{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "7e087d05-8fef-4c19-87ce-f8eba57f4cff",
   "metadata": {},
   "source": [
    "# Regime filters\n",
    "\n",
    "A regime filter is a classification to indicate the market condition we are in. They could provide some useful pointers to improve trading results. Though we cannot predict the market movements with certainty, we can atleast get an edge.\n",
    "\n",
    "We should keep these filters simple so that it is easy to understand and do not overfit too much to available data. Boolean filters are a good choice since they only have a true or false value and not prone to too much overfitting, though they may only have a slight edge."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f87a5d74-bb4b-4143-9b66-4edb5fe340f7",
   "metadata": {},
   "source": [
    "## A simple price filter\n",
    "\n",
    "Let us a simple price filter.\n",
    "\n",
    "1. Create a moving average of the price for the last 60 days.\n",
    "2. If the close price is greater than the average price, give it a value of 1.\n",
    "3. If the close price is less than the average price, give it a value of 0\n",
    "    \n",
    "We shift the average price by 1 day so that we do not include today's close price \n",
    "\n",
    "<!-- TEASER_END -->"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "284271ac-a02b-470b-b16f-4b4a4b17c854",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "sns.set()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "4ed9a13b-be52-4eba-b841-7d307166b1e3",
   "metadata": {
    "tags": [
     "parameters"
    ]
   },
   "outputs": [],
   "source": [
    "# parameters\n",
    "ma = 60"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "ff4d2846-03e6-43de-8417-9badc0027b8a",
   "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>open</th>\n",
       "      <th>high</th>\n",
       "      <th>low</th>\n",
       "      <th>close</th>\n",
       "      <th>volume</th>\n",
       "      <th>adj close</th>\n",
       "      <th>ret</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>date</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2021-11-24</th>\n",
       "      <td>4675.779785</td>\n",
       "      <td>4702.870117</td>\n",
       "      <td>4659.890137</td>\n",
       "      <td>4701.459961</td>\n",
       "      <td>2464040000</td>\n",
       "      <td>4701.459961</td>\n",
       "      <td>0.002294</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2021-11-26</th>\n",
       "      <td>4664.629883</td>\n",
       "      <td>4664.629883</td>\n",
       "      <td>4585.430176</td>\n",
       "      <td>4594.620117</td>\n",
       "      <td>2676740000</td>\n",
       "      <td>4594.620117</td>\n",
       "      <td>-0.022725</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2021-11-29</th>\n",
       "      <td>4628.750000</td>\n",
       "      <td>4672.950195</td>\n",
       "      <td>4625.259766</td>\n",
       "      <td>4655.270020</td>\n",
       "      <td>3471380000</td>\n",
       "      <td>4655.270020</td>\n",
       "      <td>0.013200</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2021-11-30</th>\n",
       "      <td>4640.250000</td>\n",
       "      <td>4646.020020</td>\n",
       "      <td>4560.000000</td>\n",
       "      <td>4567.000000</td>\n",
       "      <td>4950190000</td>\n",
       "      <td>4567.000000</td>\n",
       "      <td>-0.018961</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2021-12-01</th>\n",
       "      <td>4602.819824</td>\n",
       "      <td>4652.939941</td>\n",
       "      <td>4510.270020</td>\n",
       "      <td>4513.040039</td>\n",
       "      <td>4078260000</td>\n",
       "      <td>4513.040039</td>\n",
       "      <td>-0.011815</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                   open         high          low        close      volume  \\\n",
       "date                                                                         \n",
       "2021-11-24  4675.779785  4702.870117  4659.890137  4701.459961  2464040000   \n",
       "2021-11-26  4664.629883  4664.629883  4585.430176  4594.620117  2676740000   \n",
       "2021-11-29  4628.750000  4672.950195  4625.259766  4655.270020  3471380000   \n",
       "2021-11-30  4640.250000  4646.020020  4560.000000  4567.000000  4950190000   \n",
       "2021-12-01  4602.819824  4652.939941  4510.270020  4513.040039  4078260000   \n",
       "\n",
       "              adj close       ret  \n",
       "date                               \n",
       "2021-11-24  4701.459961  0.002294  \n",
       "2021-11-26  4594.620117 -0.022725  \n",
       "2021-11-29  4655.270020  0.013200  \n",
       "2021-11-30  4567.000000 -0.018961  \n",
       "2021-12-01  4513.040039 -0.011815  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv('/home/pi/data/sp500.csv', parse_dates=['Date']).rename(\n",
    "columns = lambda x:x.lower()).sort_values(by='date').set_index('date')\n",
    "df['ret'] = df.close.pct_change()\n",
    "df.tail()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "42aa76b8-165d-4910-88e3-a417ff0bbe31",
   "metadata": {},
   "outputs": [],
   "source": [
    "df['ma_price'] = df.close.rolling(ma).mean().shift(1)\n",
    "df['is_price'] = df.eval('close > ma_price')+0\n",
    "df['is_price'] = df.is_price.shift(1) # Shifting price since we use the signal only next day"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "b59e3a27-eba4-4b1a-ba8a-3066f7bf64da",
   "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>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>is_price</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0.0</th>\n",
