Chapter 03 - Meta-Labeling.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from collections import defaultdict\n",
    "from functools import reduce\n",
    "from path import Path\n",
    "from pprint import pprint\n",
    "\n",
    "%matplotlib inline\n",
    "plt.style.use('ggplot')\n",
    "plt.rcParams['figure.figsize'] = 16,6"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note: During one of his talks Marcos' mentioned that this is one of his favorite applications of ML in Finance"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3.1a\n",
    "\n",
    "Form dollar bars for E-mini S&P 500 futures:\n",
    "\n",
    "Apply a symmetric CUSUM filter where the threshold is the standard deviation of daily returns."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sampling import dollar_bars\n",
    "from filters import cusum\n",
    "from multiprocess import mpPandasObj\n",
    "from load_data import load_contracts\n",
    "\n",
    "es_contracts = load_contracts('@ES')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "def getDailyVol(close, span0=100):\n",
    "    # daily vol, reindexed to cloes\n",
    "    df0 = close.index.searchsorted(close.index - pd.Timedelta(days=1))\n",
    "    df0 = df0[df0>0]\n",
    "    df0 = pd.Series(close.index[df0 - 1], index=close.index[close.shape[0] - df0.shape[0]:])\n",
    "    df0 = close.loc[df0.index] / close.loc[df0.values].values - 1 # daily returns\n",
    "    df0 = df0.ewm(span=span0).std()\n",
    "    return df0\n",
    "\n",
    "es_dbars = dollar_bars(es_contracts, 100000000)\n",
    "daily_vol = getDailyVol(es_dbars['Close'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x116c08588>"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x432 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "daily_vol.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "daily_vol_mean = daily_vol.mean()\n",
    "sampled_es_dbars = cusum(es_dbars['Close'], daily_vol_mean)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3.1b\n",
    "\n",
    "Form dollar bars for E-mini S&P 500 futures:\n",
    "\n",
    "Use Snippet 3.4 on a pandas series t1, where numDays=1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "def getVerticalBarriers(close, tEvents, numDays):\n",
    "    t1 = close.index.searchsorted(tEvents+pd.Timedelta(days=numDays))\n",
    "    t1 = t1[t1 < close.shape[0]]\n",
    "    t1 = pd.Series(close.index[t1], index=tEvents[:t1.shape[0]]) # NaNs at the end\n",
    "    return t1\n",
    "\n",
    "close = es_dbars['Close']\n",
    "tEvents = sampled_es_dbars\n",
    "numDays = 1\n",
    "\n",
    "t1 = getVerticalBarriers(close, tEvents, numDays)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2016-11-30 11:09:00   2016-12-05 15:12:00\n",
       "2016-12-08 09:46:00   2016-12-09 09:57:00\n",
       "2016-12-08 13:57:00   2016-12-09 14:09:00\n",
       "2016-12-08 16:10:00   2016-12-11 18:02:00\n",
       "2016-12-09 09:18:00   2016-12-11 18:02:00\n",
       "dtype: datetime64[ns]"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "t1.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3.1c\n",
    "\n",
    "Form dollar bars for E-mini S&P 500 futures:\n",
    "\n",
    "On those sampled features, apply the triple-barrier method, where ptS1 = [1,1] and t1 is the series you created in point 1.b"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "def getEvents(close, tEvents, ptSl, trgt, minRet, numThreads=1, t1=False, side=None):\n",
    "    # 1) get target\n",
    "    trgt = trgt.loc[tEvents]\n",
    "    trgt = trgt[trgt > minRet]\n",
    "    # 2) get t1 (max holding period)\n",
    "    if t1 is False:\n",
    "        t1 = pd.Series(pd.NaT, index=tEvents)\n",
    "    # 3) form events object, apply stop loss on t1\n",
    "    if side is None:\n",
    "        side_, ptSl_ = pd.Series(1.0, index=trgt.index), [ptSl[0], ptSl[0]]\n",
    "    else:\n",
    "        side_, ptSl_ = side.loc[trgt.index], ptSl[:2]\n",
    "    events = pd.concat({'t1': t1, 'trgt': trgt, 'side': side_}, axis=1).dropna(subset=['trgt'])\n",
    "    df0 = mpPandasObj(func=applyPtSlOnT1, pdObj=('molecule', events.index), numThreads=numThreads, close=close, events=events, ptSl=ptSl_)\n",
    "    events['t1'] = df0.dropna(how='all').min(axis=1) # pd.min ignores NaN\n",
