GPlearn Zhengzhou Commodity Exchange PTA Futures  Alpha Factor(1 Minute Period Data).ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# GPlearn 1 Minute  Period Data Alpha Factor Generating\n",
    "## What is Alpha Factor?\n",
    "Alpha (α) is a term used in investing to describe an investment strategy's ability to beat the market, or its \"edge.\" Alpha is thus also often referred to as “excess return” or “abnormal rate of return,” which refers to the idea that markets are efficient, and so there is no way to systematically earn returns that exceed the broad market as a whole. Alpha is often used in conjunction with beta (the Greek letter β), which measures the broad market's overall volatility or risk, known as systematic market risk.  \n",
    "\n",
    "Alpha is used in finance as a measure of performance, indicating when a strategy, trader, or portfolio manager has managed to beat the market return over some period. Alpha, often considered the active return on an investment, gauges the performance of an investment against a market index or benchmark that is considered to represent the market’s movement as a whole.  \n",
    "\n",
    "The excess return of an investment relative to the return of a benchmark index is the investment’s alpha. Alpha may be positive or negative and is the result of active investing. Beta, on the other hand, can be earned through passive index investing.(Retrived from https://www.investopedia.com/terms/a/alpha.asp)  \n",
    "\n",
    "Alpha Factor is a mathmatical expression that tries to identify the strategy that can provide information on the future trend of the market. Alpha factor examples can be found on https://www.spglobal.com/marketintelligence/en/solutions/alpha-factor-library.  \n",
    "## What is GPlearn and symbolic transformer?\n",
    "gplearn implements Genetic Programming in Python, with a scikit-learn inspired and compatible API.\n",
    "\n",
    "While Genetic Programming (GP) can be used to perform a very wide variety of tasks, gplearn is purposefully constrained to solving symbolic regression problems. This is motivated by the scikit-learn ethos, of having powerful estimators that are straight-forward to implement.Genetic programming is capable of taking a series of totally random programs, untrained and unaware of any given target function you might have had in mind, and making them breed, mutate and evolve their way towards the truth.\n",
    "\n",
    "Think of genetic programming as a stochastic optimization process. Every time an initial population is conceived, and with every selection and evolution step in the process, random individuals from the current generation are selected to undergo random changes in order to enter the next. You can control this randomness by using the random_state parameter of the estimator.  \n",
    "\n",
    "A symbolic transformer is a supervised transformer that begins by building a population of naive random formulas to represent a relationship. The formulas are represented as tree-like structures with mathematical functions being recursively applied to variables and constants. Each successive generation of programs is then evolved from the one that came before it by selecting the fittest individuals from the population to undergo genetic operations such as crossover, mutation or reproduction. The final population is searched for the fittest individuals with the least correlation to one another."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T02:54:58.003848Z",
     "start_time": "2021-09-26T02:54:55.533777Z"
    }
   },
   "outputs": [],
   "source": [
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import graphviz\n",
    "\n",
    "from scipy.stats import rankdata\n",
    "import scipy.stats as stats\n",
    "\n",
    "from gplearn import genetic\n",
    "from gplearn.functions import make_function\n",
    "from gplearn.genetic import SymbolicTransformer, SymbolicRegressor\n",
    "from gplearn.fitness import make_fitness\n",
    "\n",
    "from sklearn.utils import check_random_state\n",
    "from sklearn.model_selection import train_test_split\n",
    "import baostock as bs\n",
    "from ta.volume import VolumeWeightedAveragePrice\n",
    "import statsmodels.api as sm\n",
    "from scipy.stats.mstats import zscore\n",
    "import talib\n",
    "\n",
    "from scipy.stats import spearmanr\n",
    "import datetime\n",
    "from numba import jit\n",
    "import scipy.stats as stats\n",
    "from sklearn.linear_model import LinearRegression\n",
    "import math\n",
    "import cloudpickle\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Import and Prepare Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T02:54:58.544797Z",
     "start_time": "2021-09-26T02:54:58.050798Z"
    }
   },
   "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>date</th>\n",
       "      <th>open</th>\n",
       "      <th>close</th>\n",
       "      <th>high</th>\n",
       "      <th>low</th>\n",
       "      <th>volume</th>\n",
       "      <th>money</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2014-03-21 09:01:00</td>\n",
       "      <td>6278.0</td>\n",
       "      <td>6272.0</td>\n",
       "      <td>6280.0</td>\n",
       "      <td>6262.0</td>\n",
       "      <td>8086.0</td>\n",
       "      <td>253796760.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2014-03-21 09:02:00</td>\n",
       "      <td>6274.0</td>\n",
       "      <td>6266.0</td>\n",
       "      <td>6278.0</td>\n",
       "      <td>6266.0</td>\n",
       "      <td>2978.0</td>\n",
       "      <td>93449640.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2014-03-21 09:03:00</td>\n",
       "      <td>6266.0</td>\n",
       "      <td>6266.0</td>\n",
       "      <td>6270.0</td>\n",
       "      <td>6262.0</td>\n",
       "      <td>6364.0</td>\n",
       "      <td>199612500.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  date    open   close    high     low  volume        money\n",
       "0  2014-03-21 09:01:00  6278.0  6272.0  6280.0  6262.0  8086.0  253796760.0\n",
       "1  2014-03-21 09:02:00  6274.0  6266.0  6278.0  6266.0  2978.0   93449640.0\n",
       "2  2014-03-21 09:03:00  6266.0  6266.0  6270.0  6262.0  6364.0  199612500.0"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_df = pd.read_csv('dataset/PTA1分钟历史数据.csv')\n",
    "data_df = data_df.rename(columns = {\"Unnamed: 0\": \"date\"})\n",
    "\n",
    "data_df.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T02:54:58.654805Z",
     "start_time": "2021-09-26T02:54:58.592797Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(                       date    open   close    high     low  volume  \\\n",
       " 633657  2021-09-24 22:58:00  5206.0  5210.0  5210.0  5206.0  8295.0   \n",
       " 633658  2021-09-24 22:59:00  5210.0  5194.0  5212.0  5194.0  8423.0   \n",
       " 633659  2021-09-24 23:00:00  5194.0  5202.0  5204.0  5192.0  6532.0   \n",
       " \n",
       "               money  \n",
       " 633657  214425750.0  \n",
       " 633658  217734550.0  \n",
       " 633659  168852200.0  ,\n",
       "                   date    open   close    high     low  volume        money\n",
       " 0  2014-03-21 09:01:00  6278.0  6272.0  6280.0  6262.0  8086.0  253796760.0\n",
       " 1  2014-03-21 09:02:00  6274.0  6266.0  6278.0  6266.0  2978.0   93449640.0\n",
       " 2  2014-03-21 09:03:00  6266.0  6266.0  6270.0  6262.0  6364.0  199612500.0,\n",
       " date      False\n",
       " open      False\n",
       " close     False\n",
       " high      False\n",
       " low       False\n",
       " volume    False\n",
       " money     False\n",
       " dtype: bool)"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fields = ['open','close','high','low','volume','money']\n",
    "length = []\n",
    "\n",
    "df = data_df.copy()\n",
    "df.high = df.high.values.astype('float64')\n",
    "df.low = df.low.values.astype('float64')\n",
    "df.close = df.close.values.astype('float64')\n",
    "df.open = df.open.values.astype('float64')\n",
    "df.volume = df.volume.values.astype('float64')\n",
    "df.amount = df.money.values.astype('float64')\n",
    "\n",
    "df.tail(3), df.head(3), df.isnull().any()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Calculate The 1 Minute Rate of Return for Backtest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T02:54:58.808797Z",
     "start_time": "2021-09-26T02:54:58.701797Z"
    }
   },
   "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>date</th>\n",
       "      <th>open</th>\n",
       "      <th>close</th>\n",
       "      <th>high</th>\n",
       "      <th>low</th>\n",
       "      <th>volume</th>\n",
       "      <th>money</th>\n",
       "      <th>1_min_return</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>633656</th>\n",
       "      <td>2021-09-24 22:57:00</td>\n",
       "      <td>5202.0</td>\n",
       "      <td>5206.0</td>\n",
       "      <td>5208.0</td>\n",
       "      <td>5200.0</td>\n",
       "      <td>5257.0</td>\n",
       "      <td>135893450.0</td>\n",
       "      <td>0.000769</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>633657</th>\n",
       "      <td>2021-09-24 22:58:00</td>\n",
       "      <td>5206.0</td>\n",
       "      <td>5210.0</td>\n",
       "      <td>5210.0</td>\n",
       "      <td>5206.0</td>\n",
       "      <td>8295.0</td>\n",
       "      <td>214425750.0</td>\n",
       "      <td>0.000768</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>633658</th>\n",
       "      <td>2021-09-24 22:59:00</td>\n",
       "      <td>5210.0</td>\n",
       "      <td>5194.0</td>\n",
       "      <td>5212.0</td>\n",
       "      <td>5194.0</td>\n",
       "      <td>8423.0</td>\n",
       "      <td>217734550.0</td>\n",
       "      <td>-0.003071</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                       date    open   close    high     low  volume  \\\n",
       "633656  2021-09-24 22:57:00  5202.0  5206.0  5208.0  5200.0  5257.0   \n",
       "633657  2021-09-24 22:58:00  5206.0  5210.0  5210.0  5206.0  8295.0   \n",
       "633658  2021-09-24 22:59:00  5210.0  5194.0  5212.0  5194.0  8423.0   \n",
       "\n",
       "              money  1_min_return  \n",
       "633656  135893450.0      0.000769  \n",
       "633657  214425750.0      0.000768  \n",
       "633658  217734550.0     -0.003071  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['1_min_return'] = df.open.pct_change(1).shift(-1)\n",
    "df = df.dropna()\n",
    "\n",
    "df.tail(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Split Training and Testing Datase# 分割训练和测试数据"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T02:54:58.933797Z",
     "start_time": "2021-09-26T02:54:58.856798Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(                  date    open   close    high     low  volume        money  \\\n",
       " 0  2014-03-21 09:01:00  6278.0  6272.0  6280.0  6262.0  8086.0  253796760.0   \n",
       " 1  2014-03-21 09:02:00  6274.0  6266.0  6278.0  6266.0  2978.0   93449640.0   \n",
       " 2  2014-03-21 09:03:00  6266.0  6266.0  6270.0  6262.0  6364.0  199612500.0   \n",
       " \n",
       "    1_min_return  \n",
       " 0     -0.000637  \n",
       " 1     -0.001275  \n",
       " 2      0.000000  ,\n",
       "                        date    open   close    high     low  volume  \\\n",
       " 537552  2020-07-31 22:58:00  3574.0  3578.0  3578.0  3574.0  1410.0   \n",
       " 537553  2020-07-31 22:59:00  3576.0  3578.0  3580.0  3576.0  3611.0   \n",
       " 537554  2020-07-31 23:00:00  3578.0  3576.0  3580.0  3576.0  3609.0   \n",