       "      <td>1889.0</td>\n",
       "      <td>0.000423</td>\n",
       "      <td>0.017703</td>\n",
       "      <td>-0.119841</td>\n",
       "      <td>-0.008244</td>\n",
       "      <td>0.000956</td>\n",
       "      <td>0.009203</td>\n",
       "      <td>0.115800</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1.0</th>\n",
       "      <td>3626.0</td>\n",
       "      <td>0.000206</td>\n",
       "      <td>0.008350</td>\n",
       "      <td>-0.058944</td>\n",
       "      <td>-0.003560</td>\n",
       "      <td>0.000547</td>\n",
       "      <td>0.004686</td>\n",
       "      <td>0.038053</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           count      mean       std       min       25%       50%       75%  \\\n",
       "is_price                                                                       \n",
       "0.0       1889.0  0.000423  0.017703 -0.119841 -0.008244  0.000956  0.009203   \n",
       "1.0       3626.0  0.000206  0.008350 -0.058944 -0.003560  0.000547  0.004686   \n",
       "\n",
       "               max  \n",
       "is_price            \n",
       "0.0       0.115800  \n",
       "1.0       0.038053  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.groupby(['is_price']).ret.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "d75489d6-8da2-4ad5-b3fe-dcacd65c411b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<seaborn.axisgrid.FacetGrid at 0x7f553f16d0>"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 414.975x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.displot(data=df, x='ret', hue='is_price', kind='kde')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4d1b5f6b-0ed8-48b5-8865-d46310d7fbe1",
   "metadata": {},
   "source": [
    "Looks there seem to be an edge when the filter is negative where the returns are a bit more.\n",
    "\n",
    "~~In the earlier version, I computed the end of the day returns which is obvious and just a explanation of events already happened. Forward returns should always be considered to find a predictive edge.~~\n",
    "\n",
    "\n",
    "<div class=\"alert alert-warning\">\n",
    "    This doesn't seem like a tradeable edge since the margin is very small\n",
    "    </div>\n",
    "\n",
    "Let us see whether the edge persists through all years.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "b8ab2ef6-e6a4-4fe2-b125-547ae6fab284",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:title={'center':'Mean returns by year'}, xlabel='year'>"
      ]
     },
     "execution_count": 7,
     "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": [
    "df['year'] = df.index.year\n",
    "df.groupby(['year', 'is_price']).ret.mean().unstack().plot.bar(stacked=True, \n",
    "                                                               title=\"Mean returns by year\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bed195a0-a222-420e-bb04-b948932d4666",
   "metadata": {},
   "source": [
    "The edge fairly persists over the years.\n",
    "\n",
    "Let us try changing it to median and see if it improves the results\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "15ba70cc-0f1b-4eaa-a25a-bd0318a955d4",
   "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>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>is_price</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0.0</th>\n",
       "      <td>1950.0</td>\n",
       "      <td>0.000485</td>\n",
       "      <td>0.017500</td>\n",
       "      <td>-0.119841</td>\n",
       "      <td>-0.008144</td>\n",
       "      <td>0.000953</td>\n",
       "      <td>0.009326</td>\n",
       "      <td>0.115800</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1.0</th>\n",
       "      <td>3565.0</td>\n",
       "      <td>0.000168</td>\n",
       "      <td>0.008332</td>\n",
       "      <td>-0.058944</td>\n",
       "      <td>-0.003545</td>\n",
       "      <td>0.000546</td>\n",
       "      <td>0.004628</td>\n",
       "      <td>0.038053</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           count      mean       std       min       25%       50%       75%  \\\n",
       "is_price                                                                       \n",
       "0.0       1950.0  0.000485  0.017500 -0.119841 -0.008144  0.000953  0.009326   \n",
       "1.0       3565.0  0.000168  0.008332 -0.058944 -0.003545  0.000546  0.004628   \n",
       "\n",
       "               max  \n",
       "is_price            \n",
       "0.0       0.115800  \n",
       "1.0       0.038053  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['ma_price'] = df.close.rolling(ma).median().shift(1)\n",
    "df['is_price'] = df.eval('close > ma_price')+0\n",
    "df['is_price'] = df.is_price.shift(1) # Shifting price since we use the signal only next day\n",
    "df.groupby(['is_price']).ret.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a7c8fea6-9720-401c-80a7-54162b68a6a8",
   "metadata": {},
   "source": [
    "Not so big. This is just a simple way to create a filter to define a market environment and see how it impacts the returns. The following could be done to improve a filter\n",
    "\n",
    " * Use quantiles instead of mean or median\n",
    " * Calculate the filter for different moving averages\n",
    " * Try simulating for periods instead of rolling over averages"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.2"
  },
  "nikola": {
   "category": "regime-change",
   "date": "2021-11-25 08:50:05 UTC",
   "slug": "simple-regime-filter",
   "tags": "filters,regimes",
   "title": "A simple regime filter"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}