    "    if side is None:\n",
    "        events = events.drop('side', axis=1)\n",
    "\n",
    "    # store for later\n",
    "    events['pt'] = ptSl[0]\n",
    "    events['sl'] = ptSl[1]\n",
    "\n",
    "    return events\n",
    "    \n",
    "def applyPtSlOnT1(close, events, ptSl, molecule):\n",
    "    # apply stop loss/profit taking, if it takes place before t1 (end of event)\n",
    "    events_ = events.loc[molecule]\n",
    "    out = events_[['t1']].copy(deep=True)\n",
    "\n",
    "    if ptSl[0] > 0:\n",
    "        pt = ptSl[0] * events_['trgt']\n",
    "    else:\n",
    "        pt = pd.Series(index=events.index) # NaNs\n",
    "\n",
    "    if ptSl[1] > 0:\n",
    "        sl = - ptSl[1] * events_['trgt']\n",
    "    else:\n",
    "        sl = pd.Series(index=events.index) # 'mo NaNs\n",
    "\n",
    "    for loc, t1 in events_['t1'].fillna(close.index[-1]).iteritems():\n",
    "        df0 = close[loc:t1] # path prices\n",
    "        df0 = (df0 / close[loc] - 1) * events_.at[loc, 'side'] # path returns\n",
    "        out.loc[loc, 'sl'] = df0[df0<sl[loc]].index.min() # earliest stop loss\n",
    "        out.loc[loc, 'pt'] = df0[df0>pt[loc]].index.min() # earliest profit take\n",
    "    return out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [],
   "source": [
    "events = getEvents(es_dbars['Close'], tEvents=tEvents, ptSl=[1,1], t1=t1, numThreads=1, trgt=daily_vol, minRet=0.01)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>t1</th>\n",
       "      <th>trgt</th>\n",
       "      <th>pt</th>\n",
       "      <th>sl</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2016-11-30 11:09:00</th>\n",
       "      <td>2016-12-05 15:12:00</td>\n",
       "      <td>0.012111</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-05 10:28:00</th>\n",
       "      <td>2018-02-05 13:02:00</td>\n",
       "      <td>0.010232</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-05 10:34:00</th>\n",
       "      <td>2018-02-05 13:09:00</td>\n",
       "      <td>0.010271</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-05 10:54:00</th>\n",
       "      <td>2018-02-05 13:02:00</td>\n",
       "      <td>0.010475</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-05 11:01:00</th>\n",
       "      <td>2018-02-05 13:05:00</td>\n",
       "      <td>0.010512</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                     t1      trgt  pt  sl\n",
       "2016-11-30 11:09:00 2016-12-05 15:12:00  0.012111   1   1\n",
       "2018-02-05 10:28:00 2018-02-05 13:02:00  0.010232   1   1\n",
       "2018-02-05 10:34:00 2018-02-05 13:09:00  0.010271   1   1\n",
       "2018-02-05 10:54:00 2018-02-05 13:02:00  0.010475   1   1\n",
       "2018-02-05 11:01:00 2018-02-05 13:05:00  0.010512   1   1"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "events.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3.1d\n",
    "\n",
    "Form dollar bars for E-mini S&P 500 futures:\n",
    "\n",
    "Apply `getBins` to generate the labels.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [],
   "source": [
    "def getBins(events, close):\n",
    "    # 1) prices aligned with events\n",
    "    events_ = events.dropna(subset=['t1'])\n",
    "    px = events_.index.union(events_['t1'].values).drop_duplicates()\n",
    "    px = close.reindex(px, method='bfill')\n",
    "    # 2) create out object\n",
    "    out = pd.DataFrame(index=events_.index)\n",
    "    out['ret'] = px.loc[events_['t1'].values].values / px.loc[events_.index] - 1\n",
    "    out['bin'] = np.sign(out['ret'])\n",
    "    return out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [],
   "source": [
    "bins = getBins(events, close)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "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>ret</th>\n",
       "      <th>bin</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2016-11-30 11:09:00</th>\n",
       "      <td>0.002844</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-05 10:28:00</th>\n",
       "      <td>-0.010336</td>\n",
       "      <td>-1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-05 10:34:00</th>\n",
       "      <td>-0.010532</td>\n",