       " \n",
       "              money  1_min_return  \n",
       " 537552  25281300.0      0.000560  \n",
       " 537553  64745230.0      0.000559  \n",
       " 537554  64709370.0      0.000559  ,\n",
       " (537555, 8))"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data=df[(df.date<'2020-08-01')]\n",
    "train_data=train_data.reset_index(drop=True)\n",
    "\n",
    "train_data.head(3), train_data.tail(3), train_data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T02:54:59.308305Z",
     "start_time": "2021-09-26T02:54:59.263305Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(                  date    open   close    high     low  volume        money  \\\n",
       " 0  2020-08-03 09:01:00  3580.0  3582.0  3588.0  3578.0  7838.0  140535340.0   \n",
       " 1  2020-08-03 09:02:00  3582.0  3578.0  3584.0  3576.0  6935.0  124344550.0   \n",
       " 2  2020-08-03 09:03:00  3578.0  3576.0  3580.0  3572.0  4178.0   74869760.0   \n",
       " \n",
       "    1_min_return  \n",
       " 0      0.000559  \n",
       " 1     -0.001117  \n",
       " 2     -0.000559  ,\n",
       " (96104, 8))"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_data=df[(df.date>='2020-08-01')]\n",
    "test_data=test_data.reset_index(drop=True)\n",
    "\n",
    "test_data.head(3), test_data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T02:54:59.805812Z",
     "start_time": "2021-09-26T02:54:59.784814Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([[6.2780000e+03, 6.2720000e+03, 6.2800000e+03, 6.2620000e+03,\n",
       "         8.0860000e+03, 2.5379676e+08],\n",
       "        [6.2740000e+03, 6.2660000e+03, 6.2780000e+03, 6.2660000e+03,\n",
       "         2.9780000e+03, 9.3449640e+07],\n",
       "        [6.2660000e+03, 6.2660000e+03, 6.2700000e+03, 6.2620000e+03,\n",
       "         6.3640000e+03, 1.9961250e+08],\n",
       "        ...,\n",
       "        [3.5740000e+03, 3.5780000e+03, 3.5780000e+03, 3.5740000e+03,\n",
       "         1.4100000e+03, 2.5281300e+07],\n",
       "        [3.5760000e+03, 3.5780000e+03, 3.5800000e+03, 3.5760000e+03,\n",
       "         3.6110000e+03, 6.4745230e+07],\n",
       "        [3.5780000e+03, 3.5760000e+03, 3.5800000e+03, 3.5760000e+03,\n",
       "         3.6090000e+03, 6.4709370e+07]]),\n",
       " (537555, 6),\n",
       " array([-0.00063715, -0.0012751 ,  0.        , ...,  0.0005596 ,\n",
       "         0.00055928,  0.00055897]),\n",
       " (537555,))"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train = train_data.drop(columns=['date','1_min_return']).to_numpy()\n",
    "y_train = train_data['1_min_return'].values\n",
    "\n",
    "X_train, X_train.shape, y_train, y_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T02:55:00.057976Z",
     "start_time": "2021-09-26T02:55:00.040976Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([[3.5800000e+03, 3.5820000e+03, 3.5880000e+03, 3.5780000e+03,\n",
       "         7.8380000e+03, 1.4053534e+08],\n",
       "        [3.5820000e+03, 3.5780000e+03, 3.5840000e+03, 3.5760000e+03,\n",
       "         6.9350000e+03, 1.2434455e+08],\n",
       "        [3.5780000e+03, 3.5760000e+03, 3.5800000e+03, 3.5720000e+03,\n",
       "         4.1780000e+03, 7.4869760e+07],\n",
       "        ...,\n",
       "        [5.2020000e+03, 5.2060000e+03, 5.2080000e+03, 5.2000000e+03,\n",
       "         5.2570000e+03, 1.3589345e+08],\n",
       "        [5.2060000e+03, 5.2100000e+03, 5.2100000e+03, 5.2060000e+03,\n",
       "         8.2950000e+03, 2.1442575e+08],\n",
       "        [5.2100000e+03, 5.1940000e+03, 5.2120000e+03, 5.1940000e+03,\n",
       "         8.4230000e+03, 2.1773455e+08]]),\n",
       " (96104, 6),\n",
       " array([ 0.00055866, -0.00111669, -0.00055897, ...,  0.00076894,\n",
       "         0.00076834, -0.00307102]),\n",
       " (96104,))"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test = test_data.drop(columns=['date', '1_min_return']).to_numpy()\n",
    "y_test = test_data['1_min_return'].values\n",
    "\n",
    "X_test, X_test.shape, y_test, y_test.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Add More Functions and a Customized Backtest Fitness Function for GPlearn\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T02:55:01.186795Z",
     "start_time": "2021-09-26T02:55:00.944082Z"
    },
    "code_folding": [
     1,
     22,
     43,
     64,
     77,
     91,
     104,
     119,
     134,
     147,
     162,
     175,
     190,
     193,
     209,
     219,
     229,
     241,
     254,
     275,
     288,
     301,
     314,
     328,
     341,
     354,
     368,
     381,
     392,
     403,
     414,
     417,
     421,
     424,
     427,
     433
    ]
   },
   "outputs": [],
   "source": [
    "init_function = ['add', 'sub', 'mul', 'div','sqrt', 'log','inv','sin','max','min']\n",
    "def _ts_beta(x1, x2, n):\n",
    "    with np.errstate(divide='ignore', invalid='ignore'):\n",
    "        try:\n",
    "            if n[0] == n[1] and n[1] == n[2]:\n",
    "                u'''need：(list1,list2,number)  return：number\n",
    "                前 n 期样本 A 对 B 做回归所得回归系数'''\n",
    "                list1 = x1.flatten().tolist()\n",
    "                list2 = x2.flatten().tolist()\n",
    "                n = int(n[0])\n",
    "                list1 = np.array(list1[-n-1:]).reshape(-1, 1)\n",
    "                list2 = np.array(list2[-n-1:]).reshape(-1, 1)\n",
    "                linreg = LinearRegression()\n",
    "                model = linreg.fit(list1, list2)\n",
    "                res = linreg.coef_.tolist()[0]\n",
    "                res = np.array(res+[0]*(len(x1)-len(linreg.coef_))).flatten()\n",
    "                return res\n",
    "            else:\n",
    "                return np.zeros(x1.shape[0])\n",
    "        except:\n",
    "            return np.zeros(x1.shape[0])\n",
    "        \n",
    "def _ts_resid(x1, x2, n):\n",
    "    with np.errstate(divide='ignore', invalid='ignore'):\n",
    "        try:\n",
    "            if n[0] == n[1] and n[1] == n[2]:\n",
    "                u'''need：(list1,list2,number)  return：number\n",
    "                前 n 期样本 A 对 B 做回归所得的残差'''\n",
    "                list1 = x1.flatten().tolist()\n",
    "                list2 = x2.flatten().tolist()\n",
    "                n = int(n[0])\n",
    "                list1 = np.array(list1[-n-1:]).reshape(-1, 1)\n",
    "                list2 = np.array(list2[-n-1:]).reshape(-1, 1)\n",
    "                linreg = LinearRegression()\n",
    "                model = linreg.fit(list1, list2)\n",
    "                res = list(linreg.intercept_)\n",
    "                res = np.array(res+[0]*(len(x1)-len([linreg.intercept_]))).flatten()\n",
    "                return res\n",
    "            else:\n",
    "                return np.zeros(x1.shape[0])\n",
    "        except:\n",
    "            return np.zeros(x1.shape[0])\n",
    "        \n",
    "def _corr(data1,data2,n):\n",
    "    with np.errstate(divide='ignore', invalid='ignore'):\n",
    "        try:\n",
    "            if n[0] == n[1] and n[1] ==n[2]:\n",
    "                window  = n[0]\n",
    "                x1 = pd.Series(data1.flatten())\n",
    "                x2 = pd.Series(data2.flatten())\n",
    "                df = pd.concat([x1,x2],axis=1)\n",
    "                temp = pd.Series()\n",
    "                for i in range(len(df)):\n",
    "                    if i<=window-2:\n",
    "                        temp[str(i)] = np.nan\n",
    "                    else:\n",
    "                        df2 = df.iloc[i-window+1:i,:]\n",
    "                        temp[str(i)] = df2.corr('spearman').iloc[1,0]\n",
    "                return np.nan_to_num(temp)\n",
    "            else:\n",
    "                return np.zeros(data1.shape[0])\n",
    "        except:\n",
    "            return np.zeros(data1.shape[0])\n",
    "        \n",
    "def SEQUENCE(n):\n",
    "    with np.errstate(divide='ignore', invalid='ignore'):\n",
    "        try:\n",
    "            if n[0] == n[1] and n[1] == n[2]:\n",
    "                a = n[0]\n",
    "                res = np.array(list(range(1, a+1)))\n",
    "                res = res.tolist()+[0]*(len(n)-len(res))\n",
    "                return np.array(res).flatten()\n",
    "            else:\n",
    "                return np.zeros(n.shape[0])\n",
    "        except:\n",
    "            return np.zeros(n.shape[0])\n",
    "\n",
    "def _ts_prod(x, n):\n",
    "    with np.errstate(divide='ignore', invalid='ignore'):\n",
    "        try:\n",
    "            if n[0] == n[1] and n[1] == n[2]:\n",
    "                x = pd.Series(x.flatten())\n",
    "                n = n[0]\n",
    "                res = x.prod(min_count=n)\n",
    "                res = np.array([res]+[0]*(len(x)-1))\n",
    "                return res\n",
    "            else:\n",
    "                return np.zeros(x.shape[0]) \n",
    "        except:\n",
    "            return np.zeros(data1.shape[0]) \n",
    "        \n",
    "def _ts_cov(x,y,window):\n",
    "    with np.errstate(divide='ignore', invalid='ignore'):\n",
    "        try:\n",
    "            if n[0] == n[1] and n1[1] == n[2]:\n",
    "                x = pd.Series(x.flatten())\n",
    "                y = pd.Series(y.flatten())\n",
    "                res = x.rolling(window).cov(y).fillna(0).values\n",
    "                return res\n",
    "            else:\n",
    "                return np.zeros(x.shape[0]) \n",
    "        except:\n",
    "            return np.zeros(x.shape[0])\n",
    "\n",
    "def _ts_lowday(x1, n1):\n",
    "    with np.errstate(divide='ignore', invalid='ignore'):\n",
    "        try:\n",
    "            if n1[0] == n1[1] and n1[1] == n1[2]:\n",
    "                '''need：(list,number)  return：number\n",
    "                计算 A 前 n 期时间序列中最小值距离当前时点的间隔'''\n",
    "                n = int(n1[0])\n",
    "                x = x1.flatten().tolist()    \n",
    "                lowday = n-x[-n:].index(min(list(x)[-n:]))-1\n",
    "                return np.array([lowday]+[0]*(len(x)-1))\n",
    "            else:\n",
    "                return np.zeros(x1.shape[0]) \n",
    "        except:\n",
    "            return np.zeros(x1.shape[0]) \n",
    "        \n",
    "def _ts_highday(x1, n1):\n",
    "    with np.errstate(divide='ignore', invalid='ignore'):\n",
    "        try:\n",
    "            if n1[0] == n1[1] and n1[1] == n1[2]:\n",
    "                u'''need：(list,number)  return：number\n",
    "                计算 A 前 n 期时间序列中最大值距离当前时点的间隔'''\n",
    "                n = int(n1[0])\n",
    "                x = x1.flatten().tolist()\n",
    "                highday = n-x[-n:].index(max(list(x)[-n:]))-1\n",
    "                return np.array([highday]+[0]*(len(x)-1))\n",
    "            else:\n",
    "                return np.zeros(x1.shape[0]) \n",
    "        except:\n",
    "            return np.zeros(x1.shape[0]) \n",
    "        \n",
    "def _ts_adv(x,d):\n",
    "    with np.errstate(divide='ignore', invalid='ignore'):\n",
    "        try:\n",
    "            if d[0] == d[1] and d[1] == d[2]:\n",
    "                d = int(d[0])\n",
    "                reslist(map(lambda t: np.mean(x.flatten()[t:t+20]),range(len(x.flatten())-d+1)))\n",
    "                res = res[numpy.isneginf(res)] = 0\n",
    "                return np.array(res)\n",
    "            else:\n",
    "                return np.zeros(x.shape[0])\n",
    "        except:\n",
    "            return np.zeros(x.shape[0])\n",
    "            \n",
