alpha/strategies/strategy_3.ipynb

{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": 40,
      "metadata": {
        "pycharm": {
          "is_executing": false,
          "metadata": false,
          "name": "#%% Trade Strategy 3: Follow the upward trend.\n"
        }
      },
      "outputs": [],
      "source": "# %%file strategy_3.py\n\nimport import_ipynb\nfrom strategy_base import TradeStrategyBase\n\nimport numpy as np\nimport pandas as pd\nimport matplotlib.pyplot as plt\n\nclass TradeStrategy3(TradeStrategyBase):\n    \"\"\"\n    Trade Strategy 3: Mean Reversion.\n    Overview:\n    According to probability density distribution (Assuming p_change follows normal distribution). So: \n    Pr(µ - 1σ \u003c\u003d X \u003c\u003d µ + 1σ) ~\u003d 0.6827\n    Pr(µ - 2σ \u003c\u003d X \u003c\u003d µ + 2σ) ~\u003d 0.9545\n    Pr(µ - 3σ \u003c\u003d X \u003c\u003d µ + 3σ) ~\u003d 0.9973\n    This should hold true for µ moving_average_x_days and corresponding σ moving_std_x_days. \n    Therefore, we can use a __std_mutiplier to control probability distribution, and use the following strategy.\n    \n    Buy when price \u003e\u003d moving_average_x_days + __std_mutiplier * moving_std_xdays \u0026 Sell when price reverted to moving_average\n    Sell when price \u003c\u003d moving_average_x_days - __std_mutiplier * moving_std_xdays \u0026 Buy when price reverted to moving_average\n    Assumptions: \n    moving_average_x_days and moving_std_x_days follows normal distribution (To be tested)\n    Requirement: \n    1. This strategy should work with any highly liquid security. Preferably Forex or Crypto.\n    2. This strategy should work whether trend is up or down, volatile or not. \n    3. The more volatile, the better, as we need to quickly close the position after mean reversion, BEFORE moving_average moved too much. \n    Constants: \n    Params:\n    _std_mutiplier: The multiplier for moving standard deviation.\n    _x_days: The number of days to calculate moving_average \u0026 moving_std. \n    \"\"\"\n    # The multiplier for moving standard deviation\n    _std_mutiplier \u003d 2.0\n    # The number of days to calculate moving_average \u0026 moving_std.\n    _x_days \u003d 5\n    # Position Decision. E.g. LONG: 1, SHORT: -1, NONE: 0\n    _position \u003d 0\n    \n    def __init__(self, x_days:int \u003d 5, std_mutiplier:float \u003d 2.0, should_log:bool \u003d False, should_plot:bool \u003d False, axs \u003d None):\n        \"\"\"\n        Initialize Strategy with given parameters\n        :type x_days: int\n        :type std_mutiplier: float \n        :type should_log: bool\n        :type should_plot: bool\n        :param x_days: The number of days to calculate moving_average \u0026 moving_std. E.g. 5 \u003d 5 days.\n        :param std_mutiplier: The multiplier for moving standard deviation. E.g. 2.0 means µ +- 2σ \n        :param should_log Whether or not log the strategy\n        :param should_plot Whether or not plot the strategy\n        \"\"\"\n        self._std_mutiplier \u003d std_mutiplier\n        self._x_days \u003d x_days\n        self._should_log \u003d should_log\n        self._should_plot \u003d should_plot\n        self._axs \u003d axs\n        \n        self._position \u003d 0\n        self._long_tradedays \u003d  pd.DataFrame()\n        self._short_tradedays \u003d pd.DataFrame()\n        \n    def trade(self, date:pd.Timestamp, tradeday: pd.Series):\n        \"\"\"\n        Perform Analysis, Decide on Trade, and Execute Trade\n        :type date: pd.Timestamp\n        :type tradeday: pd.Series\n        :param date: The pandas DataFrame date\n        :param tradeday: The pandas DateFrame Series representing a trade day\n        :return: \n        \"\"\"\n        \n        \"\"\"\n        Validate Data, Check tradeday has column \u0027ma_x\u0027 \u0026 \u0027mov_std_x\u0027 \u0026 \u0027ewm_std_x\u0027.