       "      <td>-1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-05 10:54:00</th>\n",
       "      <td>-0.010785</td>\n",
       "      <td>-1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-05 11:01:00</th>\n",
       "      <td>-0.011426</td>\n",
       "      <td>-1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                          ret  bin\n",
       "2016-11-30 11:09:00  0.002844  1.0\n",
       "2018-02-05 10:28:00 -0.010336 -1.0\n",
       "2018-02-05 10:34:00 -0.010532 -1.0\n",
       "2018-02-05 10:54:00 -0.010785 -1.0\n",
       "2018-02-05 11:01:00 -0.011426 -1.0"
      ]
     },
     "execution_count": 67,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bins.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       " 1.0    997\n",
       "-1.0    747\n",
       " 0.0      4\n",
       "Name: bin, dtype: int64"
      ]
     },
     "execution_count": 68,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bins['bin'].value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3.2\n",
    "\n",
    "From exercise 1, use Snippet 3.8 to drop rare labels."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [],
   "source": [
    "def dropLabels(events, mitPct=0.05):\n",
    "    # apply weights, drop labels with insufficient examples\n",
    "    while True:\n",
    "        df0 = events['bin'].value_counts(normalize=True)\n",
    "        if df0.min() > mitPct or df0.shape[0] < 3:\n",
    "            break\n",
    "        print(\"Dropped label\", df0.argmin(), df0.min())\n",
    "        events = events[events['bin'] != df0.argmin()]\n",
    "    return events\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dropped label 0.0 0.002288329519450801\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/doda/anaconda3/envs/zippy/lib/python3.5/site-packages/ipykernel_launcher.py:7: FutureWarning: 'argmin' is deprecated, use 'idxmin' instead. The behavior of 'argmin'\n",
      "will be corrected to return the positional minimum in the future.\n",
      "Use 'series.values.argmin' to get the position of the minimum now.\n",
      "  import sys\n",
      "/Users/doda/anaconda3/envs/zippy/lib/python3.5/site-packages/ipykernel_launcher.py:8: FutureWarning: 'argmin' is deprecated, use 'idxmin' instead. The behavior of 'argmin'\n",
      "will be corrected to return the positional minimum in the future.\n",
      "Use 'series.values.argmin' to get the position of the minimum now.\n",
      "  \n"
     ]
    },
    {
     "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>ret</th>\n",
       "      <th>bin</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2016-11-30 11:09:00</th>\n",
       "      <td>0.002844</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-05 10:28:00</th>\n",
       "      <td>-0.010336</td>\n",
       "      <td>-1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-05 10:34:00</th>\n",
       "      <td>-0.010532</td>\n",
       "      <td>-1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-05 10:54:00</th>\n",
       "      <td>-0.010785</td>\n",
       "      <td>-1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-05 11:01:00</th>\n",
       "      <td>-0.011426</td>\n",
       "      <td>-1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                          ret  bin\n",
       "2016-11-30 11:09:00  0.002844  1.0\n",
       "2018-02-05 10:28:00 -0.010336 -1.0\n",
       "2018-02-05 10:34:00 -0.010532 -1.0\n",
       "2018-02-05 10:54:00 -0.010785 -1.0\n",
       "2018-02-05 11:01:00 -0.011426 -1.0"
      ]
     },
     "execution_count": 70,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dropLabels(bins).head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3.3\n",
    "\n",
    "Adjust the `getBins` function (Snippet 3.5) to return a 0 whenever the vertical barrier is the first one touched."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [],
   "source": [
    "def barrier_touched(out_df, events):\n",
    "    # We'll graciously use the barrier_touched method from\n",
    "    # https://github.com/hudson-and-thames/mlfinlab/blob/master/mlfinlab/labeling/labeling.py#L164\n",
    "    store = []\n",
    "    for date_time, values in out_df.iterrows():\n",
    "        ret = values['ret']\n",
    "        target = values['trgt']\n",