    "def _ts_wma(x1,d):\n",
    "        with np.errstate(divide='ignore', invalid='ignore'):\n",
    "            try:\n",
    "                if d[0] == d[1] and d[1] == d[2]:\n",
    "                    d = int(d[0])\n",
    "                    x1 = pd.Series(x1.flatten())\n",
    "                    weights = np.arange(1,d+1)\n",
    "                    wma = x1.rolling(d).apply(lambda prices: np.dot(prices, weights)/weights.sum(), raw=True)\n",
    "                    wma = wma.fillna(0).values\n",
    "                    return wma\n",
    "                else:\n",
    "                    return np.zeros(x1.shape[0])\n",
    "            except:\n",
    "                return np.zeros(x1.shape[0])\n",
    "\n",
    "def _ts_mean(X, d):\n",
    "    with np.errstate(divide='ignore', invalid='ignore'):\n",
    "        try:\n",
    "            if d[0] == d[1] and d[1] == d[2]:\n",
    "                X = pd.Series(X.flatten())\n",
    "                res = X.rolling(window=d[0]).mean()\n",
    "                res = res.fillna(0).values\n",
    "                return res\n",
    "            else:\n",
    "                return np.zeros(X.shape[0])\n",
    "        except:\n",
    "            return np.zeros(X.shape[0])\n",
    "\n",
    "def _decay_linear(data1,n):\n",
    "    with np.errstate(divide='ignore', invalid='ignore'):\n",
    "        try:\n",
    "            if n[0] == n[1] and n[1] == n[2]:\n",
    "                n = int(n[0])\n",
    "                w = np.arange(1,n+1)\n",
    "                w = w[::-1]\n",
    "                w = np.array(w)/np.sum(w)\n",
    "                res = list(map(lambda t: np.sum(np.multiply(np.array(data1.flatten()[t:t+n]),w)),range(len(data1.flatten())-n+1)))\n",
    "                return np.array(res).flatten()\n",
    "            else:\n",
    "                return np.zeros(data1.shape[0])\n",
    "        except:\n",
    "            return np.zeros(data1.shape[0])\n",
    "\n",
    "def _abs(x1):\n",
    "    return abs(x1.flatten())\n",
    "\n",
    "def _ts_abs(data1, n):\n",
    "    with np.errstate(divide='ignore', invalid='ignore'):\n",
    "        try:\n",
    "            if n[0] == n[1] and n[1] == n[2]:\n",
    "                window = int(n[0])\n",
    "                x1 = data1.flatten()\n",
    "                temp1 = x1[-window:]\n",
    "                temp1 = np.abs(temp1)\n",
    "                temp2 = x1[:-window]\n",
    "                result = np.append(temp2,temp1)\n",
    "                return result\n",
    "            else:\n",
    "                return np.zeros(data1.shape[0])\n",
    "        except:\n",
    "            return np.zeros(data1.shape[0]) \n",
    "\n",
    "def _ts_scale(data1, n):\n",
    "        with np.errstate(divide='ignore', invalid='ignore'):\n",
    "            try:\n",
    "                if n[0] == n[1] and n[1] == n[2]:\n",
    "                    return (int(n[0])*x1 / np.nansum(np.abs(data1.flatten())))\n",
    "                else:\n",
    "                    return np.zeros(data1.shape[0])\n",
    "            except:\n",
    "                return np.zeros(data1.shape[0])\n",
    "            \n",
    "def _signedpower(data1, n):\n",
    "    with np.errstate(divide='ignore', invalid='ignore'):\n",
    "        try:\n",
    "            if n[0] == n[1] and n[1] == n[2]:\n",
    "                return np.sign(data1.flatten()) * np.power(np.abs(data1.flatten()), int(n[0]))\n",
    "            else:\n",
    "                return np.zeros(data1.shape[0])\n",
    "        except:\n",
    "            return np.zeros(data1.shape[0]) \n",
    "\n",
    "def _ts_delta(data1,n):\n",
    "    with np.errstate(divide='ignore', invalid='ignore'):\n",
    "        try:\n",
    "            if n[0] == n[1] and n[1] == n[2]:\n",
    "                a1 = data1-pd.Series(data1.flatten()).shift(periods=int(n[0]))\n",
    "                a1 = a1.fillna(0).values\n",
    "                return a1\n",
    "            else:\n",
    "                return np.zeros(data1.shape[0])\n",
    "        except:\n",
    "            return np.zeros(data1.shape[0]) \n",
    "\n",
    "def _ts_delay(data1,n):\n",
    "    with np.errstate(divide='ignore', invalid='ignore'):\n",
    "        try:\n",
    "            if n[0] == n[1] and n[1] == n[2]:\n",
    "                a1 = pd.Series(data1.flatten())\n",
    "                a1 = a1.shift(periods=int(n[0]))\n",
    "                a1 = a1.fillna(0).values\n",
    "                return a1\n",
    "            else:\n",
    "                return np.zeros(data1.shape[0])\n",
    "        except:\n",
    "            return np.zeros(data1.shape[0]) \n",
    "\n",
    "def _ts_corr(data1,data2,n):\n",
    "    with np.errstate(divide='ignore', invalid='ignore'):\n",
    "        try:\n",
    "            if n[0] == n[1] and n[1] ==n[2]:\n",
    "                window = int(n[0])\n",
    "                x1 = pd.Series(data1.flatten())\n",
    "                x2 = pd.Series(data2.flatten())\n",
    "                df = pd.concat([x1,x2],axis=1)\n",
    "                temp = pd.Series()\n",
    "                for i in range(len(df)):\n",
    "                    if i<=window-2:\n",
    "                        temp[str(i)] = np.nan\n",
    "                    else:\n",
    "                        df2 = df.iloc[i-window+1:i,:]\n",
    "                        temp[str(i)] = df2.corr('spearman').iloc[1,0]\n",
    "                return np.nan_to_num(temp)\n",
    "            else:\n",
    "                return np.zeros(data1.shape[0])\n",
    "        except:\n",
    "            return np.zeros(data1.shape[0]) \n",
    "\n",
    "def _ts_sum(data,n):\n",
    "    with np.errstate(divide='ignore', invalid='ignore'):\n",
    "        try:\n",
    "            if n[0] == n[1] and n[1] ==n[2]:\n",
    "                window = int(n[0])\n",
    "                value = np.array(pd.Series(data.flatten()).rolling(window).sum().tolist())\n",
    "                value = np.nan_to_num(value)\n",
    "                return value\n",
    "            else:\n",
    "                return np.zeros(data.shape[0])\n",
    "        except:\n",
    "            return np.zeros(data.shape[0])\n",
    "        \n",
    "def _ts_sma(data,n):\n",
    "    with np.errstate(divide='ignore', invalid='ignore'):   \n",
    "        try:\n",
    "            if n[0] == n[1] and n[1] ==n[2]:\n",
    "                window = int(n[0])\n",
    "                value = np.array(pd.Series(data.flatten()).rolling(window).mean().tolist())\n",
    "                value = np.nan_to_num(value)\n",
    "                return value\n",
    "            else:\n",
    "                return np.zeros(data.shape[0])\n",
    "        except:\n",
    "            return np.zeros(data.shape[0])\n",
    "\n",
    "def _ts_stddev(data,n):   \n",
    "    with np.errstate(divide='ignore', invalid='ignore'): \n",
    "        try:\n",
    "            if n[0] == n[1] and n[1] ==n[2]:\n",
    "                window = int(n[0])\n",
    "                value = np.array(pd.Series(data.flatten()).rolling(window).std().tolist())\n",
    "                value = np.nan_to_num(value)\n",
    "                return value\n",
    "            else:\n",
    "                return np.zeros(data.shape[0])\n",
    "        except:\n",
    "            return np.zeros(data.shape[0])\n",
    "\n",
    "def _ts_rank(x1,n):\n",
    "    with np.errstate(divide='ignore', invalid='ignore'):\n",
    "        try:\n",
    "            if n[0] == n[1] and n[1] ==n[2]:\n",
    "                n = int(n[0])\n",
    "                x2 = pd.Series(x1.flatten()[-n:])\n",
    "                res = x2.rank(pct=True).values\n",
    "                res = res.tolist()+[0]*(x1.shape[0] - res.shape[0])\n",
    "                return np.array(res)\n",
    "            else:\n",
    "                return np.zeros(x1.shape[0]) \n",
    "        except:\n",
    "            return np.zeros(x1.shape[0]) \n",
    "        \n",
    "def _ts_argmin(data,n):\n",
    "    with np.errstate(divide='ignore', invalid='ignore'):\n",
    "        try:\n",
    "            if n[0] == n[1] and n[1] ==n[2]:\n",
    "                window = int(n[0])\n",
    "                value = pd.Series(data.flatten()).rolling(window).apply(np.argmin) + 1 \n",
    "                value = np.nan_to_num(value)\n",
    "                return value\n",
    "            else:\n",
    "                return np.zeros(data.shape[0])  \n",
    "        except:\n",
    "            return np.zeros(data.shape[0])  \n",
    "\n",
    "def _ts_argmax(data,n):\n",
    "    with np.errstate(divide='ignore', invalid='ignore'):\n",
    "        try:\n",
    "            if n[0] == n[1] and n[1] ==n[2]:\n",
    "                window = int(n[0])\n",
    "                value = pd.Series(data.flatten()).rolling(window).apply(np.argmax) + 1 \n",
    "                value = np.nan_to_num(value)\n",
    "                return value\n",
    "            else:\n",
    "                return np.zeros(data.shape[0])\n",
    "        except:\n",
    "            return np.zeros(data.shape[0]) \n",
    "        \n",
    "def _ts_min(data,n):\n",
    "    import numpy as np\n",
    "    with np.errstate(divide='ignore', invalid='ignore'):\n",
    "        try:\n",
    "            if n[0] == n[1] and n[1] ==n[2]:\n",
    "                window = int(n[0])\n",
    "                value = np.array(pd.Series(data.flatten()).rolling(window).min().tolist())\n",
    "                value = np.nan_to_num(value)\n",
    "                return value\n",
    "            else:\n",
    "                return np.zeros(data.shape[0])\n",
    "        except:\n",
    "            return np.zeros(data.shape[0])\n",
    "\n",
    "def _ts_max(data,n):\n",
    "    with np.errstate(divide='ignore', invalid='ignore'):\n",
    "        try:\n",
    "            if n[0] == n[1] and n[1] ==n[2]:\n",
    "                window = int(n[0])\n",
    "                value = np.array(pd.Series(data.flatten()).rolling(window).max().tolist())\n",
    "                value = np.nan_to_num(value)\n",
    "                return value\n",
    "            else:\n",
    "                return np.zeros(data.shape[0])\n",
    "        except:\n",
    "            return np.zeros(data.shape[0])\n",
    "@jit\n",
    "def _beta(x1, x2):\n",
    "    list1 = x1.flatten().tolist()\n",
    "    list2 = x2.flatten().tolist()\n",
    "    list1 = np.array(list1).reshape(-1, 1)\n",
    "    list2 = np.array(list2).reshape(-1, 1)\n",
    "    linreg = LinearRegression()\n",
    "    model = linreg.fit(list1, list2)\n",
    "    res = linreg.coef_.tolist()[0]\n",
    "    res = np.array(res+[0]*(len(x1)-len(linreg.coef_))).flatten()\n",
    "    return res\n",
    "\n",
    "def _resid(x1, x2):\n",
    "    list1 = x1.flatten().tolist()\n",
    "    list2 = x2.flatten().tolist()\n",
    "    list1 = np.array(list1).reshape(-1, 1)\n",
    "    list2 = np.array(list2).reshape(-1, 1)\n",
    "    linreg = LinearRegression()\n",
    "    model = linreg.fit(list1, list2)\n",
    "    res = linreg.coef_.tolist()[0]\n",
    "    res = np.array(res+[0]*(len(x1)-len(linreg.intercept_))).flatten()\n",
    "    return res \n",
    "\n",
    "def _beta(x1, x2):\n",
    "    list1 = x1.flatten().tolist()\n",
    "    list2 = x2.flatten().tolist()\n",
    "    list1 = np.array(list1).reshape(-1, 1)\n",
    "    list2 = np.array(list2).reshape(-1, 1)\n",
    "    linreg = LinearRegression()\n",
