如何使用Python計算還原股價

最近在開發量化投資系統，要做量化投資最重要的就是對股票進行回測，而回測計算報酬率的時候，必須計算還原股價 ，估計公司沒有除權息調整時的股價，所以這篇文章就來教大家如何用Python計算還原股價。

資料的方面我們使用FindMind所提供的開源數據集，來做計算~

一. 取得股票各年股利分配情況

首先先取得股票的各年股利分配情況的資料，那我們這邊使用台積電(2330)，作為範例，參數中的token需要到FindMind的官網去 申請帳號取得。

	import requests
	import pandas as pd
	def get_dividend_data():
	url = "https://api.finmindtrade.com/api/v4/data"
	parameter = {
	"dataset": "TaiwanStockDividend",
	"data_id": "2330",
	"start_date": "2000-01-01",
	"token":'your token'
	}
	dividend_data = requests.get(url, params=parameter)
	dividend_data = dividend_data.json()
	dividend_data = pd.DataFrame(dividend_data['data'])
	return dividend_data

view raw get_dividend_data.py hosted with ❤ by GitHub

二. 取得股票的股價資料

一樣也是從FindMind官網去取得資料，和取得股利資料一樣。

	def get_stock_data():
	url = "https://api.finmindtrade.com/api/v4/data"
	parameter = {
	"dataset": "TaiwanStockPrice",
	"data_id": "2330",
	"start_date": "2006-01-01",
	"token":'your_token'
	}
	resp = requests.get(url, params=parameter)
	stock_data = resp.json()
	stock_data = pd.DataFrame(stock_data["data"])
	return stock_data

view raw get_stock_data.py hosted with ❤ by GitHub

三. 股利資料欄位介紹以及計算調整股價

首先我們計算還原股價會用到的是以下四個欄位:

1. CashExDividendTradingDate : 除息交易日

2. StockExDividendTradingDate : 除權交易日

3. CashEarningsDistribution : 現金股利

4. StockEarningsDistribution + StockStatutorySurplus : 股票股利

計算調整股價公式:

主要分成兩部分:

a. 計算除權因子: 在發放股票股利那天以前(包含當天)的股價都要乘上除權因子去調整股價。

$$\mathrm{除}\mathrm{權}\mathrm{因}\mathrm{子}=1/(1+\mathrm{股}\mathrm{票}\mathrm{股}\mathrm{利}/10)$$

b. 計算除息因子:

這裡有人可能會直接用收盤價扣現金股利，但這樣會導致股價調整前和調整後日收益率改變，但用下面的方法就可以讓日收益率不變， 因為我們是去計算發放現金股利後與原本股價的改變比例，而這裡t-1代表的是昨日的價格，在發放現金股利那天以前的股價 都要乘上除息因子去調整股價。

$$\mathrm{除}\mathrm{息}\mathrm{因}\mathrm{子}=(\mathrm{收}\mathrm{盤}{\mathrm{價}}_{t-1}-\mathrm{現}\mathrm{金}\mathrm{股}\mathrm{利})/(\mathrm{收}\mathrm{盤}{\mathrm{價}}_{t-1})$$

程式計算說明

首先將股利資料分別分成現金股利以及股票股利，並且將這兩個資料合併到股票價格資料上，因為股利資料會有先公布的資料， 也就是未來的資料，所以合併的同時也可以使未來的資料刪掉。

	dividend_data = get_dividend_data()
	stock_data = get_stock_data()
	# 將資料分成現金股利和股票股利
	stock_dividend_data = dividend_data[['StockExDividendTradingDate', 'StockEarningsDistribution', 'StockStatutorySurplus']]
	stock_dividend_data = stock_dividend_data.rename(columns={"StockExDividendTradingDate":'date'})
	cash_dividend_data = dividend_data[['CashExDividendTradingDate', 'CashEarningsDistribution']]
	cash_dividend_data = cash_dividend_data.rename(columns={"CashExDividendTradingDate":'date'})
	# 順便也將未來要發股利的去掉了
	stock_data = stock_data.merge(stock_dividend_data, on='date', how='left')
	stock_data = stock_data.merge(cash_dividend_data, on='date', how='left')

view raw cal_adjusted_price1.py hosted with ❤ by GitHub

然後在按照公式分別計算除權因子和除息因子

	# 計算除權因子
	stock_dividend_factor = 1 / (1 + (stock_data.loc[stock_data['StockEarningsDistribution'].notnull(), 'StockEarningsDistribution'] + stock_data.loc[stock_data['StockEarningsDistribution'].notnull(), 'StockStatutorySurplus']) / 10)
	stock_data.loc[stock_dividend_factor.index-1, 'stock_dividend_factor'] = stock_dividend_factor.values
	# 計算除息因子
	stock_data['previous_close_price'] = stock_data['close'].shift(1)
	cash_dividend_factor = (stock_data.loc[stock_data['CashEarningsDistribution'].notnull(), 'previous_close_price'] - stock_data.loc[stock_data['CashEarningsDistribution'].notnull(), 'CashEarningsDistribution']) / stock_data.loc[stock_data['CashEarningsDistribution'].notnull(), 'previous_close_price']
	stock_data.loc[cash_dividend_factor.index-1, 'cash_dividend_factor'] = cash_dividend_factor.values
	cash_dividend_index = stock_data.loc[stock_data['CashEarningsDistribution'].notnull(), 'cash_dividend_factor'].index

view raw cal_adjusted_price2.py hosted with ❤ by GitHub

然後將每一期的除息因子都做累積乘積，這樣就可以將每一期調整的程度反映到過去的股價上，算完除息因子後乘上除權因子 就可以得到總因子，再乘上原始股價就可以得到調整股價，因為有些是在除權後的股價，因此就是填充為原始股價。