    "\n",
    "        pt_level_reached = ret > target * events.loc[date_time, 'pt']\n",
    "        sl_level_reached = ret < -target * events.loc[date_time, 'sl']\n",
    "\n",
    "        if ret > 0.0 and pt_level_reached:\n",
    "            # Top barrier reached\n",
    "            store.append(1)\n",
    "        elif ret < 0.0 and sl_level_reached:\n",
    "            # Bottom barrier reached\n",
    "            store.append(-1)\n",
    "        else:\n",
    "            # Vertical barrier reached\n",
    "            store.append(0)\n",
    "\n",
    "    # Save to 'bin' column and return\n",
    "    out_df['bin'] = store\n",
    "    return out_df\n",
    "\n",
    "\n",
    "def getBins(events, close):\n",
    "    '''\n",
    "    Compute event's outcome (including side information, if provided).\n",
    "    events is a DataFrame where:\n",
    "    -events.index is event's starttime\n",
    "    -events['t1'] is event's endtime\n",
    "    -events['trgt'] is event's target\n",
    "    -events['side'] (optional) implies the algo's position side\n",
    "    Case 1: ('side' not in events): bin in (-1, 1) <- label by price action\n",
    "    Case 2: ('side' in events): bin in (0, 1) <- label by pnl (meta-labeling)\n",
    "    '''\n",
    "    # 1) prices aligned with events\n",
    "    events_ = events.dropna(subset=['t1'])\n",
    "    px = events_.index.union(events_['t1'].values).drop_duplicates()\n",
    "    px = close.reindex(px, method='bfill')\n",
    "    # 2) create out object\n",
    "    out = pd.DataFrame(index=events_.index)\n",
    "\n",
    "    out['ret'] = px.loc[events_['t1'].values].values / px.loc[events_.index] - 1\n",
    "    if 'side' in events_:\n",
    "        out['ret'] *= events_['side']  # meta-labeling\n",
    "\n",
    "    out['trgt'] = events_['trgt']\n",
    "    out = barrier_touched(out, events)\n",
    "\n",
    "    if 'side' in events_:\n",
    "        out.loc[out['ret'] <= 0, 'bin'] = 0\n",
    "        \n",
    "    if 'side' in events_:\n",
    "        out['side'] = events['side']\n",
    "    return out\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [],
   "source": [
    "bins = getBins(events, close)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       " 0    825\n",
       "-1    482\n",
       " 1    441\n",
       "Name: bin, dtype: int64"
      ]
     },
     "execution_count": 73,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bins['bin'].value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3.4a\n",
    "\n",
    "Develop a trend-following strategy based on a popular technical analysis statistic (e.g. MA Crossover). For each observation, the model suggests a side, but not a size of the bet.\n",
    "\n",
    "Derive meta-labels for `ptSl = [1,2]` and `t1` where `numDays=1`. Use as `trgt` the daily standard deviation as computed by Snippet 3.1."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = es_dbars.copy().truncate(before='2017')\n",
    "tEvents = cusum(df['Close'], getDailyVol(df['Close']).mean())\n",
    "t1 = getVerticalBarriers(df['Close'], tEvents, numDays=1)\n",
    "events = getEvents(close, tEvents=tEvents, ptSl=[1,2], t1=t1, trgt=daily_vol, minRet=0.01)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {},
   "outputs": [],
   "source": [
    "slow_ma = close.rolling(100).mean()\n",
    "fast_ma = close.rolling(10).mean()\n",
    "\n",
    "long_signals = (fast_ma >= slow_ma)\n",
    "short_signals = (fast_ma < slow_ma)\n",
    "close = df['Close']\n",
    "\n",
    "events.loc[long_signals, 'side'] = 1\n",
    "events.loc[short_signals, 'side'] = -1\n",
    "events['side'] = events['side'].shift()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [],
   "source": [
    "bins = getBins(events, df['Close'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "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>t1</th>\n",
       "      <th>trgt</th>\n",
       "      <th>pt</th>\n",
       "      <th>sl</th>\n",
       "      <th>side</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2018-02-05 10:28:00</th>\n",
       "      <td>2018-02-05 13:02:00</td>\n",
       "      <td>0.010232</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-05 10:34:00</th>\n",
