Fin_Quant_Pairs_Trading_Part_1_Spread.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Modelo de Pairs Trading Baseado em Cointegração"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Autores: Alexandre Fernandes Theoharidis, Débora Darin e Nelly Pires Colnaghi**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Este trabalho contém:**\n",
    "\n",
    "**1)** Importação da série de preços das ações PETR3 e PETR4 e tratamento de dados;\n",
    "\n",
    "**2)** Cálculo da regressão linear dos preços das ações & computação dos resíduos da regressão (spreads) para verificação de desvios da média;\n",
    "\n",
    "Os seguintes itens serão apresentados no notebook parte 2 (Fin_Quant_Pairs_Trading_Part_2_Backtest):\n",
    "\n",
    "**3)** Backtest da estratégia de pairs trading;\n",
    "\n",
    "**4)** Cálculo do retorno para o par PETR3 x PETR4."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Introdução"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Neste notebook, foi desenvolvido um modelo de pairs trading de ações baseado em cointegração, inspirado no artigo de Caldeira e Moura (2013). A estratégia tem como objetivo identificar, no curto prazo, desvios entre preços de pares de ativos que estejam relacionados no longo prazo. Assim, A cointegração se verifica quando os resíduos de um modelo de regressão entre duas séries são estacionários - dessa forma, dizemos que as séries são cointegradas.\n",
    "\n",
    "Assim, a primeira etapa da estratégia resume-se a encontrar um par de ações com spread estacionário. Matematicamente, sejam $A_t$ e $B_t$, $t = 1, 2,…, T$, a série de preços das ações A e B, respectivamente. Uma abordagem simples para verificar se essas ações são elegíveis consiste no seguinte modelo de regressão (1):\n",
    "\n",
    "\\begin{align}\n",
    "ln A_t = \\alpha + \\beta ln B_t + \\varepsilon_t\n",
    "\\end{align}\n",
    "\n",
    "onde $\\varepsilon_t$ é o spread do par, $\\beta$ é a proporção do ativo A em relação ao ativo B e $\\alpha$ é uma constante. Teoricamente, o spread deve ter média zero. Os log preços dos ativos são usados em vez dos seus preços de mercado, de modo a impor uma relação entre os retornos desses ativos, semelhante aos modelos de fatores aplicado aos retornos das ações, como os introduzidos por Carhart (1997) e Fama e French (1993), relacionando o retorno esperado da ação A ao retorno esperado da ação B, sendo E o operador de valor esperado (2):\n",
    "\n",
    "\\begin{align}\n",
    "E[ln(A_t/A_{t-1})] = \\beta E[ln(B_t/B_{t-1})]\n",
    "\\end{align}\n",
    "\n",
    "Esse resultado é uma condição necessária para a implementação de estratégias de negociação de pares usando um único par de ações, pois garante que os retornos contínuos esperados das ações A e B sejam proporcionais. Portanto, se o spread seguir um processo estocástico de reversão média, é possível explorar desvios de curto prazo em relação às equações (1) e (2) negociando essas ações por meio de posições compradas e vendidas.\n",
    "\n",
    "Com o objetivo de ilustrar o conceito por trás dessa estratégia, suponhamos que, em um determinado momento, o $\\beta$ e o spread sejam positivos e distantes de sua média. Por hipótese, o spread é estacionário e a equação (1) sugere que a reversão à média só pode ocorrer se a diferença $[ln A_t - \\beta(ln B_t)]$ diminuir, pois $\\alpha$ é uma constante. Assim, existem apenas cinco cenários nos quais isso ocorre: (i) $ln A_t$ declina e $ln B_t$ aumenta; (ii) $ln A_t$ declina, enquanto $ln B_t$ permanece inalterado; (iii) $ln B_t$ aumenta, enquanto  $ln A_t$ permanece inalterado; (iv) ambos os preços logarítmicos aumentam, mas $ln A_t$ cresce menos que $\\beta (ln B_t)$; (v) ambos os log preços caem, mas $\\beta (ln B_t)$ cai mais lentamente que em $ln A_t$. Assim, a estratégia de vender ativo A e comprar ativo B em uma proporção adequada pode produzir retornos positivos em qualquer desses cenários. Assim, para cada real apostado na depreciação da ação A, deve-se apostar $\\beta$ reais na valorização do ação B.\n",
    "\n",
    "Uma vez modelado o spread, a estratégia de negociação de pares pode ser implementada. O modelo mais usado com essa intenção é o de limite fixo (Fixed Threshold), uma abordagem não paramétrica que estabelece como orientação de negociação dois limites externos fixos equivalentes em módulo ao desvio padrão histórico do spread multiplicado por um fator predeterminado, geralmente definido igual a 2, mas com sinais opostos (Gatev et al. 2006). Devido à sua essência e simplicidade não paramétricas, a implementação do modelo é direta. De fato, analogamente ao exemplo discutido anteriormente, para $\\beta > 0$, sempre que o spread ultrapassa o limite superior (positivo), uma posição comprada na ação B e uma posição vendida na ação A são abertas, com pesos variando de acordo com o beta encontrado via equação (1), esperando a convergência do spread para seu valor médio. Segue-se naturalmente que essas posições devem ser invertidas quando o spread atingir o limiar oposto (negativo). Para $\\beta < 0$, as posições na ação B devem ser invertidas em cada caso, ou seja, a posição da ação deve ser reduzida quando o spread ultrapassar o limite superior, enquanto uma posição comprada deve ser aberta quando o spread atingir o limite inferior.\n",