\n        \"\"\"\n        if not hasattr(tradeday, \u0027ma\u0027):\n            # Data Incomplete, do not trade\n            return\n            \n        \"\"\"\n        Firstly, add tradeday to position DataFrame\n        \"\"\"\n        if self._position \u003d\u003d 1:\n            # Add Long days to _long_tradedays\n            self._long_tradedays \u003d self._long_tradedays.append(tradeday)\n        elif self._position \u003d\u003d -1:\n            # Add Short days to _short_tradedays\n            self._short_tradedays \u003d self._long_tradedays.append(tradeday)\n        \n        \"\"\"\n        Secondly, Closing\n        \"\"\"\n        if self._evaluate_closing(date\u003ddate,\n                                 tradeday\u003dtradeday):\n            self._execute_closing(date\u003ddate,\n                                 tradeday\u003dtradeday)\n        \n        \"\"\"\n        Thirdly, Buying\n        \"\"\"\n        if self._evaluate_buying(date\u003ddate,\n                                 tradeday\u003dtradeday):\n            self._execute_buying(date\u003ddate,\n                                 tradeday\u003dtradeday)\n        \n        \"\"\"\n        Finally, Selling\n        \"\"\"\n        if self._evaluate_selling(date\u003ddate,\n                                  tradeday\u003dtradeday):\n            self._execute_selling(date\u003ddate,\n                                  tradeday\u003dtradeday)\n        \n        \n    def _evaluate_closing(self, date:pd.Timestamp, tradeday: pd.Series):\n        \"\"\"\n        Evaluate whether or not BUY should be executed\n        :type date: pd.Timestamp\n        :type tradeday: pd.Series\n        :param date: The pandas DataFrame date\n        :param tradeday: The pandas DateFrame Series representing a trade day\n        \"\"\"\n        ma \u003d tradeday[\"ma\"]\n        mov_std \u003d tradeday[\"mov_std\"]\n        emv_std \u003d tradeday[\"ewm_std\"]\n        \n        if self._position \u003d\u003d -1:\n            # Currently have short position\n            close_threshold \u003d ma # Close threshold is set to ma_x\n            if tradeday.close \u003e\u003d close_threshold:\n                # Price reverted to above close threshold\n                return True\n        elif self._position \u003d\u003d 1:\n            # Currently have long position\n            close_threshold \u003d ma # Close threshold is set to ma_x\n            if tradeday.close \u003c\u003d close_threshold:\n                # Price reverted to below close threshold\n                return True\n        return False\n        \n    def _execute_closing(self, date:pd.Timestamp, tradeday: pd.Series):\n        \"\"\"\n        How BUY should be executed\n        :type date: pd.Timestamp\n        :type tradeday: pd.Series\n        :param date: The pandas DataFrame date\n        :param tradeday: The pandas DateFrame Series representing a trade day\n        \"\"\"\n        if self._should_log:\n            print(\"- Closed on {} at ${}.\".format(date, tradeday.close))\n        self._position \u003d 0\n        \n        \n    def _evaluate_buying(self, date:pd.Timestamp, tradeday: pd.Series):\n        \"\"\"\n        Evaluate whether or not BUY should be executed\n        :type date: pd.Timestamp\n        :type tradeday: pd.Series\n        :param date: The pandas DataFrame date\n        :param tradeday: The pandas DateFrame Series representing a trade day\n        \"\"\"\n        ma \u003d tradeday[\"ma\"]\n        mov_std \u003d tradeday[\"mov_std\"]\n        emv_std \u003d tradeday[\"ewm_std\"]\n        buy_threshold \u003d ma - self._std_mutiplier * mov_std\n        # print(\"Close {} \u003e {} ({} - {} * {})\".format(tradeday.close, buy_threshold, ma, self._std_mutiplier, mov_std))\n        \n        if tradeday.open \u003c buy_threshold:\n            # When BUY agrees\n            if self._position \u003d\u003d 0:\n                # AND no current position\n                return True\n        return False\n        \n    def _execute_buying(self, date:pd.Timestamp, tradeday: pd.Series):\n        \"\"\"\n        How BUY should be executed\n        :type date: pd.Timestamp\n        :type tradeday: pd.Series\n        :param date: The pandas DataFrame date\n        :param tradeday: The pandas DateFrame Series representing a trade day\n        \"\"\"\n        if self._should_log:\n            print(\"Bought on {} at ${}.