       "      <td>2018-02-05 13:09:00</td>\n",
       "      <td>0.010271</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>-1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-05 10:54:00</th>\n",
       "      <td>2018-02-05 13:02:00</td>\n",
       "      <td>0.010475</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>-1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-05 11:01:00</th>\n",
       "      <td>2018-02-05 13:05:00</td>\n",
       "      <td>0.010512</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>-1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-05 11:10:00</th>\n",
       "      <td>2018-02-05 13:14:00</td>\n",
       "      <td>0.010515</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>-1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                     t1      trgt  pt  sl  side\n",
       "2018-02-05 10:28:00 2018-02-05 13:02:00  0.010232   1   2   NaN\n",
       "2018-02-05 10:34:00 2018-02-05 13:09:00  0.010271   1   2  -1.0\n",
       "2018-02-05 10:54:00 2018-02-05 13:02:00  0.010475   1   2  -1.0\n",
       "2018-02-05 11:01:00 2018-02-05 13:05:00  0.010512   1   2  -1.0\n",
       "2018-02-05 11:10:00 2018-02-05 13:14:00  0.010515   1   2  -1.0"
      ]
     },
     "execution_count": 112,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "events.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    1302\n",
       "1     445\n",
       "Name: bin, dtype: int64"
      ]
     },
     "execution_count": 113,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bins['bin'].value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3.4b\n",
    "\n",
    "Develop a trend-following strategy based on a popular technical analysis statistic (e.g. MA Crossover). For each observation, the model suggests a side, but not a size of the bet.\n",
    "\n",
    "Train a random forest to decide whether to trade or not. Note: The decision is whether to trade or not, {0,1}, since the underlying model (the crossing moving average) has decided the side, {-1, 1}."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/doda/anaconda3/envs/zippy/lib/python3.5/site-packages/ipykernel_launcher.py:9: FutureWarning: Currently, 'apply' passes the values as ndarrays to the applied function. In the future, this will change to passing it as Series objects. You need to specify 'raw=True' to keep the current behaviour, and you can pass 'raw=False' to silence this warning\n",
      "  if __name__ == '__main__':\n",
      "/Users/doda/anaconda3/envs/zippy/lib/python3.5/site-packages/ipykernel_launcher.py:10: FutureWarning: Currently, 'apply' passes the values as ndarrays to the applied function. In the future, this will change to passing it as Series objects. You need to specify 'raw=True' to keep the current behaviour, and you can pass 'raw=False' to silence this warning\n",
      "  # Remove the CWD from sys.path while we load stuff.\n",
      "/Users/doda/anaconda3/envs/zippy/lib/python3.5/site-packages/ipykernel_launcher.py:11: FutureWarning: Currently, 'apply' passes the values as ndarrays to the applied function. In the future, this will change to passing it as Series objects. You need to specify 'raw=True' to keep the current behaviour, and you can pass 'raw=False' to silence this warning\n",
      "  # This is added back by InteractiveShellApp.init_path()\n"
     ]
    }
   ],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "rf = RandomForestClassifier(n_estimators=1000, random_state=42)\n",
    "\n",
    "# We'll use the features described in 3.5b (vol, serial corr., ma crosses)\n",
    "df['log_ret'] = np.log(close).diff()\n",
    "df['vol5'] = df['log_ret'].rolling(5).std()\n",
    "df['vol10'] = df['log_ret'].rolling(10).std()\n",
    "df['vol15'] = df['log_ret'].rolling(15).std()\n",
    "\n",
    "df['serialcorr20-1'] = df['log_ret'].rolling(20).apply(lambda x: pd.Series(x).autocorr(lag=1))\n",
    "df['serialcorr20-2'] = df['log_ret'].rolling(20).apply(lambda x: pd.Series(x).autocorr(lag=2))\n",
    "df['serialcorr20-3'] = df['log_ret'].rolling(20).apply(lambda x: pd.Series(x).autocorr(lag=3))\n",
    "\n",
    "df.loc[long_signals, 'side'] = 1\n",
    "df.loc[short_signals, 'side'] = -1\n",