    "\n",
    "Assim, nesse projeto, o cálculo dos spreads entre as ações do par foi realizado utilizando o método de filtro de Kalman e a estratégia implementada baseou-se em modelo com thresholds para identificação dos pontos de abertura e fechamento do trade, já citado anteriormente. No caso do filtro de Kalman, este foi configurado de tal forma que os coeficientes da regressão linear fossem variáveis no tempo e assumissem o comportamento de um passeio aleatório, conforme verificado abaixo em (3) e (4):\n",
    "\n",
    "\\begin{align}\n",
    "\\alpha_t = \\delta_1 + \\alpha_{t-1} + w_t \\\\\n",
    "\\beta_t = \\delta_2 + \\beta_{t-1} + z_t\n",
    "\\end{align}\n",
    "\n",
    "Onde $\\delta_1$ e $\\delta_2$ são parâmetros a serem estimados e $w_t$ e $z_t$ são ruídos brancos não correlacionados.\n",
    "\n",
    "Os dados utilizados no projeto compreendem as séries históricas de preços das ações ordinárias e preferenciais da Petrobras (PETR3 e PETR4). Foram coletados preços diários de janeiro de 2008 a março de 2020. A escolha dessas ações se justifica por serem ações da mesma empresa e, portanto, deveriam apresentar cointegração. \n",
    "\n",
    "Os parâmetros utilizados nesse trabalho consistem em uma definição das janelas de estimação dos parâmetros da regressão e de negociação do par. O tamanho da janela de estimação, definida no trabalho como janela_dados, foi calibrado de modo que houvesse um volume de dados razoável para ajuste do filtro de Kalman. Foi escolhida uma janela de 1 ano em dias úteis para essa análise, equivalente a 252 dias. \n",
    "\n",
    "Além disso, foi definida uma janela para determinar se o spread era estacionário ou não, chamadada de janela de cointegração (janela_coint). Nessa janela, a conclusão de cointegração para o dia em análise é encontrada somente se, em todos os dias da janela, o teste ADF indicar estacionariedade do spread. Para isso, foi escolhida empiricamente uma janela de 5 dias.\n",
    "\n",
    "Ademais, como premissa do trabalho, assumimos um custo de execução igual a zero e um nível de significância alfa igual a 1% (a ser utilizado no teste ADF para verificação da cointegração). Para a regra de trading, também foi definido um threshold de distância de 0.25 desvio-padrão da média para abertura de posição. Por fim, também foi definida uma variável para reduzir a frequência de ajustes de posição (delta_beta_lim), que determina um intervalo de variação para o beta dentro do qual não há ajuste das posições em relação àquelas quando o par foi aberto - para ilustração: um delta_beta_lim igual a 0.10 significaria que a posição não sofreria alteração se o beta abrisse para menos desse valor. Com esta medida, impedimos que haja variação para qualquer movimento do beta dentro deste intervalo, limitando-o.\n",
    "\n",
    "A figura abaixo ilustra a estratégia de pairs trading com thresholds. Ela mostra que as posições são abertas quando o spread se situa nas regiões inferior ou superior de fora da banda delimitada pelos threholds, definidos com um múltiplo do desvio-padrão do spread. Ademais, as posições são fechadas quando o spread cruza o eixo 0. \n",
    "\n",
    "Quando um par é aberto, o tamanho da posição não sofre ajuste até que haja convergência do spread para  a média, mesmo que o spread continue se desviando da média, tornando a estratégia teoricamente mais atrativa. O único ajuste implementado é um ajuste decorrente da variação do beta ao longo do tempo, que determina a proporção relativa das ações do par que devem ser compradas."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<PIL.PngImagePlugin.PngImageFile image mode=RGBA size=1590x784 at 0x256B0703F88>"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from PIL import Image\n",
    "str_dir = 'D:/Estudos/2 Mestrado/Insper/Disciplinas/Finanças Quantitativas/Projeto/'\n",
    "myImage = Image.open(str_dir + 'Figura_Pairs_Trading.png')\n",
    "myImage"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Bibliotecas"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Abaixo estão todas as bibliotecas utilizadas no código."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import scipy as sp\n",