    "    model = linreg.fit(list1, list2)\n",
    "    res = linreg.coef_.tolist()[0]\n",
    "    res = np.array(res+[0]*(len(x1)-len(linreg.coef_))).flatten()\n",
    "    return res\n",
    "\n",
    "def _abs(x1):\n",
    "    return abs(x1.flatten())\n",
    "\n",
    "def _rank(x1):\n",
    "    x1 = pd.Series(x1.flatten())\n",
    "    return x1.rank(pct=True).values\n",
    "        \n",
    "def _cube(data):\n",
    "    return np.square(data.flatten())*data.flatten()\n",
    "\n",
    "def _square(data):\n",
    "    return np.square(data.flatten())\n",
    "\n",
    "def _ts_argmaxmin(data,n):\n",
    "    try:\n",
    "        return _ts_argmax(data,n) - _ts_argmin(data,n)\n",
    "    except:\n",
    "        return np.zeros(data.shape[0])\n",
    "\n",
    "def _my_metric_backtest(y, y_pred, w):\n",
    "    pred = pd.Series(y_pred.flatten()).fillna(0)\n",
    "    y = pd.Series(y.flatten()).fillna(0)\n",
    "    short_profit = 0\n",
    "    long_profit = 0\n",
    "    held_long = 0\n",
    "    held_short = 0\n",
    "    profit = []\n",
    "    profit_temp = 0\n",
    "    \n",
    "    for i in range(len(pred)):\n",
    "        current_pred = pred.iloc[i]\n",
    "        current_return = y.iloc[i]\n",
    "        #open long\n",
    "        if current_pred>=0.75 and held_long==0 and held_short==0:\n",
    "            held_long = 1\n",
    "            held_short = 0\n",
    "            long_profit += current_return\n",
    "                    \n",
    "        #hold long\n",
    "        elif current_pred>=0.75 and held_long==1 and held_short==0:\n",
    "            held_long = 1\n",
    "            held_short = 0\n",
    "            long_profit += current_return\n",
    "            \n",
    "        #open long and close short\n",
    "        elif current_pred>=0.75 and held_long==0 and held_short==1:\n",
    "            #close short and record profit\n",
    "            held_long = 1\n",
    "            held_short = 0\n",
    "            short_profit += (-current_return)\n",
    "            profit.append(short_profit)\n",
    "            #open long\n",
    "            long_profit += current_return\n",
    "                    \n",
    "        #open short\n",
    "        elif current_pred<=-0.75and held_long==0 and held_short==0:\n",
    "            held_long = 0\n",
    "            held_short = 1\n",
    "            short_profit += (-current_return)\n",
    "            \n",
    "        #keep short\n",
    "        elif current_pred<=-0.75 and held_long==0 and held_short==1:\n",
    "            held_long = 0\n",
    "            held_short = 1\n",
    "            short_profit += (-current_return)\n",
    "  \n",
    "        #close long and open short\n",
    "        elif current_pred <=-0.75 and held_long==1 and held_short==0:\n",
    "            #close long\n",
    "            held_long = 0\n",
    "            held_short = 1\n",
    "            long_profit += current_return\n",
    "            profit.append(long_profit)\n",
    "            #open short\n",
    "            short_profit += (-current_return)\n",
    "                    \n",
    "        #closeout long \n",
    "        elif current_pred<0.75 and current_pred>-0.75 and held_long==1 and held_short==0:\n",
    "            held_long = 0\n",
    "            held_short = 0\n",
    "            long_profit += current_return\n",
    "            profit.append(long_profit)      \n",
    "            \n",
    "        #closeout short    \n",
    "        elif current_pred<0.75 and current_pred>-0.75 and held_long==0 and held_short==1:\n",
    "            held_long = 0\n",
    "            held_short = 0\n",
    "            short_profit += (-current_return)\n",
    "            profit.append(short_profit)\n",
    "    try:        \n",
    "        total_return = profit[-1]\n",
    "    except:\n",
    "        total_return = 0\n",
    "        \n",
    "    result = total_return\n",
    "    \n",
    "    del pred\n",
    "    del y\n",
    "    del short_profit\n",
    "    del long_profit\n",
    "    del held_long\n",
    "    del held_short\n",
    "    del profit\n",
    "    \n",
    "    return result"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Activate The Customized Fuctions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T02:55:01.854328Z",
     "start_time": "2021-09-26T02:55:01.751319Z"
    }
   },
   "outputs": [],
   "source": [
    "ts_beta = make_function(function = _ts_beta, name='ts_beta', arity=3, wrap=False)\n",
    "beta = make_function(function = _beta, name='beta', arity=2, wrap=False)\n",
    "ts_resid = make_function(function = _ts_resid, name='ts_resid', arity=3, wrap=False)\n",
    "resid = make_function(function = _resid, name='resid', arity=2, wrap=False)\n",
    "corr = make_function(function = _corr, name='corr', arity=3, wrap=False) \n",
    "SEQUENCE = make_function(function=SEQUENCE, name='SEQUENCE', arity=1, wrap=False) \n",
    "ts_cov = make_function(function=_ts_cov, name='ts_cov', arity=3, wrap=False) \n",
    "ts_lowday = make_function(function=_ts_lowday, name='ts_lowday', arity=2, wrap=False) \n",
    "ts_highday = make_function(function=_ts_highday, name='ts_highday', arity=2, wrap=False) \n",
    "ts_adv = make_function(function=_ts_adv, name='ts_adv', arity=2, wrap=False)  \n",
    "ts_wma = make_function(function=_ts_wma, name='ts_wma', arity=2, wrap=False)       \n",
    "ts_mean = make_function(function=_ts_mean, name='ts_mean', arity=2, wrap=False) \n",
    "decay_linear = make_function(function=_decay_linear, name='decay_linear', arity=2, wrap=False) \n",
    "_abs = make_function(function=_abs, name='_abs', arity=1, wrap=False) \n",
    "ts_abs = make_function(function=_ts_abs, name='ts_abs', arity=2, wrap=False) \n",
    "rank = make_function(function=_rank, name='rank', arity=1, wrap=False) \n",
    "ts_scale = make_function(function=_ts_scale, name='ts_scale', arity=2, wrap=False) \n",
    "ts_delta = make_function(function=_ts_delta, name='ts_delta', arity=2, wrap=False) \n",
    "ts_delay = make_function(function=_ts_delay, name='ts_delay', arity=2, wrap=False)  \n",
    "ts_sma = make_function(function=_ts_sma, name='ts_sma', arity=2, wrap=False)     \n",
    "ts_std = make_function(function=_ts_stddev, name='ts_std', arity=2, wrap=False)     \n",
    "ts_rank = make_function(function=_ts_rank, name='ts_rank', arity=2, wrap=False)\n",
    "ts_stddev = make_function(function=_ts_stddev, name='ts_stddev', arity=2, wrap=False)\n",
    "ts_sum = make_function(function=_ts_sum, name='ts_sum', arity=2, wrap=False)\n",
    "ts_corr = make_function(function=_ts_corr, name='ts_corr', arity=3, wrap=False)\n",
    "ts_min = make_function(function=_ts_min, name='ts_min', arity=2, wrap=False)\n",
    "cube = make_function(function=_cube, name='cube', arity=1, wrap=False)\n",
    "square = make_function(function=_square, name='square', arity=1, wrap=False)\n",
    "ts_argmaxmin = make_function(function=_ts_argmaxmin, name='ts_argmaxmin', arity=2, wrap=False)\n",
    "ts_argmax = make_function(function=_ts_argmax, name='ts_argmax', arity=2)\n",
    "ts_argmin = make_function(function=_ts_argmin, name='ts_argmin', arity=2, wrap=False)\n",
    "ts_min = make_function(function=_ts_min, name='ts_min', arity=2, wrap=False)\n",
    "ts_max = make_function(function=_ts_max, name='ts_max', arity=2, wrap=False)\n",
    "#decay_linear,ts_corr(replaced by corr),\n",
    "user_function = [SEQUENCE, ts_cov, ts_lowday, ts_highday, ts_adv, ts_wma, ts_mean, _abs, rank, ts_scale, ts_delta, ts_delay, \n",
    "                 ts_sma, ts_std, ts_rank, ts_sum, corr, ts_min, cube, square, ts_argmaxmin, ts_argmax, ts_argmin, ts_min, ts_max, ts_beta,\n",
    "                beta, ts_resid, resid]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Activate The Customized Fitness Function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T02:55:02.636910Z",
     "start_time": "2021-09-26T02:55:02.620910Z"
    }
   },
   "outputs": [],
   "source": [
    "my_metric = make_fitness(function=_my_metric_backtest, greater_is_better=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Fit The Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T04:11:59.550569Z",
     "start_time": "2021-09-26T02:55:05.200943Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "    |   Population Average    |             Best Individual              |\n",
      "---- ------------------------- ------------------------------------------ ----------\n",
      " Gen   Length          Fitness   Length          Fitness      OOB Fitness  Time Left\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=8)]: Using backend LokyBackend with 8 concurrent workers.\n",
      "[Parallel(n_jobs=8)]: Done   2 out of   8 | elapsed: 69.8min remaining: 209.3min\n",
      "[Parallel(n_jobs=8)]: Done   8 out of   8 | elapsed: 75.7min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "   0    31.46       -0.0188075       27          12.3871              N/A   7496.11m\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "SymbolicTransformer(feature_names=['open', 'close', 'high', 'low', 'volume',\n",
       "                                   'money'],\n",
       "                    function_set=['add', 'sub', 'mul', 'div', 'sqrt', 'log',\n",
       "                                  'inv', 'sin', 'max', 'min',\n",
       "                                  <gplearn.functions._Function object at 0x000001340D115F10>,\n",
       "                                  <gplearn.functions._Function object at 0x000001340DECF3D0>,\n",
       "                                  <gplearn.functions._Function object at 0x000001340DECF6A0>,\n",
       "                                  <gplearn.functions._Functi...\n",
       "                                  <gplearn.functions._Function object at 0x000001340DECF8B0>,\n",
       "                                  <gplearn.functions._Function object at 0x000001340DECF700>, ...],\n",
       "                    generations=100,\n",
       "                    metric=<gplearn.fitness._Fitness object at 0x000001340D141DF0>,\n",
       "                    n_jobs=8, p_crossover=0.4, p_hoist_mutation=0,\n",
       "                    p_point_replace=0.4, parsimony_coefficient=0.0001,\n",
       "                    population_size=5000, random_state=5, tournament_size=30,\n",
       "                    verbose=2)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "function_set = init_function + user_function\n",
    "population_size = 5000\n",
    "generations = 100\n",
    "random_state= 5\n",
    "est_gp = SymbolicTransformer(\n",
    "                            feature_names=fields, \n",
    "                            function_set=function_set,\n",
    "                            generations=generations,\n",
    "                            metric=my_metric,\n",
    "                            population_size=population_size,\n",
    "                            tournament_size=30, \n",
    "                            random_state=random_state,\n",
    "                            verbose=2, hall_of_fame=100,\n",
    "                            parsimony_coefficient=0.0001,\n",
    "                            p_crossover = 0.4,\n",
    "                            p_subtree_mutation = 0.01,\n",
    "                            p_hoist_mutation = 0,\n",
    "                            p_point_mutation = 0.01,\n",
    "                            p_point_replace = 0.4,\n",
    "                            n_jobs = 8)\n",
    "     \n",
    "X_train = np.nan_to_num(X_train)\n",
    "y_train = np.nan_to_num(y_train)\n",
