黄金期货研究201901.zip

# encoding: gbk import pandas as pd import numpy as np import arrow import re import matplotlib.pyplot as plt import time # 统计包 # from scipy import stats import statsmodels.api as sm import statsmodels.stats.diagnostic '''黄金建模专用脚本''' # 伦敦金万德获取代码 # w_wsd_data = vba_wsd("SPTAUUSDOZ.IDC","close","2016-01-01","2019-01-11", "TradingCalendar=SHFE", w_wsd_codes, w_wsd_fields, w_wsd_times, w_wsd_errorid) # 上期黄金获取代码 # w_wsd_data = vba_wsd("AU1606.SHF,AU1612.SHF,AU1706.SHF,AU1712.SHF,AU1806.SHF,AU1812.SHF,AU1906.SHF","close","2016-01-01","2019-01-11", "TradingCalendar=SHFE", w_wsd_codes, w_wsd_fields, w_wsd_times, w_wsd_errorid) # 上期黄金成交量获取代码 # w_wsd_data = vba_wsd("AU1606.SHF,AU1612.SHF,AU1706.SHF,AU1712.SHF,AU1806.SHF,AU1812.SHF,AU1906.SHF","volume","2016-01-01","2019-01-11", "TradingCalendar=SHFE", w_wsd_codes, w_wsd_fields, w_wsd_times, w_wsd_errorid) # 常量 # 文件名 shfeDataFileName = 'shfeAu20160101_20190111.csv' shfeIndexDataFileName = 'SHFEAuIndex.csv' xauUsdDataFileName = 'xauusd20160101_20190111.csv' forexCNYUSDFileName = 'CNYUSDforex.csv' # 综合月 adjMonthDict = {'5月':5, '11月':11} # 合约顺序 ctrOrderList = ['AU1606.SHF','AU1612.SHF','AU1706.SHF','AU1712.SHF','AU1806.SHF','AU1812.SHF','AU1906.SHF'] volOrderList = ['AU1606.VOL','AU1612.VOL','AU1706.VOL','AU1712.VOL','AU1806.VOL','AU1812.VOL','AU1906.VOL'] # 列名 dateTimeName = 'DateTime' indexPriceName = 'indexPrice' xauusdPriceName = 'CLOSE' # 先处理数据 # 品种指数编制：参照中信建投期货的主力合约换月规则，拼接数据 class shfeFuturesIndexGenerate(): def __init__(self,dataFileName=str(),ctrOrder=ctrOrderList,volOrder=volOrderList): self.data = pd.read_csv(dataFileName,encoding='gbk') self.ctrOrderList = ctrOrder self.volOrderList = volOrder def dataCombine(self): # 按合约顺序提取和拼接数据 catIndexData = {dateTimeName:[],indexPriceName:[]} # 根据价格列名进行循环 j = 0 for i in range(len(ctrOrderList)): # # 按照合约顺序开始抓合约价格数据和日期数据 # tempdata = self.data.loc[:,[dateTimeName,self.ctrOrderList[i]]] # 目前处理的是合约名字 tempCatName = self.ctrOrderList[i] tempVolName = self.volOrderList[i] # 获得当前合约的交割月 settleMonth = int(re.sub("\D","",tempCatName)[-2:]) # 获得当前合约的年份 settleYear = float(re.sub("\D","",tempCatName))/100 settleYear = round(settleYear) # 日期循环 while True: # 抓日期和价格 tempDate = self.data[dateTimeName][j] tempPrice = self.data[tempCatName][j] # 抓日期的月份 tempDateMonth = arrow.get(tempDate).month # 抓日期的年份 tempDateYear = int(str(arrow.get(tempDate).year)[-2:]) # 对月份和年份进行判断，看是否要换月 if tempDateMonth >= settleMonth: if tempDateYear >= settleYear: break # 对月份进行判断，看是否需要加权 if tempDateMonth == adjMonthDict['5月'] or tempDateMonth == adjMonthDict['11月']: # 目前合约的价格和成交量 tempPrice1 = tempPrice tempVol1 = self.data[tempVolName][j] # 抓下一个合约的价格和成交量 catName2 = self.ctrOrderList[i + 1] volName2 = self.volOrderList[i + 