    "from statsmodels.tsa.stattools import adfuller\n",
    "from statsmodels.graphics.tsaplots import plot_acf, plot_pacf\n",
    "from pykalman import KalmanFilter\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Configurações"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As configurações necessárias às bibliotecas empregadas no modelo estão consolidadas abaixo."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Comando para que os gráficos sejam exibidos no corpo do texto\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.rcParams['figure.dpi'] = 200 # Resolução dos gráficos impressos no pdf (dpi)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Estilo dos gráficos\n",
    "sns.set(color_codes = True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Parâmetros"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Esses são os parâmetros referentes aos modelos de pairs trading que são implementados no código, a serem detalhados mais a frente."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "alfa = 0.01 # Nível de significância utilizado nos testes de hipótese \n",
    "janela_dados = 252 # tamanho da janela de dados utilizada para estimação dos modelos\n",
    "janela_coint = 5 # janela de dias que será analisada para determinação da cointegração das ações de um par\n",
    "cost = 0 # custo de execução \n",
    "th_KF = 0.25 # Número de desvios-padrão de distância da média para abertura de posição usando KF\n",
    "delta_beta_lim = 0 #variável para reduzir a frequência de ajustes de posição"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Importação e Tratamento dos Dados"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Leitura da planilha com as séries de preços diários das ações de cada par\n",
    "\n",
    "str_plan_bd_stocks = 'BD_Stocks_v4'\n",
    "\n",
    "df_prices = pd.read_csv(filepath_or_buffer = str_dir + str_plan_bd_stocks + '.csv', sep=';')\n",
    "df_prices.index = df_prices['Dates']\n",
    "df_prices.index = pd.to_datetime(df_prices['Dates'])\n",
    "df_prices = df_prices.drop(columns=['Dates'])\n",
    "df_prices = df_prices.dropna()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>PETR3 BZ Equity</th>\n",
       "      <th>PETR4 BZ Equity</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Dates</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2008-01-02</th>\n",
       "      <td>3.954220</td>\n",
       "      <td>3.772761</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2008-01-03</th>\n",
       "      <td>3.939638</td>\n",
       "      <td>3.756538</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2008-01-04</th>\n",
       "      <td>3.878466</td>\n",
       "      <td>3.706228</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2008-01-07</th>\n",
       "      <td>3.843530</td>\n",
       "      <td>3.676427</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2008-01-08</th>\n",
       "      <td>3.901973</td>\n",
       "      <td>3.729901</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            PETR3 BZ Equity  PETR4 BZ Equity\n",
       "Dates                                       \n",
       "2008-01-02         3.954220         3.772761\n",
       "2008-01-03         3.939638         3.756538\n",
       "2008-01-04         3.878466         3.706228\n",
       "2008-01-07         3.843530         3.676427\n",
       "2008-01-08         3.901973         3.729901"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Log preços - serão usados nos testes de cointegração\n",
    "df_log_prices = np.log(df_prices)\n",
    "df_log_prices.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Cálculo dos Spreads"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Regressão Usando Filtro de Kalman"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "A função abaixo implementa a regressão do log preço de uma ação contra o log preço da outra ação do par utilizando o filtro de Kalman. Ela retorna os coeficientes estimados ao longo do tempo, o resíduo (spread), as matrizes de covariância dos coeficientes e um objeto contendo o filtro de Kalman ajustado.\n",
    "\n",
    "Essa função é empregada na função spread_KF para computar o spread de cada par de ações da base."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "'''Parâmetros da função:\n",
    "-> y, x : log preços do par de ações.\n",
    "-> delta: parâmetro relacionado à variância. Quanto mais próximo de 1, mais voláteis são os betas, melhor o ajuste\n",
    "porém pior a capacidade preditiva do filtro. Quanto mais próximo de zero, mais estáveis são os betas, porém a capacidade \n",