    "est_gp.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Resulting Alpha Factors"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T04:11:59.659567Z",
     "start_time": "2021-09-26T04:11:59.644568Z"
    }
   },
   "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>fitness</th>\n",
       "      <th>expression</th>\n",
       "      <th>depth</th>\n",
       "      <th>length</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>alpha_10</th>\n",
       "      <td>0.156346</td>\n",
       "      <td>sin(cube(close))</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>alpha_9</th>\n",
       "      <td>0.325947</td>\n",
       "      <td>mul(low, sin(money))</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>alpha_7</th>\n",
       "      <td>0.383441</td>\n",
       "      <td>sub(add(ts_sum(_abs(min(close, money)), SEQUEN...</td>\n",
       "      <td>5</td>\n",
       "      <td>57</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>alpha_6</th>\n",
       "      <td>0.388041</td>\n",
       "      <td>sub(inv(ts_std(volume, 0.670)), div(min(open, ...</td>\n",
       "      <td>3</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>alpha_8</th>\n",
       "      <td>0.388041</td>\n",
       "      <td>sub(ts_lowday(min(close, high), ts_sum(low, hi...</td>\n",
       "      <td>3</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>alpha_5</th>\n",
       "      <td>0.404158</td>\n",
       "      <td>min(cube(sin(close)), corr(ts_scale(open, volu...</td>\n",
       "      <td>3</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>alpha_4</th>\n",
       "      <td>0.834534</td>\n",
       "      <td>sin(mul(open, high))</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>alpha_3</th>\n",
       "      <td>6.596452</td>\n",
       "      <td>log(square(sin(low)))</td>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>alpha_2</th>\n",
       "      <td>6.71933</td>\n",
       "      <td>inv(sin(high))</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>alpha_1</th>\n",
       "      <td>12.384383</td>\n",
       "      <td>log(sub(add(rank(inv(low)), ts_rank(div(close,...</td>\n",
       "      <td>5</td>\n",
       "      <td>27</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            fitness                                         expression depth  \\\n",
       "alpha_10   0.156346                                   sin(cube(close))     2   \n",
       "alpha_9    0.325947                               mul(low, sin(money))     2   \n",
       "alpha_7    0.383441  sub(add(ts_sum(_abs(min(close, money)), SEQUEN...     5   \n",
       "alpha_6    0.388041  sub(inv(ts_std(volume, 0.670)), div(min(open, ...     3   \n",
       "alpha_8    0.388041  sub(ts_lowday(min(close, high), ts_sum(low, hi...     3   \n",
       "alpha_5    0.404158  min(cube(sin(close)), corr(ts_scale(open, volu...     3   \n",
       "alpha_4    0.834534                               sin(mul(open, high))     2   \n",
       "alpha_3    6.596452                              log(square(sin(low)))     3   \n",
       "alpha_2     6.71933                                     inv(sin(high))     2   \n",
       "alpha_1   12.384383  log(sub(add(rank(inv(low)), ts_rank(div(close,...     5   \n",
       "\n",
       "         length  \n",
       "alpha_10      3  \n",
       "alpha_9       4  \n",
       "alpha_7      57  \n",
       "alpha_6      11  \n",
       "alpha_8      11  \n",
       "alpha_5      14  \n",
       "alpha_4       4  \n",
       "alpha_3       4  \n",
       "alpha_2       3  \n",
       "alpha_1      27  "
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "best_programs = est_gp._best_programs\n",
    "best_programs_dict = {}\n",
    "\n",
    "for p in best_programs:\n",
    "    factor_name = 'alpha_' + str(best_programs.index(p) + 1)\n",
    "    best_programs_dict[factor_name] = {'fitness':p.fitness_, 'expression':str(p), 'depth':p.depth_, 'length':p.length_}\n",
    "     \n",
    "best_programs_dict = pd.DataFrame(best_programs_dict).T\n",
    "best_programs_dict = best_programs_dict.sort_values(by='fitness')\n",
    "best_programs_dict"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Save The Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T04:12:06.269981Z",
     "start_time": "2021-09-26T04:12:06.012909Z"
    }
   },
   "outputs": [],
   "source": [
    "filename = 'gplearn_PTA_1min_factors.pkl'\n",
    "with open(filename, 'wb') as f:\n",
    "     cloudpickle.dump(est_gp, f)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Read The Model File"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T04:12:07.035997Z",
     "start_time": "2021-09-26T04:12:06.983811Z"
    }
   },
   "outputs": [],
   "source": [
    "filename = 'gplearn_PTA_1min_factors.pkl'\n",
    "with open(filename, 'rb') as f:\n",
    "    est_gp = cloudpickle.load(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T04:12:07.530387Z",
     "start_time": "2021-09-26T04:12:07.519388Z"
    }
   },
   "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>fitness</th>\n",
       "      <th>expression</th>\n",
       "      <th>depth</th>\n",
       "      <th>length</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>alpha_10</th>\n",
       "      <td>0.156346</td>\n",
       "      <td>sin(cube(close))</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>alpha_9</th>\n",
       "      <td>0.325947</td>\n",
       "      <td>mul(low, sin(money))</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>alpha_7</th>\n",
       "      <td>0.383441</td>\n",
       "      <td>sub(add(ts_sum(_abs(min(close, money)), SEQUEN...</td>\n",
       "      <td>5</td>\n",
       "      <td>57</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>alpha_6</th>\n",
       "      <td>0.388041</td>\n",
       "      <td>sub(inv(ts_std(volume, 0.670)), div(min(open, ...</td>\n",
       "      <td>3</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>alpha_8</th>\n",
       "      <td>0.388041</td>\n",
       "      <td>sub(ts_lowday(min(close, high), ts_sum(low, hi...</td>\n",
       "      <td>3</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>alpha_5</th>\n",
       "      <td>0.404158</td>\n",
       "      <td>min(cube(sin(close)), corr(ts_scale(open, volu...</td>\n",
       "      <td>3</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>alpha_4</th>\n",
       "      <td>0.834534</td>\n",
       "      <td>sin(mul(open, high))</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>alpha_3</th>\n",
       "      <td>6.596452</td>\n",
       "      <td>log(square(sin(low)))</td>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>alpha_2</th>\n",
       "      <td>6.71933</td>\n",
       "      <td>inv(sin(high))</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>alpha_1</th>\n",
       "      <td>12.384383</td>\n",
       "      <td>log(sub(add(rank(inv(low)), ts_rank(div(close,...</td>\n",
       "      <td>5</td>\n",
       "      <td>27</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            fitness                                         expression depth  \\\n",
       "alpha_10   0.156346                                   sin(cube(close))     2   \n",
       "alpha_9    0.325947                               mul(low, sin(money))     2   \n",
       "alpha_7    0.383441  sub(add(ts_sum(_abs(min(close, money)), SEQUEN...     5   \n",
       "alpha_6    0.388041  sub(inv(ts_std(volume, 0.670)), div(min(open, ...     3   \n",
       "alpha_8    0.388041  sub(ts_lowday(min(close, high), ts_sum(low, hi...     3   \n",
       "alpha_5    0.404158  min(cube(sin(close)), corr(ts_scale(open, volu...     3   \n",
       "alpha_4    0.834534                               sin(mul(open, high))     2   \n",
       "alpha_3    6.596452                              log(square(sin(low)))     3   \n",
       "alpha_2     6.71933                                     inv(sin(high))     2   \n",
       "alpha_1   12.384383  log(sub(add(rank(inv(low)), ts_rank(div(close,...     5   \n",
       "\n",
       "         length  \n",
       "alpha_10      3  \n",
       "alpha_9       4  \n",
       "alpha_7      57  \n",
       "alpha_6      11  \n",
       "alpha_8      11  \n",
       "alpha_5      14  \n",
       "alpha_4       4  \n",
       "alpha_3       4  \n",
       "alpha_2       3  \n",
       "alpha_1      27  "
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "best_programs = est_gp._best_programs\n",
    "best_programs_dict = {}\n",
    "\n",
    "for p in best_programs:\n",
    "    factor_name = 'alpha_' + str(best_programs.index(p) + 1)\n",
    "    best_programs_dict[factor_name] = {'fitness':p.fitness_, 'expression':str(p), 'depth':p.depth_, 'length':p.length_}\n",
    "     \n",
    "best_programs_dict = pd.DataFrame(best_programs_dict).T\n",
    "best_programs_dict = best_programs_dict.sort_values(by='fitness')\n",
    "best_programs_dict"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Visualize The Alpha Factor Expressions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T04:12:11.459831Z",
     "start_time": "2021-09-26T04:12:11.230868Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "log(sub(add(rank(inv(low)), ts_rank(div(close, low), SEQUENCE(money))), resid(ts_scale(SEQUENCE(low), ts_argmaxmin(low, -0.620)), mul(ts_cov(open, high, high), min(open, -0.012)))))\n",
      "fitness: 12.38438262210482, depth: 5, length: 27\n"
     ]
    },
    {
     "data": {
      "image/svg+xml": [
       "<?xml version=\"1.0\" encoding=\"UTF-8\" standalone=\"no\"?>\r\n",
       "<!DOCTYPE svg PUBLIC \"-//W3C//DTD SVG 1.1//EN\"\r\n",
       " \"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd\">\r\n",
       "<!-- Generated by graphviz version 2.47.0 (20210316.0004)\r\n",
       " -->\r\n",
       "<!-- Title: program Pages: 1 -->\r\n",
       "<svg width=\"930pt\" height=\"404pt\"\r\n",
       " viewBox=\"0.00 0.00 929.80 404.00\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\">\r\n",
       "<g id=\"graph0\" class=\"graph\" transform=\"scale(1 1) rotate(0) translate(4 400)\">\r\n",
       "<title>program</title>\r\n",
       "<polygon fill=\"white\" stroke=\"transparent\" points=\"-4,4 -4,-400 925.8,-400 925.8,4 -4,4\"/>\r\n",
       "<!-- 0 -->\r\n",
       "<g id=\"node1\" class=\"node\">\r\n",
       "<title>0</title>\r\n",
       "<ellipse fill=\"#136ed4\" stroke=\"black\" cx=\"274\" cy=\"-378\" rx=\"27\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"274\" y=\"-374.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">log</text>\r\n",
       "</g>\r\n",
       "<!-- 1 -->\r\n",
       "<g id=\"node2\" class=\"node\">\r\n",
       "<title>1</title>\r\n",
       "<ellipse fill=\"#136ed4\" stroke=\"black\" cx=\"274\" cy=\"-306\" rx=\"27\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"274\" y=\"-302.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">sub</text>\r\n",
       "</g>\r\n",
       "<!-- 0&#45;&gt;1 -->\r\n",
       "<g id=\"edge26\" class=\"edge\">\r\n",
       "<title>0&#45;&gt;1</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M274,-359.7C274,-351.98 274,-342.71 274,-334.11\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"277.5,-334.1 274,-324.1 270.5,-334.1 277.5,-334.1\"/>\r\n",
       "</g>\r\n",
       "<!-- 2 -->\r\n",
       "<g id=\"node3\" class=\"node\">\r\n",
       "<title>2</title>\r\n",
       "<ellipse fill=\"#136ed4\" stroke=\"black\" cx=\"184\" cy=\"-234\" rx=\"27\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"184\" y=\"-230.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">add</text>\r\n",