    "df['side'] = df['side'].shift()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_features = df.loc[events.index][['side', 'vol5', 'vol10', 'vol15', 'serialcorr20-1', 'serialcorr20-2', 'serialcorr20-3']]\n",
    "train_labels = bins['bin']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',\n",
       "            max_depth=None, max_features='auto', max_leaf_nodes=None,\n",
       "            min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "            min_samples_leaf=1, min_samples_split=2,\n",
       "            min_weight_fraction_leaf=0.0, n_estimators=1000, n_jobs=None,\n",
       "            oob_score=False, random_state=42, verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 116,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rf.fit(train_features, train_labels)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3.5a\n",
    "\n",
    "Develop a mean-reverting strategy based on Bollinger Bands. For each observation, the model suggests a side, but not a size of the bet.\n",
    "\n",
    "Derive meta-labels for `ptSl = [0,2]` and `t1` where `numDays=1`. Use as `trgt` the daily standard deviation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "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>t1</th>\n",
       "      <th>trgt</th>\n",
       "      <th>pt</th>\n",
       "      <th>sl</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2018-02-05 10:28:00</th>\n",
       "      <td>2018-02-06 10:30:00</td>\n",
       "      <td>0.010232</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-05 10:34:00</th>\n",
       "      <td>2018-02-06 10:36:00</td>\n",
       "      <td>0.010271</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-05 10:54:00</th>\n",
       "      <td>2018-02-06 10:55:00</td>\n",
       "      <td>0.010475</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-05 11:01:00</th>\n",
       "      <td>2018-02-06 11:05:00</td>\n",
       "      <td>0.010512</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2018-02-05 11:10:00</th>\n",
       "      <td>2018-02-06 11:11:00</td>\n",
       "      <td>0.010515</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                     t1      trgt  pt  sl\n",
       "2018-02-05 10:28:00 2018-02-06 10:30:00  0.010232   0   2\n",
       "2018-02-05 10:34:00 2018-02-06 10:36:00  0.010271   0   2\n",
       "2018-02-05 10:54:00 2018-02-06 10:55:00  0.010475   0   2\n",
       "2018-02-05 11:01:00 2018-02-06 11:05:00  0.010512   0   2\n",
       "2018-02-05 11:10:00 2018-02-06 11:11:00  0.010515   0   2"
      ]
     },
     "execution_count": 117,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "events = getEvents(close, tEvents=tEvents, ptSl=[0,2], t1=t1, trgt=daily_vol, minRet=0.01)\n",
    "events.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3.5b\n",
    "\n",
    "Develop a mean-reverting strategy based on Bollinger Bands. For each observation, the model suggests a side, but not a size of the bet.\n",
    "\n",
    "Train a random forest to decide whether to trade or not. Use as features: volatility, serial correlation, and the crossing moving averages from exercise 2.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {},
   "outputs": [],
   "source": [
    "def bbands(close, window, stdev=2):\n",
    "    ma = close.rolling(window).mean()\n",
    "    std = close.rolling(window).std()\n",
    "    return (ma + stdev * std, ma - stdev * std)\n",
    "    \n",
    "df['bb_upper_band'], df['bb_lower_band'] = bbands(close, 100, 2)\n",
    "short_signals = close > df['bb_upper_band']\n",
    "long_signals = close < df['bb_lower_band']\n",
    "\n",
    "df['side'] = np.nan\n",
    "df.loc[long_signals, 'side'] = 1\n",
    "df.loc[short_signals, 'side'] = -1\n",
    "df['side'] = df['side'].ffill() # long until we're short\n",
    "df = df.shift() # No look-ahead"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "-1.0    14021\n",
       " 1.0    11564\n",
       "Name: side, dtype: int64"
      ]
     },
     "execution_count": 119,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.side.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 131,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',\n",
       "            max_depth=5, max_features='auto', max_leaf_nodes=None,\n",