    "preditiva é ruim. Há então um ponto ótimo no meio desse intervalo. \n",
    "-> obs_cov: matriz de covariância das observações - variáveis de estado.\n",
    "-> init_mean, init_cov: são os valores iniciais da média e da matriz de covariância dos betas para inicialização do filtro.\n",
    "-> intercept: se você deseja rodar regressão com (TRUE) ou sem (FALSE) intercepto.\n",
    "-> em: se você deseja utilizar o algoritmo \"em\" para estimar os valores iniciais da média e da matriz de covariância dos betas.\n",
    "''' \n",
    "\n",
    "def KFReg(y, x, delta, obs_cov, init_mean, init_cov, intercept = False, em = False):\n",
    "    if intercept == True: #caso regressão utilize intercepto\n",
    "        obs_mat = np.vstack([x, np.ones(x.shape)]).T[:, np.newaxis] \n",
    "        if em == False:\n",
    "            trans_cov = delta / (1 - delta) * np.eye(2)\n",
    "            kf = KalmanFilter(n_dim_obs = 1, \n",
    "                              n_dim_state = 2, \n",
    "                              initial_state_mean = init_mean,\n",
    "                              initial_state_covariance = init_cov,\n",
    "                              transition_matrices = np.eye(2), # Essa é a matriz que define o random walk\n",
    "                              observation_matrices = obs_mat,\n",
    "                              observation_covariance = obs_cov,\n",
    "                              transition_covariance = trans_cov)\n",
    "            state_means, state_covs = kf.filter(y.values)\n",
    "        else:\n",
    "            kf = KalmanFilter(n_dim_obs = 1, \n",
    "                              n_dim_state = 2, \n",
    "                              initial_state_mean = init_mean, \n",
    "                              initial_state_covariance = init_cov,\n",
    "                              observation_matrices = obs_mat)\n",
    "            state_means, state_covs = kf.filter(y.values)\n",
    "        beta0 = state_means[:,1]\n",
    "        beta1 = state_means[:,0]\n",
    "        resid = y - beta0 - beta1*x\n",
    "        return beta0, beta1, resid, state_covs, kf\n",
    "    else:\n",
    "        obs_mat = np.vstack([x]).T[:, np.newaxis]\n",
    "        if em == False:\n",
    "            trans_cov = delta / (1 - delta)\n",
    "            kf = KalmanFilter(n_dim_obs = 1, \n",
    "                              n_dim_state = 1, \n",
    "                              initial_state_mean = init_mean,\n",
    "                              initial_state_covariance = init_cov,\n",
    "                              transition_matrices = 1,\n",
    "                              observation_matrices = obs_mat,\n",
    "                              observation_covariance = obs_cov,\n",
    "                              transition_covariance = trans_cov)\n",
    "            state_means, state_covs = kf.filter(y.values)\n",
    "        else:\n",
    "            kf = KalmanFilter(n_dim_obs = 1, n_dim_state = 1, \n",
    "                              initial_state_mean = init_mean, \n",
    "                              initial_state_covariance = init_cov,\n",
    "                              observation_matrices = obs_mat)\n",
    "            state_means, state_covs = kf.em(y.values).filter(y.values)\n",
    "        beta1 = state_means[:,0]\n",
    "        resid = y - beta1*x\n",
    "        return beta1, resid, state_covs, kf"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Cálculo do Spread Usando Filtro de Kalman"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Os parâmetros são os mesmos utilizados na função da regressão do filtro de Kalman, com exceção de:\n",
    "# janela_dados: define o tamanho da janela móvel\n",
    "\n",
    "def spread_KF(log_price1, log_price2, janela_dados, delta, obs_cov, init_mean0, init_cov, intercept = False, em = False):\n",
    "    \n",
    "    y = log_price1\n",
    "    x = log_price2\n",
    "    T = len(y)\n",
    "    \n",
    "    s = pd.DataFrame(np.NaN, index = y.index, columns = ['Spread'])\n",
    "    adf_res = pd.DataFrame(np.NaN, index = np.arange(0,T), columns = ['ADF'])\n",
    "    beta0 = pd.DataFrame(np.NaN, index = np.arange(0,T), columns = ['Beta0'])\n",
    "    beta1 = pd.DataFrame(np.NaN, index = np.arange(0,T), columns = ['Beta1'])\n",
    "    \n",
    "    inicio = 0\n",
    "    fim = janela_dados\n",
    "    i = 0\n",
    "    init_mean = init_mean0\n",
    "    \n",
    "    while fim <= T:\n",
    "        \n",
    "        x_array = x.iloc[np.arange(inicio,fim)]\n",
    "        y_array = y.iloc[np.arange(inicio,fim)]\n",
    "        res = KFReg(y = y_array, x = x_array, \n",
    "                    delta = delta, obs_cov = obs_cov, \n",