       "</g>\r\n",
       "<!-- 1&#45;&gt;2 -->\r\n",
       "<g id=\"edge25\" class=\"edge\">\r\n",
       "<title>1&#45;&gt;2</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M257.07,-291.83C243.5,-281.27 224.24,-266.3 208.91,-254.37\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"210.89,-251.48 200.85,-248.11 206.6,-257.01 210.89,-251.48\"/>\r\n",
       "</g>\r\n",
       "<!-- 12 -->\r\n",
       "<g id=\"node13\" class=\"node\">\r\n",
       "<title>12</title>\r\n",
       "<ellipse fill=\"#136ed4\" stroke=\"black\" cx=\"433\" cy=\"-234\" rx=\"28.7\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"433\" y=\"-230.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">resid</text>\r\n",
       "</g>\r\n",
       "<!-- 1&#45;&gt;12 -->\r\n",
       "<g id=\"edge24\" class=\"edge\">\r\n",
       "<title>1&#45;&gt;12</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M296.11,-295.27C323.29,-283.3 369.63,-262.9 400.78,-249.19\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"402.57,-252.22 410.31,-244.99 399.75,-245.82 402.57,-252.22\"/>\r\n",
       "</g>\r\n",
       "<!-- 3 -->\r\n",
       "<g id=\"node4\" class=\"node\">\r\n",
       "<title>3</title>\r\n",
       "<ellipse fill=\"#136ed4\" stroke=\"black\" cx=\"78\" cy=\"-162\" rx=\"27\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"78\" y=\"-158.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">rank</text>\r\n",
       "</g>\r\n",
       "<!-- 2&#45;&gt;3 -->\r\n",
       "<g id=\"edge9\" class=\"edge\">\r\n",
       "<title>2&#45;&gt;3</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M165.54,-220.81C148.84,-209.78 124.05,-193.41 105.11,-180.9\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"106.88,-177.88 96.61,-175.29 103.02,-183.72 106.88,-177.88\"/>\r\n",
       "</g>\r\n",
       "<!-- 6 -->\r\n",
       "<g id=\"node7\" class=\"node\">\r\n",
       "<title>6</title>\r\n",
       "<ellipse fill=\"#136ed4\" stroke=\"black\" cx=\"184\" cy=\"-162\" rx=\"37.89\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"184\" y=\"-158.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">ts_rank</text>\r\n",
       "</g>\r\n",
       "<!-- 2&#45;&gt;6 -->\r\n",
       "<g id=\"edge8\" class=\"edge\">\r\n",
       "<title>2&#45;&gt;6</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M184,-215.7C184,-207.98 184,-198.71 184,-190.11\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"187.5,-190.1 184,-180.1 180.5,-190.1 187.5,-190.1\"/>\r\n",
       "</g>\r\n",
       "<!-- 4 -->\r\n",
       "<g id=\"node5\" class=\"node\">\r\n",
       "<title>4</title>\r\n",
       "<ellipse fill=\"#136ed4\" stroke=\"black\" cx=\"48\" cy=\"-90\" rx=\"27\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"48\" y=\"-86.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">inv</text>\r\n",
       "</g>\r\n",
       "<!-- 3&#45;&gt;4 -->\r\n",
       "<g id=\"edge2\" class=\"edge\">\r\n",
       "<title>3&#45;&gt;4</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M70.89,-144.41C67.34,-136.13 62.97,-125.92 59,-116.66\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"62.21,-115.28 55.06,-107.47 55.78,-118.04 62.21,-115.28\"/>\r\n",
       "</g>\r\n",
       "<!-- 5 -->\r\n",
       "<g id=\"node6\" class=\"node\">\r\n",
       "<title>5</title>\r\n",
       "<ellipse fill=\"#60a6f6\" stroke=\"black\" cx=\"27\" cy=\"-18\" rx=\"27\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"27\" y=\"-14.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">low</text>\r\n",
       "</g>\r\n",
       "<!-- 4&#45;&gt;5 -->\r\n",
       "<g id=\"edge1\" class=\"edge\">\r\n",
       "<title>4&#45;&gt;5</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M42.92,-72.05C40.52,-64.06 37.6,-54.33 34.92,-45.4\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"38.26,-44.36 32.04,-35.79 31.56,-46.37 38.26,-44.36\"/>\r\n",
       "</g>\r\n",
       "<!-- 7 -->\r\n",
       "<g id=\"node8\" class=\"node\">\r\n",
       "<title>7</title>\r\n",
       "<ellipse fill=\"#136ed4\" stroke=\"black\" cx=\"127\" cy=\"-90\" rx=\"27\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"127\" y=\"-86.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">div</text>\r\n",
       "</g>\r\n",
       "<!-- 6&#45;&gt;7 -->\r\n",
       "<g id=\"edge7\" class=\"edge\">\r\n",
       "<title>6&#45;&gt;7</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M170.78,-144.76C163.38,-135.68 154.03,-124.19 145.87,-114.18\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"148.4,-111.73 139.37,-106.19 142.97,-116.15 148.4,-111.73\"/>\r\n",
       "</g>\r\n",
       "<!-- 10 -->\r\n",
       "<g id=\"node11\" class=\"node\">\r\n",
       "<title>10</title>\r\n",
       "<ellipse fill=\"#136ed4\" stroke=\"black\" cx=\"230\" cy=\"-90\" rx=\"57.69\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"230\" y=\"-86.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">SEQUENCE</text>\r\n",
       "</g>\r\n",
       "<!-- 6&#45;&gt;10 -->\r\n",
       "<g id=\"edge6\" class=\"edge\">\r\n",
       "<title>6&#45;&gt;10</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M194.9,-144.41C200.38,-136.08 207.13,-125.8 213.25,-116.49\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"216.29,-118.24 218.86,-107.96 210.44,-114.39 216.29,-118.24\"/>\r\n",
       "</g>\r\n",
       "<!-- 8 -->\r\n",
       "<g id=\"node9\" class=\"node\">\r\n",
       "<title>8</title>\r\n",
       "<ellipse fill=\"#60a6f6\" stroke=\"black\" cx=\"102\" cy=\"-18\" rx=\"29.8\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"102\" y=\"-14.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">close</text>\r\n",
       "</g>\r\n",
       "<!-- 7&#45;&gt;8 -->\r\n",
       "<g id=\"edge4\" class=\"edge\">\r\n",
       "<title>7&#45;&gt;8</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M121.08,-72.41C118.22,-64.42 114.73,-54.64 111.52,-45.64\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"114.72,-44.2 108.06,-35.96 108.12,-46.55 114.72,-44.2\"/>\r\n",
       "</g>\r\n",
       "<!-- 9 -->\r\n",
       "<g id=\"node10\" class=\"node\">\r\n",
       "<title>9</title>\r\n",
       "<ellipse fill=\"#60a6f6\" stroke=\"black\" cx=\"177\" cy=\"-18\" rx=\"27\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"177\" y=\"-14.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">low</text>\r\n",
       "</g>\r\n",
       "<!-- 7&#45;&gt;9 -->\r\n",
       "<g id=\"edge3\" class=\"edge\">\r\n",
       "<title>7&#45;&gt;9</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M138.1,-73.46C144.51,-64.49 152.7,-53.02 159.92,-42.92\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"162.97,-44.66 165.93,-34.49 157.27,-40.59 162.97,-44.66\"/>\r\n",
       "</g>\r\n",
       "<!-- 11 -->\r\n",
       "<g id=\"node12\" class=\"node\">\r\n",
       "<title>11</title>\r\n",
       "<ellipse fill=\"#60a6f6\" stroke=\"black\" cx=\"257\" cy=\"-18\" rx=\"35.19\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"257\" y=\"-14.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">money</text>\r\n",
       "</g>\r\n",
       "<!-- 10&#45;&gt;11 -->\r\n",
       "<g id=\"edge5\" class=\"edge\">\r\n",
       "<title>10&#45;&gt;11</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M236.54,-72.05C239.62,-64.06 243.37,-54.33 246.82,-45.4\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"250.19,-46.38 250.52,-35.79 243.66,-43.86 250.19,-46.38\"/>\r\n",
       "</g>\r\n",
       "<!-- 13 -->\r\n",
       "<g id=\"node14\" class=\"node\">\r\n",
       "<title>13</title>\r\n",
       "<ellipse fill=\"#136ed4\" stroke=\"black\" cx=\"433\" cy=\"-162\" rx=\"40.09\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"433\" y=\"-158.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">ts_scale</text>\r\n",
       "</g>\r\n",
       "<!-- 12&#45;&gt;13 -->\r\n",
       "<g id=\"edge23\" class=\"edge\">\r\n",
       "<title>12&#45;&gt;13</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M433,-215.7C433,-207.98 433,-198.71 433,-190.11\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"436.5,-190.1 433,-180.1 429.5,-190.1 436.5,-190.1\"/>\r\n",
       "</g>\r\n",
       "<!-- 19 -->\r\n",
       "<g id=\"node20\" class=\"node\">\r\n",
       "<title>19</title>\r\n",
       "<ellipse fill=\"#136ed4\" stroke=\"black\" cx=\"662\" cy=\"-162\" rx=\"27\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"662\" y=\"-158.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">mul</text>\r\n",
       "</g>\r\n",
       "<!-- 12&#45;&gt;19 -->\r\n",
       "<g id=\"edge22\" class=\"edge\">\r\n",
       "<title>12&#45;&gt;19</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M458.38,-225.24C499.83,-212.57 582.03,-187.44 628.31,-173.3\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"629.35,-176.64 637.89,-170.37 627.3,-169.95 629.35,-176.64\"/>\r\n",
       "</g>\r\n",
       "<!-- 14 -->\r\n",
       "<g id=\"node15\" class=\"node\">\r\n",
       "<title>14</title>\r\n",
       "<ellipse fill=\"#136ed4\" stroke=\"black\" cx=\"364\" cy=\"-90\" rx=\"57.69\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"364\" y=\"-86.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">SEQUENCE</text>\r\n",
       "</g>\r\n",
       "<!-- 13&#45;&gt;14 -->\r\n",
       "<g id=\"edge14\" class=\"edge\">\r\n",
       "<title>13&#45;&gt;14</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M417.34,-145.12C408.47,-136.12 397.2,-124.68 387.31,-114.65\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"389.59,-111.97 380.07,-107.31 384.6,-116.89 389.59,-111.97\"/>\r\n",
       "</g>\r\n",
       "<!-- 16 -->\r\n",
       "<g id=\"node17\" class=\"node\">\r\n",
       "<title>16</title>\r\n",
       "<ellipse fill=\"#136ed4\" stroke=\"black\" cx=\"502\" cy=\"-90\" rx=\"61.99\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"502\" y=\"-86.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">ts_argmaxmin</text>\r\n",
       "</g>\r\n",
       "<!-- 13&#45;&gt;16 -->\r\n",
       "<g id=\"edge13\" class=\"edge\">\r\n",
       "<title>13&#45;&gt;16</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M448.66,-145.12C457.4,-136.24 468.48,-125.01 478.27,-115.08\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"480.92,-117.38 485.44,-107.8 475.93,-112.46 480.92,-117.38\"/>\r\n",
       "</g>\r\n",
       "<!-- 15 -->\r\n",
       "<g id=\"node16\" class=\"node\">\r\n",
       "<title>15</title>\r\n",
       "<ellipse fill=\"#60a6f6\" stroke=\"black\" cx=\"357\" cy=\"-18\" rx=\"27\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"357\" y=\"-14.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">low</text>\r\n",
       "</g>\r\n",
       "<!-- 14&#45;&gt;15 -->\r\n",
       "<g id=\"edge10\" class=\"edge\">\r\n",
       "<title>14&#45;&gt;15</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M362.27,-71.7C361.5,-63.98 360.57,-54.71 359.71,-46.11\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"363.19,-45.71 358.71,-36.1 356.22,-46.4 363.19,-45.71\"/>\r\n",
       "</g>\r\n",
       "<!-- 17 -->\r\n",
       "<g id=\"node18\" class=\"node\">\r\n",
       "<title>17</title>\r\n",
       "<ellipse fill=\"#60a6f6\" stroke=\"black\" cx=\"429\" cy=\"-18\" rx=\"27\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"429\" y=\"-14.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">low</text>\r\n",
       "</g>\r\n",
       "<!-- 16&#45;&gt;17 -->\r\n",
       "<g id=\"edge12\" class=\"edge\">\r\n",
       "<title>16&#45;&gt;17</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M484.7,-72.41C474.66,-62.78 461.9,-50.55 451.14,-40.23\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"453.35,-37.5 443.71,-33.11 448.51,-42.56 453.35,-37.5\"/>\r\n",