\".format(date, tradeday.close))\n        self._long_tradedays \u003d self._long_tradedays.append(tradeday)\n        self._position \u003d 1\n\n    def _evaluate_selling(self, date:pd.Timestamp, tradeday: pd.Series):\n        \"\"\"\n        Evaluate whether or not SELL should be executed\n        :type date: pd.Timestamp\n        :type tradeday: pd.Series\n        :param date: The pandas DataFrame date\n        :param tradeday: The pandas DateFrame Series representing a trade day\n        \"\"\"\n        ma \u003d tradeday[\"ma\"]\n        mov_std \u003d tradeday[\"mov_std\"]\n        emv_std \u003d tradeday[\"ewm_std\"]\n        sell_threshold \u003d ma + self._std_mutiplier * mov_std\n        \n        if tradeday.open \u003e sell_threshold:\n            # When SELL agrees\n            if self._position \u003d\u003d 0:\n                # AND no current position\n                return True\n        return False\n            \n    def _execute_selling(self, date:pd.Timestamp, tradeday: pd.Series):\n        \"\"\"\n        How SELL should be executed\n        :type date: pd.Timestamp\n        :type tradeday: pd.Series\n        :param date: The pandas DataFrame date\n        :param tradeday: The pandas DateFrame Series representing a trade day\n        \"\"\"\n        if self._should_log:\n            print(\"Sold on {} at ${}.\".format(date, tradeday.close))\n        if self._should_plot:\n            self._plot_trade(axs\u003dself._axs,\n                             buy_date\u003dself._last_buy_date, \n                             sell_date\u003ddate)\n        self._short_tradedays \u003d self._short_tradedays.append(tradeday)\n        self._position \u003d -1\n\n    def _plot_trade(self, axs, buy_date, sell_date):\n\n        print(\"Todo: Plotting {} {}\".format(buy_date, sell_date))\n\n        # buy_tradeday \u003d self._held_tradedays[buy_date.date()]\n        # sell_tradeday \u003d self._held_tradedays[sell_date.date()]\n        # print(\"Bought at {}, Sold at {}\".format(buy_tradeday.close, sell_tradeday.close))\n\n        # drawer \u003d plt if axs is None else axs\n        # \n        # # Mark red for up and green for down, slicing using start and end\n        # if sell_tradeday.close \u003c buy_tradeday.close:\n        #   drawer.fill_between(self._held_tradedays.index[buy_date:sell_date], 0,\n        #                   self._held_tradedays[\u0027close\u0027][buy_date:sell_date], color\u003d\u0027red\u0027,\n        #                   alpha\u003d.38)\n        #   is_win \u003d False\n        # else:\n        #   drawer.fill_between(amex_df.index[buy_date:sell_date], 0,\n        #                   amex_df[\u0027close\u0027][buy_date:sell_date], color\u003d\u0027green\u0027,\n        #                   alpha\u003d.38)\n        #   is_win \u003d True\n        # \n        # return is_win\n        \n    @property\n    def trade_profit(self):\n        long_profit \u003d 0.0\n        if \u0027p_change\u0027 in self._long_tradedays.columns:\n            profit_array \u003d self._long_tradedays.p_change/100 + 1\n            long_profit \u003d np.product(profit_array) - 1\n        short_profit \u003d 0.0\n        if \u0027p_change\u0027 in self._short_tradedays.columns:\n            profit_array \u003d self._short_tradedays.p_change/100 - 1\n            short_profit \u003d np.abs(np.product(profit_array)) - 1\n        return long_profit + short_profit\n    \n    @property\n    def p_change_threshold(self):\n        return self._p_change_threshold\n    \n    @property\n    def keep_stock_threshold(self):\n        return self._hold_stock_threshold\n   "
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "outputs": [],
      "source": "",
      "metadata": {
        "pycharm": {
          "metadata": false,
          "name": "#%%\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": "3.7.3"
    },
    "stem_cell": {
      "cell_type": "raw",
      "metadata": {
        "pycharm": {
          "metadata": false
        }
      },
      "source": ""
    }
  },
  "nbformat": 4,
  "nbformat_minor": 1
}