    "                    init_mean = init_mean, init_cov = init_cov, \n",
    "                    intercept = intercept, em = em)\n",
    "        if intercept == False:\n",
    "            beta1.iloc[np.arange(inicio,fim),:] = pd.DataFrame(res[0]).values\n",
    "            s.iloc[np.arange(inicio,fim),:] = pd.DataFrame(res[1]).values\n",
    "            adf_res.iloc[fim-1,0] = 1*(adfuller(x = res[1], \n",
    "                                            regression = 'c', \n",
    "                                            autolag = 'BIC')[1] < alfa)\n",
    "        else:\n",
    "            beta0.iloc[np.arange(inicio,fim),:] = pd.DataFrame(res[0]).values\n",
    "            beta1.iloc[np.arange(inicio,fim),:] = pd.DataFrame(res[1]).values\n",
    "            s.iloc[np.arange(inicio,fim),:] = pd.DataFrame(res[2]).values\n",
    "            adf_res.iloc[fim-1,0] = 1*(adfuller(x = res[2], \n",
    "                                            regression = 'c', \n",
    "                                            autolag = 'BIC')[1] < alfa)\n",
    "        \n",
    "        inicio = inicio + 1\n",
    "        if intercept == False:\n",
    "            init_mean = beta1.iloc[inicio,0]\n",
    "        else:\n",
    "            init_mean = np.array([beta1.iloc[inicio,0],\n",
    "                                    beta0.iloc[inicio,0]])\n",
    "        fim = fim + 1\n",
    "        i = i + 1\n",
    "    \n",
    "    s = s.dropna()\n",
    "    \n",
    "    if intercept == False:\n",
    "        return s, adf_res, beta1\n",
    "    else:\n",
    "        return s, adf_res, beta0, beta1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Exemplo de Cálculo de Spread - PETR3 x PETR4"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Nessa seção, realizamos um exemplo de cálculo de spread usando o par PETR3 com PETR4. Os resultados serão salvos em um arquivo .csv, a ser utilizado no notebook parte 2."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Cálculo do Retorno Acumulado para as Séries de Preço"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Cálculo do retorno acumulado para as séries de preço\n",
    "x = df_log_prices['PETR4 BZ Equity']\n",
    "x = x.diff()\n",
    "x[0] = 0 #Para casos de não intercepto na regressão do filtro de Kalman, as séries de preço são normalizadas começando em zero\n",
    "x = x.cumsum()\n",
    "y = df_log_prices['PETR3 BZ Equity']\n",
    "y = y.diff()\n",
    "y[0] = 0 #Para casos de não intercepto na regressão do filtro de Kalman, as séries de preço são normalizadas começando em zero\n",
    "y = y.cumsum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Cálculo do Spread das Séries de Preço usando Filtro de Kalman\n",
    "\n",
    "Foi utilizado Filtro de Kalman como intercepto, em concordância com o modelo que foi apresentado na introdução."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Cálculo do spread usando KF com intercepto\n",
    "# Valores calibrados empiricamente\n",
    "res_KF = spread_KF(y, x, janela_dados, \n",
    "                delta = 1e-2, obs_cov = 0.25, \n",
    "                init_mean0 = np.array([1,0]), init_cov = np.matrix([[0.25,0],[0,0.25]]), \n",
    "                intercept = True, em = False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Variáveis que Guardam Resultados do Cálculo do Spread"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Variáveis que guardam os resultados do cálculo do spread usando KF\n",
    "s_KF = res_KF[0]\n",
    "adf_KF = res_KF[1]\n",
    "beta_KF = res_KF[3]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Rotina para Guardar os Resultados em .csv"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "s_KF.to_csv(path_or_buf = str_dir + 's_KF.csv')\n",
    "adf_KF.to_csv(path_or_buf = str_dir + 'adf_KF.csv')\n",
    "beta_KF.to_csv(path_or_buf = str_dir + 'beta_KF.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Referências Bibliográficas\n",
    "\n",
    "CALDEIRA, J. F.; MOURA, G. V. Selection of a portfolio of pairs based in cointegration: A statistical arbitrage strategy. Revista Brasileira de Finanças, v. 11, n. 1, p. 49-80, mar. 2013.\n",
    "\n",
    "CARHART, M. M., On persistence in mutual fund performance. Journal of Finance, v. 52, n. 1, p. 57-82, mar. 1997.\n",
    "\n",
    "FAMA, E. F., FRENCH, K. F., Common risk factors in the returns on stocks and bonds. Journal of Financial Economics, v.33, n.1, p. 3-56, fev. 1993.\n",
    "\n",
    "GATEV, E., $et.$ $al$. Pairs trading: performance of a relative-value arbitrage rule. Review of Financial Studies, v. 19, n. 3, p. 797-827, mar. 2006."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}