       "</g>\r\n",
       "<!-- 18 -->\r\n",
       "<g id=\"node19\" class=\"node\">\r\n",
       "<title>18</title>\r\n",
       "<ellipse fill=\"#60a6f6\" stroke=\"black\" cx=\"508\" cy=\"-18\" rx=\"33.6\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"508\" y=\"-14.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">&#45;0.620</text>\r\n",
       "</g>\r\n",
       "<!-- 16&#45;&gt;18 -->\r\n",
       "<g id=\"edge11\" class=\"edge\">\r\n",
       "<title>16&#45;&gt;18</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M503.48,-71.7C504.14,-63.98 504.94,-54.71 505.68,-46.11\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"509.17,-46.37 506.53,-36.1 502.19,-45.77 509.17,-46.37\"/>\r\n",
       "</g>\r\n",
       "<!-- 20 -->\r\n",
       "<g id=\"node21\" class=\"node\">\r\n",
       "<title>20</title>\r\n",
       "<ellipse fill=\"#136ed4\" stroke=\"black\" cx=\"662\" cy=\"-90\" rx=\"36\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"662\" y=\"-86.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">ts_cov</text>\r\n",
       "</g>\r\n",
       "<!-- 19&#45;&gt;20 -->\r\n",
       "<g id=\"edge21\" class=\"edge\">\r\n",
       "<title>19&#45;&gt;20</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M662,-143.7C662,-135.98 662,-126.71 662,-118.11\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"665.5,-118.1 662,-108.1 658.5,-118.1 665.5,-118.1\"/>\r\n",
       "</g>\r\n",
       "<!-- 24 -->\r\n",
       "<g id=\"node25\" class=\"node\">\r\n",
       "<title>24</title>\r\n",
       "<ellipse fill=\"#136ed4\" stroke=\"black\" cx=\"808\" cy=\"-90\" rx=\"27\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"808\" y=\"-86.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">min</text>\r\n",
       "</g>\r\n",
       "<!-- 19&#45;&gt;24 -->\r\n",
       "<g id=\"edge20\" class=\"edge\">\r\n",
       "<title>19&#45;&gt;24</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M683.23,-150.82C708.1,-138.9 749.48,-119.06 777.72,-105.52\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"779.24,-108.67 786.74,-101.19 776.21,-102.36 779.24,-108.67\"/>\r\n",
       "</g>\r\n",
       "<!-- 21 -->\r\n",
       "<g id=\"node22\" class=\"node\">\r\n",
       "<title>21</title>\r\n",
       "<ellipse fill=\"#60a6f6\" stroke=\"black\" cx=\"588\" cy=\"-18\" rx=\"28.7\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"588\" y=\"-14.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">open</text>\r\n",
       "</g>\r\n",
       "<!-- 20&#45;&gt;21 -->\r\n",
       "<g id=\"edge17\" class=\"edge\">\r\n",
       "<title>20&#45;&gt;21</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M645.94,-73.81C635.65,-64.08 622.13,-51.29 610.75,-40.52\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"612.94,-37.77 603.27,-33.44 608.13,-42.86 612.94,-37.77\"/>\r\n",
       "</g>\r\n",
       "<!-- 22 -->\r\n",
       "<g id=\"node23\" class=\"node\">\r\n",
       "<title>22</title>\r\n",
       "<ellipse fill=\"#60a6f6\" stroke=\"black\" cx=\"662\" cy=\"-18\" rx=\"27\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"662\" y=\"-14.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">high</text>\r\n",
       "</g>\r\n",
       "<!-- 20&#45;&gt;22 -->\r\n",
       "<g id=\"edge16\" class=\"edge\">\r\n",
       "<title>20&#45;&gt;22</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M662,-71.7C662,-63.98 662,-54.71 662,-46.11\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"665.5,-46.1 662,-36.1 658.5,-46.1 665.5,-46.1\"/>\r\n",
       "</g>\r\n",
       "<!-- 23 -->\r\n",
       "<g id=\"node24\" class=\"node\">\r\n",
       "<title>23</title>\r\n",
       "<ellipse fill=\"#60a6f6\" stroke=\"black\" cx=\"734\" cy=\"-18\" rx=\"27\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"734\" y=\"-14.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">high</text>\r\n",
       "</g>\r\n",
       "<!-- 20&#45;&gt;23 -->\r\n",
       "<g id=\"edge15\" class=\"edge\">\r\n",
       "<title>20&#45;&gt;23</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M677.98,-73.46C687.94,-63.78 700.91,-51.17 711.85,-40.53\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"714.32,-43.02 719.05,-33.54 709.44,-38 714.32,-43.02\"/>\r\n",
       "</g>\r\n",
       "<!-- 25 -->\r\n",
       "<g id=\"node26\" class=\"node\">\r\n",
       "<title>25</title>\r\n",
       "<ellipse fill=\"#60a6f6\" stroke=\"black\" cx=\"808\" cy=\"-18\" rx=\"28.7\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"808\" y=\"-14.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">open</text>\r\n",
       "</g>\r\n",
       "<!-- 24&#45;&gt;25 -->\r\n",
       "<g id=\"edge19\" class=\"edge\">\r\n",
       "<title>24&#45;&gt;25</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M808,-71.7C808,-63.98 808,-54.71 808,-46.11\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"811.5,-46.1 808,-36.1 804.5,-46.1 811.5,-46.1\"/>\r\n",
       "</g>\r\n",
       "<!-- 26 -->\r\n",
       "<g id=\"node27\" class=\"node\">\r\n",
       "<title>26</title>\r\n",
       "<ellipse fill=\"#60a6f6\" stroke=\"black\" cx=\"888\" cy=\"-18\" rx=\"33.6\" ry=\"18\"/>\r\n",
       "<text text-anchor=\"middle\" x=\"888\" y=\"-14.3\" font-family=\"Times New Roman,serif\" font-size=\"14.00\">&#45;0.012</text>\r\n",
       "</g>\r\n",
       "<!-- 24&#45;&gt;26 -->\r\n",
       "<g id=\"edge18\" class=\"edge\">\r\n",
       "<title>24&#45;&gt;26</title>\r\n",
       "<path fill=\"none\" stroke=\"black\" d=\"M823.81,-75.17C835.12,-65.27 850.57,-51.75 863.47,-40.46\"/>\r\n",
       "<polygon fill=\"black\" stroke=\"black\" points=\"865.9,-42.99 871.12,-33.77 861.29,-37.72 865.9,-42.99\"/>\r\n",
       "</g>\r\n",
       "</g>\r\n",
       "</svg>\r\n"
      ],
      "text/plain": [
       "<graphviz.files.Source at 0x1341432f490>"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def alpha_factor_graph(num):\n",
    "    # 打印指定num的表达式图\n",
    "\n",
    "    factor = best_programs[num-1]\n",
    "    print(factor)\n",
    "    print('fitness: {0}, depth: {1}, length: {2}'.format(factor.fitness_, factor.depth_, factor.length_))\n",
    "\n",
    "    dot_data = factor.export_graphviz()\n",
    "    graph = graphviz.Source(dot_data)\n",
    "    graph.render('images/alpha_factor_graph', format='png', cleanup=True)\n",
    "    \n",
    "    return graph\n",
    "\n",
    "graph1 = alpha_factor_graph(1)\n",
    "graph1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T04:12:11.645879Z",
     "start_time": "2021-09-26T04:12:11.642879Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "log(sub(add(rank(inv(low)), ts_rank(div(close, low), SEQUENCE(money))), resid(ts_scale(SEQUENCE(low), ts_argmaxmin(low, -0.620)), mul(ts_cov(open, high, high), min(open, -0.012)))))\n",
      "fitness: 12.38438262210482, depth: 5, length: 27\n"
     ]
    }
   ],
   "source": [
    "factor = best_programs[0]\n",
    "print(factor)\n",
    "print('fitness: {0}, depth: {1}, length: {2}'.format(factor.fitness_, factor.depth_, factor.length_))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Alpha Factor Analysis"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 因子回测函数 "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T04:12:14.143970Z",
     "start_time": "2021-09-26T04:12:14.128971Z"
    },
    "code_folding": [
     1
    ]
   },
   "outputs": [],
   "source": [
    "@jit\n",
    "def _factor_backtest(factor_perd, market_price):\n",
    "    pred = pd.Series(factor_perd.flatten()).fillna(0)\n",
    "    evaluation = []\n",
    "    slippage = 4\n",
    "    shares = 25\n",
    "    comission = 0.00025\n",
    "     \n",
    "    backtest_data = market_price\n",
    "    trades = pred\n",
    "\n",
    "    short_open = 0\n",
    "    long_open = 0\n",
    "    held_long = 0\n",
    "    held_short = 0\n",
    "    profit = []\n",
    "    profit_temp = 0\n",
    "    initial_assets = shares*max(backtest_data.open.values)\n",
    "    initial_cash = shares*max(backtest_data.open.values)\n",
    "    net_worth = [initial_cash]\n",
    "    \n",
    "    for i in range(len(trades)):\n",
    "        current_pred = trades.iloc[i]\n",
    "        current_close = backtest_data.iloc[i].open.astype('float')\n",
    "        #open long\n",
    "        if current_pred>=0.75 and held_long==0 and held_short==0:\n",
    "            held_long = 1\n",
    "            held_short = 0\n",
    "            long_open = current_close+slippage\n",
    "            short_open = 0\n",
    "            #print('open long')\n",
    "                    \n",
    "        #hold long\n",
    "        elif current_pred>=0.75 and held_long==1 and held_short==0:\n",
    "            #print('hold long')\n",
    "            pass\n",
    "                \n",
    "            #open long and close short\n",
    "        elif current_pred>=0.75 and held_long==0 and held_short==1:\n",
    "            held_long = 1\n",
    "            #close short and calculate profit\n",
    "            held_short = 0\n",
    "            profit_temp = (short_open-(current_close+slippage))*shares*(1-comission)\n",
    "            initial_cash += profit_temp\n",
    "            profit.append(profit_temp)\n",
    "            net_worth.append(initial_cash)\n",
    "            #open long\n",
    "            short_open = 0\n",
    "            long_open = current_close+slippage\n",
    "            #print('open long and close short')\n",
    "                    \n",
    "        #open short\n",
    "        elif current_pred<=-0.75and held_long==0 and held_short==0:\n",
    "            held_long = 0\n",
    "            held_short = 1\n",
    "            long_open = long_open\n",
    "            short_open = current_close+slippage\n",
    "            profit = profit\n",
    "            #print('open short')\n",
    "            \n",
    "        #keep short\n",
    "        elif current_pred<=-0.75 and held_long==0 and held_short==1:\n",
    "            #print('keep short')\n",
    "            pass\n",
    "                \n",
    "        #close long and open short\n",
    "        elif current_pred <=-0.75 and held_long==1 and held_short==0:\n",
    "            #close long\n",
    "            held_long = 0\n",
    "            held_short = 1\n",
    "            profit_temp = ((current_close-slippage)-long_open)*shares*(1-comission)\n",
    "            initial_cash += profit_temp\n",
    "            profit.append(profit_temp)\n",
    "            net_worth.append(initial_cash)\n",
    "            #open short\n",
    "            long_open = 0\n",
    "            short_open = current_close-slippage\n",
    "            profit = profit\n",
    "            #print('close long and open short')\n",
    "                    \n",
    "        #closeout long\n",
    "        elif current_pred<0.75 and current_pred>-0.75 and held_long==1 and held_short==0:\n",
    "            held_long = 0\n",
    "            held_short = 0\n",
    "            profit_temp = ((current_close-slippage)-long_open)*shares*(1-comission)\n",
    "            initial_cash += profit_temp\n",
    "            profit.append(profit_temp)\n",
    "            net_worth.append(initial_cash)\n",
    "            short_open = 0\n",
    "            long_open = 0\n",
    "            #print('closeout long')        \n",
    "            \n",
    "        #closeout short    \n",
    "        elif current_pred<0.75 and current_pred>-0.75 and held_long==0 and held_short==1:\n",
    "            held_long = 0\n",
    "            held_short = 0\n",
    "            profit_temp = (short_open-(current_close+slippage))*shares*(1-comission)\n",