	# 累積總因子
	cum_prod_cash_dividend = stock_data.loc[stock_data['cash_dividend_factor'].notnull(), 'cash_dividend_factor'][::-1].cumprod()[::-1]
	stock_data.loc[cash_dividend_factor.index-1, 'cash_dividend_factor'] = cum_prod_cash_dividend.values
	stock_data['cash_dividend_factor'] = stock_data['cash_dividend_factor'].fillna(method='backfill')
	stock_data['stock_dividend_factor'] = stock_data['stock_dividend_factor'].fillna(method='backfill')
	stock_data['cash_dividend_factor'] = stock_data['cash_dividend_factor'].fillna(method='backfill')
	stock_data['stock_dividend_factor'] = stock_data['stock_dividend_factor'].fillna(method='backfill').fillna(1)
	stock_data['total_factor'] = stock_data['cash_dividend_factor'] * stock_data['stock_dividend_factor']
	stock_data['adjust_price'] = stock_data['close'] * stock_data['total_factor']
	# 還沒到除息日的股價用現在的股價填充
	stock_data.loc[stock_data['adjust_price'].isnull(), 'adjust_price'] = stock_data.loc[stock_data['adjust_price'].isnull(), 'close']

view raw cal_adjusted_price3.py hosted with ❤ by GitHub

那以下就是完整的程式碼，如果有什麼問題歡迎在下面留言~

	import requests
	import pandas as pd
	def get_dividend_data():
	url = "https://api.finmindtrade.com/api/v4/data"
	parameter = {
	"dataset": "TaiwanStockDividend",
	"data_id": "2330",
	"start_date": "2000-01-01",
	"token":'your token'
	}
	dividend_data = requests.get(url, params=parameter)
	dividend_data = dividend_data.json()
	dividend_data = pd.DataFrame(dividend_data['data'])
	return dividend_data
	def get_stock_data():
	url = "https://api.finmindtrade.com/api/v4/data"
	parameter = {
	"dataset": "TaiwanStockPrice",
	"data_id": "2330",
	"start_date": "2006-01-01",
	"token":'your token'
	}
	resp = requests.get(url, params=parameter)
	stock_data = resp.json()
	stock_data = pd.DataFrame(stock_data["data"])
	return stock_data
	def cal_adjusted_price(stock_data, dividend_data):
	stock_dividend_data = dividend_data[['StockExDividendTradingDate', 'StockEarningsDistribution', 'StockStatutorySurplus']]
	stock_dividend_data = stock_dividend_data.rename(columns={"StockExDividendTradingDate":'date'})
	cash_dividend_data = dividend_data[['CashExDividendTradingDate', 'CashEarningsDistribution']]
	cash_dividend_data = cash_dividend_data.rename(columns={"CashExDividendTradingDate":'date'})
	# 順便也將未來要發股利的去掉了
	stock_data = stock_data.merge(stock_dividend_data, on='date', how='left')
	stock_data = stock_data.merge(cash_dividend_data, on='date', how='left')
	# 計算除權因子
	stock_dividend_factor = 1 / (1 + (stock_data.loc[stock_data['StockEarningsDistribution'].notnull(), 'StockEarningsDistribution'] + stock_data.loc[stock_data['StockEarningsDistribution'].notnull(), 'StockStatutorySurplus']) / 10)
	stock_data.loc[stock_data['StockEarningsDistribution'].notnull(), 'stock_dividend_factor'] = stock_dividend_factor
	# 計算除息因子
	stock_data['previous_close_price'] = stock_data['close'].shift(1)
	cash_dividend_factor = (stock_data.loc[stock_data['CashEarningsDistribution'].notnull(), 'previous_close_price'] - stock_data.loc[stock_data['CashEarningsDistribution'].notnull(), 'CashEarningsDistribution']) / stock_data.loc[stock_data['CashEarningsDistribution'].notnull(), 'previous_close_price']
	stock_data.loc[stock_data['CashEarningsDistribution'].notnull(), 'cash_dividend_factor'] = cash_dividend_factor
	cash_dividend_index = stock_data.loc[stock_data['CashEarningsDistribution'].notnull(), 'cash_dividend_factor'].index
	# 累積總因子
	cum_prod_cash_dividend = stock_data.loc[stock_data['cash_dividend_factor'].notnull(), 'cash_dividend_factor'][::-1].cumprod()[::-1]
	cum_prod_cash_dividend.index = cash_dividend_index
	stock_data.loc[stock_data['CashEarningsDistribution'].notnull(), 'cash_dividend_factor'] = cum_prod_cash_dividend
	stock_data['cash_dividend_factor'] = stock_data['cash_dividend_factor'].fillna(method='backfill')
	stock_data['stock_dividend_factor'] = stock_data['stock_dividend_factor'].fillna(method='backfill')
	stock_data['cash_dividend_factor'] = stock_data['cash_dividend_factor'].fillna(method='backfill')
	stock_data['stock_dividend_factor'] = stock_data['stock_dividend_factor'].fillna(method='backfill').fillna(1)
	stock_data['total_factor'] = stock_data['cash_dividend_factor'] * stock_data['stock_dividend_factor']
	stock_data['adjust_price'] = stock_data['close'] * stock_data['total_factor']
	# 還沒到除息日的股價用現在的股價填充
	stock_data.loc[stock_data['adjust_price'].isnull(), 'adjust_price'] = stock_data.loc[stock_data['adjust_price'].isnull(), 'close']
	return stock_data
	if __name__ == "__main__":
	dividend_data = get_dividend_data()
	stock_data = get_stock_data()
	stock_data = cal_adjusted_price(stock_data, dividend_data)

view raw cal_adjusted_price.py hosted with ❤ by GitHub

喜歡我的文章可以幫我拍拍手哦~~~