \n",
       "1704668      -25.551    -25.551   18.412   18.412  0.628  ...   1.239000   \n",
       "1787717      141.180    143.533   32.996   32.996  3.899  ...   1.335000   \n",
       "233585        21.504     22.185  305.000  321.000  4.653  ...   5.321000   \n",
       "293957       245.000    252.500  -39.825  -38.854  2.470  ...   1.149000   \n",
       "571291        33.143     33.389   39.539   39.539  7.755  ...  19.753000   \n",
       "\n",
       "         PEG_trailing  DIVYIELD  PEG_1yrforward  PEG_ltgforward    sprtrn  \\\n",
       "51919        0.220000  0.022435        0.486003        1.591086  1.785936   \n",
       "52199        0.220000  0.022435        0.486003        1.591086  1.785936   \n",
       "1155825     -0.045000  0.022435        0.486003        1.591086  1.785936   \n",
       "1638810      2.329234  0.022435        0.486003        1.591086  1.785936   \n",
       "1653933      2.329234  0.041600        0.486003        1.591086  1.785936   \n",
       "1704668      0.009000  0.124000        0.486003        1.591086  1.785936   \n",
       "1787717      2.329234  0.013900        0.486003        1.591086  1.785936   \n",
       "233585       2.329234  0.025700        0.486003        1.591086  1.785936   \n",
       "293957       2.329234  0.031400        0.486003        1.591086  1.785936   \n",
       "571291       0.978000  0.005590        0.486003        1.591086  1.785936   \n",
       "\n",
       "         win  decision_date    tic predictions  \n",
       "51919      1     2018-12-01  BRK.A       0.935  \n",
       "52199      1     2018-12-01  BRK.B       0.935  \n",
       "1155825    1     2018-12-01   HLIT       0.930  \n",
       "1638810    1     2018-12-01   OMCL       0.930  \n",
       "1653933    0     2018-12-01   CCOI       0.920  \n",
       "1704668    1     2018-12-01   CNSL       0.920  \n",
       "1787717    1     2018-12-01   LOGM       0.920  \n",
       "233585     1     2018-12-01    JNJ       0.910  \n",
       "293957     0     2018-12-01    MUR       0.900  \n",
       "571291     1     2018-12-01    ZTS       0.900  \n",
       "\n",
       "[10 rows x 21 columns]"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results2[0].nlargest(10, 'predictions')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The compounded return of the portfolio over the period is: 3.57 %\n",
      "The compounded return of the SP500 over the period is: 1.79 %\n"
     ]
    }
   ],
   "source": [
    "monthly_yields=[]\n",
    "SP500_monthly_yields=[]\n",
    "compounded_portfolio=1\n",
    "compounded_SP500=1\n",
    "for i in range(0,len(results2)):\n",
    "    monthly_yields.append((results2[i].nlargest(5, 'predictions')).trt1m.mean().round(2))\n",
    "    SP500_monthly_yields.append(results2[i].sprtrn.mean().round(2))\n",
    "    compounded_portfolio=(compounded_portfolio*(1+(monthly_yields[i]/100)))\n",
    "    compounded_SP500=compounded_SP500*(1+(SP500_monthly_yields[i]/100))\n",
    "print('The compounded return of the portfolio over the period is:',round(((compounded_portfolio-1))*100,2),'%')\n",
    "print('The compounded return of the SP500 over the period is:',round((compounded_SP500-1)*100,2),'%')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "results3,feature_imp3,importances3,train_feature_list3=rfstockpicker_backtest(clean_removed_db,(\"2018-12-01\"),7,(12*15),100,24)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The compounded return of the portfolio over the period is: 572.72 %\n",
      "The compounded return of the SP500 over the period is: 260.95 %\n"
     ]
    }
   ],
   "source": [
    "monthly_yields=[]\n",
    "monthly_date=[]\n",
    "SP500_monthly_yields=[]\n",
    "ret_model=[]\n",
    "ret_SP500=[]\n",
    "compounded_portfolio=1\n",
    "compounded_SP500=1\n",
    "for i in range(len(results3) - 1, -1,-1):\n",
    "    monthly_yields.append((results3[i].nlargest(10, 'predictions')).trt1m.mean().round(2))\n",
    "    SP500_monthly_yields.append(results3[i].sprtrn.mean().round(2))\n",
    "    monthly_date.append(results3[i].datadate.iloc[0])\n",
    "    \n",
    "for i in range(0,len(monthly_yields)):\n",
    "    compounded_portfolio=(compounded_portfolio*(1+(monthly_yields[i]/100)))\n",
    "    compounded_SP500=(compounded_SP500*(1+(SP500_monthly_yields[i]/100)))\n",
    "    \n",
    "    ret_model.append(compounded_portfolio)\n",
    "    ret_SP500.append(compounded_SP500)\n",
    "    \n",
    "print('The compounded return of the portfolio over the period is:',round(((compounded_portfolio)*100),2),'%')\n",
    "print('The compounded return of the SP500 over the period is:',round((compounded_SP500)*100,2),'%')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "results4,feature_imp4,importances4,train_feature_list4=rfstockpicker_backtest(new_db,(\"2018-12-01\"),7,(12*15),100,24)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The compounded return of the portfolio over the period is: 1420.52 %\n",
      "The compounded return of the SP500 over the period is: 260.95 %\n"
     ]
    }
   ],
   "source": [
    "monthly_yields=[]\n",
    "monthly_date=[]\n",
    "SP500_monthly_yields=[]\n",
    "ret_model=[]\n",
    "ret_SP500=[]\n",
    "compounded_portfolio=1\n",
    "compounded_SP500=1\n",
    "for i in range(len(results4) - 1, -1,-1):\n",
    "    monthly_yields.append((results4[i].nlargest(10, 'predictions')).trt1m.mean().round(2))\n",
    "    SP500_monthly_yields.append(results4[i].sprtrn.mean().round(2))\n",
    "    monthly_date.append(results4[i].datadate.iloc[0])\n",
    "for i in range(0,len(monthly_yields)):\n",
    "    compounded_portfolio=(compounded_portfolio*(1+(monthly_yields[i]/100)))\n",
    "    compounded_SP500=(compounded_SP500*(1+(SP500_monthly_yields[i]/100)))\n",
    "    \n",
    "    ret_model.append(compounded_portfolio)\n",
    "    ret_SP500.append(compounded_SP500)\n",
    "    \n",
    "print('The compounded return of the portfolio over the period is:',round(((compounded_portfolio)*100),2),'%')\n",
    "print('The compounded return of the SP500 over the period is:',round((compounded_SP500)*100,2),'%')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7f15c4cb0cd0>"
      ]
     },
     "execution_count": 71,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt \n",
    "\n",
    "plt.plot(monthly_date,SP500_monthly_yields,label=\"Return of the SP500\")\n",
    "plt.plot(monthly_date,monthly_yields,label=\"Return of the model\")\n",
    "\n",
    "plt.ylabel('Monthly return %')\n",
    "plt.title('Performance from 2003 to 2019')\n",
    "plt.grid(True)\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7f15c84ac450>"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(monthly_date,ret_SP500,label=\"Return of the SP500\")\n",
    "plt.plot(monthly_date,ret_model,label=\"Return of the model\")\n",
    "\n",
    "plt.ylabel('')\n",
    "plt.title('Growth of hypothetical 1$ from December 2003 to December 2018')\n",
    "plt.grid(True)\n",
    "plt.legend()"
   ]
  },
  {
   "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": 4
}