    "            initial_cash += profit_temp\n",
    "            profit.append(profit_temp)\n",
    "            net_worth.append(initial_cash)\n",
    "            short_open = 0\n",
    "            long_open = 0\n",
    "            #print('closeout short')\n",
    "            \n",
    "    total_return = (initial_cash-initial_assets)/initial_assets\n",
    "    print('总收益率', total_return)\n",
    "    shaprpe_df = pd.Series(profit)\n",
    "    sharpe_temp = (shaprpe_df - shaprpe_df.shift(1))/shaprpe_df.shift(1)\n",
    "    sharpe = sharpe_temp.mean()/sharpe_temp.std()*np.sqrt(len(profit))\n",
    "          \n",
    "    a = np.maximum.accumulate(net_worth)\n",
    "    l = np.argmax((np.maximum.accumulate(net_worth) - net_worth)/np.maximum.accumulate(net_worth))\n",
    "    k = np.argmax(net_worth[:l])\n",
    "    max_draw = (net_worth[k] - net_worth[l])/(net_worth[l]) \n",
    "    print('最大回撤', max_draw)        \n",
    "        \n",
    "    win_count = 0\n",
    "    loss_count = 0\n",
    "    initial_profit = 0\n",
    "    for i in range(len(net_worth)):\n",
    "        current_profit = net_worth[i]\n",
    "        if i==0:\n",
    "            if current_profit > initial_assets:\n",
    "                win_count += 1\n",
    "            else:\n",
    "                loss_count += 1\n",
    "        else:\n",
    "            last_profit = net_worth[i-1]\n",
    "            if current_profit > last_profit:\n",
    "                win_count += 1\n",
    "            else:\n",
    "                loss_count += 1\n",
    "    win_rate = win_count/len(net_worth)\n",
    "    print('胜率', win_rate)\n",
    "    \n",
    "    total_gain = 0\n",
    "    total_loss = 0\n",
    "    for i in range(len(profit)):\n",
    "        if profit[i] >0:\n",
    "            total_gain += profit[i]\n",
    "        else:\n",
    "            total_loss += profit[i]\n",
    "    gain_loss_ratio = (total_gain/win_count)/(abs(total_loss)/loss_count)\n",
    "    print('盈亏比', gain_loss_ratio)\n",
    "    \n",
    "    result = total_return*np.nan_to_num(sharpe,nan=1)*win_rate*gain_loss_ratio*(1-max_draw)\n",
    "    \n",
    "    x = np.array(net_worth).reshape(len(net_worth),)\n",
    "    y = np.arange(len(net_worth))\n",
    "    \n",
    "    plt.rcParams['font.sans-serif'] = ['KaiTi'] # 指定默认字体\n",
    "    plt.rcParams['axes.unicode_minus'] = False # 解决保存图像是负号'-'显示为方块的问题\n",
    "    fig = plt.figure(figsize=(16,9))\n",
    "    plt.plot(y, x)\n",
    "    plt.title('因子资金曲线',fontsize=20)\n",
    "    plt.xlabel('交易次数',fontsize=20)\n",
    "    plt.ylabel('账户净值',fontsize=20)\n",
    "    plt.show()\n",
    "\n",
    "    return result"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Prepare Alpha Factor Backtest Results (In-Sample)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T04:12:15.887437Z",
     "start_time": "2021-09-26T04:12:15.082423Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((537555, 10), (537555, 6))"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "factors_pred = est_gp.transform(X_train)\n",
    "factors_pred.shape, X_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T04:12:16.026423Z",
     "start_time": "2021-09-26T04:12:15.981422Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(               0          1         2         3         4         5  \\\n",
       " 0       0.100368  22.882979 -0.653085  0.900637  0.000000 -0.999682   \n",
       " 1      -1.264464   1.123289 -0.008746  0.989437  0.000000 -0.999363   \n",
       " 2      -0.146115  -1.723635 -0.653085 -0.999216  0.000000 -0.999362   \n",
       " 3       0.114995  -1.274463 -2.220488  0.825151 -0.483079 -0.998407   \n",
       " 4      -0.148354   1.123289 -0.036217  0.989437 -0.483079 -0.999363   \n",
       " ...          ...        ...       ...       ...       ...       ...   \n",
       " 537550  0.667190   1.311040 -0.181889  0.967394  0.000000 -0.998881   \n",
       " 537551  0.512656   1.311040 -0.198645 -0.997367  0.000000 -1.000000   \n",
       " 537552  0.633991   3.695278 -0.198645  0.706805  0.000000 -0.998882   \n",
       " 537553  0.512296  -1.012161 -0.541641  0.920581  0.000000 -0.998883   \n",
       " 537554  0.112663  -1.012161 -0.541641 -0.757335  0.000000 -0.999441   \n",
       " \n",
       "                    6              7            8         9  \n",
       " 0      -1.163984e+12 -515272.962015 -3847.113515 -0.999276  \n",
       " 1      -1.578445e+11     -63.984116  4730.223871  0.979288  \n",
       " 2      -7.205184e+11     -76.709788  2945.151011  0.979288  \n",
       " 3      -4.120335e+10     -53.696368 -5973.986256 -0.997546  \n",
       " 4      -2.418245e+10     -49.862443  1866.359562 -0.997546  \n",
       " ...              ...            ...          ...       ...  \n",
       " 537550 -7.685231e+10     -62.710442 -3448.591131  0.930671  \n",
       " 537551 -1.878220e+10     -52.049074 -3337.868024  0.930671  \n",
       " 537552 -2.021835e+10     -52.582004 -2713.469694 -0.931886  \n",
       " 537553 -1.326058e+11     -67.104605  2171.787445 -0.931886  \n",
       " 537554 -1.324589e+11     -67.095528  2099.418035  0.930671  \n",
       " \n",
       " [537555 rows x 10 columns],\n",
       "                0\n",
       " 0       0.100368\n",
       " 1      -1.264464\n",
       " 2      -0.146115\n",
       " 3       0.114995\n",
       " 4      -0.148354\n",
       " ...          ...\n",
       " 537550  0.667190\n",
       " 537551  0.512656\n",
       " 537552  0.633991\n",
       " 537553  0.512296\n",
       " 537554  0.112663\n",
       " \n",
       " [537555 rows x 1 columns])"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pred_data = pd.DataFrame(factors_pred).T.T\n",
    "pred_data, pred_data.iloc[:,[0]]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Alpha Factor Backtest Results (In-Sample)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T04:13:20.214386Z",
     "start_time": "2021-09-26T04:12:16.722885Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "总收益率 18.505672747580395\n",
      "最大回撤 0.002646899619426275\n",
      "胜率 0.712678746716897\n",
      "盈亏比 7.897330418750164\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x648 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "103.87904216943294"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Total Rate of Return: 总收益率\n",
    "#Max drawdown: 最大回撤\n",
    "#Wining Rate: 胜率\n",
    "#Win-loss ratio:盈亏比\n",
    "#Graph Title: Factor Backtest Account Networth cruve\n",
    "#Graph X Axis: Number of Trades\n",
    "#Graph y Axis: Account Networth\n",
    "_factor_backtest(pred_data.iloc[:,[0]].values, train_data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Prepare Alpha Factor Backtest Results (Out-of-Sample)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T04:13:20.477401Z",
     "start_time": "2021-09-26T04:13:20.339386Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((96104, 10), (96104, 6))"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "factors_pred_1 = est_gp.transform(X_test)\n",
    "factors_pred_1.shape, X_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T04:13:20.647386Z",
     "start_time": "2021-09-26T04:13:20.618386Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(              0          1         2         3         4         5  \\\n",
       " 0      0.524718   3.370902 -2.614112 -0.638455  0.000000 -0.997770   \n",
       " 1      0.367585   1.890978 -0.541641  0.863733  0.000000 -0.999442   \n",
       " 2      0.527684  -1.012161  0.000000 -0.757335  0.000000 -0.999441   \n",
       " 3      0.527684   3.695278  0.000000  0.234697  0.000000 -0.999441   \n",
       " 4      0.527684   1.311040  0.000000  0.091613  0.000000 -0.999441   \n",
       " ...         ...        ...       ...       ...       ...       ...   \n",
       " 96099 -1.594632   1.151090 -0.118088 -0.888143 -0.837670 -0.999230   \n",
       " 96100  0.055283  -2.176269 -0.118088 -0.976671 -0.234197 -0.997309   \n",
       " 96101 -0.085450  -1.450599 -0.967729  0.995723 -0.051406 -0.998848   \n",
       " 96102 -0.303794   1.057549 -1.978670 -0.578769  0.000000 -0.999232   \n",
       " 96103 -1.602842 -10.242552 -0.416836  0.510597 -0.535125 -0.999616   \n",
       " \n",
       "                   6             7            8         9  \n",
       " 0     -6.247722e+11 -21968.402408   891.993268  0.935941  \n",
       " 1     -4.891073e+11    -78.222285 -3516.647952 -0.931886  \n",
       " 2     -1.774213e+11    -69.515373  2240.298450  0.930671  \n",
       " 3     -1.072262e+11    -65.380076 -3536.327905  0.930671  \n",
       " 4     -3.904640e+10    -57.467009 -2149.987316  0.930671  \n",
       " ...             ...           ...          ...       ...  \n",
       " 96099 -1.264734e+11    -63.762524  1732.942538  0.619167  \n",
       " 96100 -1.253970e+12    -83.396394   634.151738  0.995431  \n",
       " 96101 -4.051981e+11    -73.398900  4932.086547 -0.246165  \n",
       " 96102 -1.008844e+12    -81.421896  1253.035948 -0.202595  \n",
       " 96103 -1.040220e+12    -81.698483 -2661.092380  0.577035  \n",
       " \n",
       " [96104 rows x 10 columns],\n",
       "               0\n",
       " 0      0.524718\n",
       " 1      0.367585\n",
       " 2      0.527684\n",
       " 3      0.527684\n",
       " 4      0.527684\n",
       " ...         ...\n",
       " 96099 -1.594632\n",
       " 96100  0.055283\n",
       " 96101 -0.085450\n",
       " 96102 -0.303794\n",
       " 96103 -1.602842\n",
       " \n",
       " [96104 rows x 1 columns])"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pred_data_1 = pd.DataFrame(factors_pred_1).T.T\n",
    "pred_data_1, pred_data_1.iloc[:,[0]]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Alpha Factor Backtest Results (Out-of-Sample)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-09-26T04:13:30.897424Z",
     "start_time": "2021-09-26T04:13:20.786386Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "总收益率 5.333650336045313\n",
      "最大回撤 0.009763898723050963\n",
      "胜率 0.7675511751326762\n",
      "盈亏比 6.328575792104121\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x648 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "25.65527196285825"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Total Rate of Return: 总收益率\n",
    "#Max drawdown: 最大回撤\n",
    "#Wining Rate: 胜率\n",
    "#Win-loss ratio:盈亏比\n",
    "#Graph Title: Factor Backtest Account Networth cruve\n",
    "#Graph X Axis: Number of Trades\n",
    "#Graph y Axis: Account Networth\n",
    "_factor_backtest(pred_data_1.iloc[:,[0]].values, test_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "hide_input": false,
  "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": "3.9.7"
  },
  "latex_envs": {
   "LaTeX_envs_menu_present": true,
   "autoclose": false,
   "autocomplete": true,
   "bibliofile": "biblio.bib",
   "cite_by": "apalike",
   "current_citInitial": 1,
   "eqLabelWithNumbers": true,
   "eqNumInitial": 1,
   "hotkeys": {
    "equation": "Ctrl-E",
    "itemize": "Ctrl-I"
   },
   "labels_anchors": false,
   "latex_user_defs": false,
   "report_style_numbering": false,
   "user_envs_cfg": false
  },
  "nbTranslate": {
   "displayLangs": [
    "*"
   ],
   "hotkey": "alt-t",
   "langInMainMenu": true,
   "sourceLang": "en",
   "targetLang": "fr",
   "useGoogleTranslate": true
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": false,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "MarkDown菜单",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": false
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}