       "            min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "            min_samples_leaf=1, min_samples_split=2,\n",
       "            min_weight_fraction_leaf=0.0, n_estimators=100, n_jobs=None,\n",
       "            oob_score=False, random_state=10, verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 131,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_features = df.loc[events.index][['side', 'vol5', 'vol10', 'vol15', 'serialcorr20-1', 'serialcorr20-2', 'serialcorr20-3']]\n",
    "train_labels = bins['bin']\n",
    "\n",
    "rf = RandomForestClassifier(n_estimators=100, random_state=10, max_depth=5)\n",
    "rf.fit(train_features, train_labels)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3.5c\n",
    "\n",
    "Develop a mean-reverting strategy based on Bollinger Bands. For each observation, the model suggests a side, but not a size of the bet.\n",
    "\n",
    "What is the accuracy of predictions from the primary model (i.e. if the secondary model does not filter the bets)? What are the precision, recall, and F-1 scores?\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.00      0.00      0.00      1302\n",
      "           1       0.25      1.00      0.41       445\n",
      "\n",
      "   micro avg       0.25      0.25      0.25      1747\n",
      "   macro avg       0.13      0.50      0.20      1747\n",
      "weighted avg       0.06      0.25      0.10      1747\n",
      "\n",
      "Confusion Matrix\n",
      "[[   0 1302]\n",
      " [   0  445]]\n",
      "\n",
      "Accuracy\n",
      "0.25472238122495705\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/doda/anaconda3/envs/zippy/lib/python3.5/site-packages/sklearn/metrics/classification.py:1143: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import roc_curve\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "primary_forecast = pd.DataFrame(bins['bin'])\n",
    "primary_forecast['pred'] = 1\n",
    "primary_forecast.columns = ['actual', 'pred']\n",
    "\n",
    "# Performance Metrics\n",
    "actual = primary_forecast['actual']\n",
    "pred = primary_forecast['pred']\n",
    "print(classification_report(y_true=actual, y_pred=pred))\n",
    "\n",
    "print(\"Confusion Matrix\")\n",
    "print(confusion_matrix(actual, pred))\n",
    "\n",
    "print('')\n",
    "print(\"Accuracy\")\n",
    "print(accuracy_score(actual, pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3.5d\n",
    "\n",
    "Develop a mean-reverting strategy based on Bollinger Bands. For each observation, the model suggests a side, but not a size of the bet.\n",
    "\n",
    "What is the accuracy of prediction from the secondary model? What are the precision, recall, and F1-scores?\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 133,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.80      0.99      0.89      1302\n",
      "           1       0.88      0.29      0.44       445\n",
      "\n",
      "   micro avg       0.81      0.81      0.81      1747\n",
      "   macro avg       0.84      0.64      0.66      1747\n",
      "weighted avg       0.82      0.81      0.77      1747\n",
      "\n",
      "Confusion Matrix\n",
      "[[1284   18]\n",
      " [ 315  130]]\n",
      "\n",
      "Accuracy\n",
      "0.8093875214653692\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x432 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_train = bins['bin']\n",
    "X_train = train_features\n",
    "# Performance Metrics\n",
    "y_pred_rf = rf.predict_proba(X_train)[:, 1]\n",
    "y_pred = rf.predict(X_train)\n",
    "fpr_rf, tpr_rf, _ = roc_curve(y_train, y_pred_rf)\n",
    "print(classification_report(y_train, y_pred))\n",
    "\n",
    "print(\"Confusion Matrix\")\n",
    "print(confusion_matrix(y_train, y_pred))\n",
    "\n",
    "print('')\n",
    "print(\"Accuracy\")\n",
    "print(accuracy_score(y_train, y_pred))\n",
    "\n",
    "plt.figure(1)\n",
    "plt.plot([0, 1], [0, 1], 'k--')\n",
    "plt.plot(fpr_rf, tpr_rf, label='RF')\n",
    "plt.xlabel('False positive rate')\n",
    "plt.ylabel('True positive rate')\n",
    "plt.title('ROC curve')\n",
    "plt.legend(loc='best')\n",
    "plt.show()\n"
   ]
  }
 ],
 "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": "3.5.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}