1] tempPrice2 = self.data[catName2][j] tempVol2 = self.data[volName2][j] # 合成加权平均价格 weight1 = tempVol1 / (tempVol1 + tempVol2) weight2 = tempVol2 / (tempVol1 + tempVol2) tempPrice = tempPrice1 * weight1 + tempPrice2 * weight2 tempPrice = round(tempPrice,2) # 先插入价格 catIndexData[indexPriceName].append(tempPrice) # 最后加入日期 tempDate = time.strptime(tempDate,'%Y/%m/%d') tempDate = time.strftime('%Y/%m/%d',tempDate) catIndexData[dateTimeName].append(tempDate) # 行数递增 j += 1 if j == self.data.shape[0]: break catIndexData = pd.DataFrame(catIndexData) return catIndexData def getNewDateFile(self): data = self.dataCombine() data.to_csv('SHFEAuIndex.csv',index=False) # 然后开始建模 class getVarModel(): # def __init__(self, shfeAuIndexFileName=shfeIndexDataFileName, xauusdFileName=xauUsdDataFileName, inSampleStart='2016-1-4',inSampleEnd='2018-9-28'): def __init__(self, shfeAuIndexFileName=shfeIndexDataFileName, xauusdFileName=xauUsdDataFileName, inSampleStart='2018-1-1',inSampleEnd='2018-9-28'): self.shfeIndexData = pd.read_csv(shfeAuIndexFileName) self.xauusdData = pd.read_csv(xauusdFileName) self.forexData = pd.read_csv(forexCNYUSDFileName) # 处理数据，用日期做index self.shfeIndexData[dateTimeName] = pd.to_datetime(self.shfeIndexData[dateTimeName]) self.shfeIndexData = self.shfeIndexData.set_index(dateTimeName) self.xauusdData[dateTimeName] = pd.to_datetime(self.xauusdData[dateTimeName]) self.xauusdData = self.xauusdData.set_index(dateTimeName) self.forexData[dateTimeName] = pd.to_datetime(self.forexData[dateTimeName]) self.forexData = self.forexData.set_index(dateTimeName) # 样本内数据,构造Series类型数据 self.shfeIndexInSampleData = self.shfeIndexData[inSampleStart:inSampleEnd] self.shfeIndexInSampleData = pd.Series(self.shfeIndexInSampleData[indexPriceName], index=self.shfeIndexInSampleData.index) self.xauusdDataInSampleData = self.xauusdData[inSampleStart:inSampleEnd] self.xauusdDataInSampleData = pd.Series(self.xauusdDataInSampleData[xauusdPriceName], index=self.xauusdDataInSampleData.index) self.forexDataInSampleData = self.forexData[inSampleStart:inSampleEnd] self.forexDataInSampleData = pd.Series(self.forexDataInSampleData['OPEN'], index=self.forexDataInSampleData.index) # 样本个数 self.sampleNum = self.shfeIndexInSampleData.shape[0] # 价格走势图 def price_trend_draw(self): # 处理数据,化为人民币和1g # changeXAUUSD = np.array(self.xauusdDataInSampleData) / 28.3495 * 6.78 font1 = {'size':23} changeXAUUSD = np.multiply(np.array(self.xauusdDataInSampleData) / 28.3495 , np.array(self.forexDataInSampleData)) changeXAUUSD = np.round(changeXAUUSD,2) shfeXAU = np.array(self.shfeIndexInSampleData) # 算相关系数，一定要是pd.Series类型数据才可以 correlation = round(pd.Series(shfeXAU).corr(pd.Series(changeXAUUSD)),6) # print(changeXAUUSD) fig = plt.figure(figsize=(12,8)) plt.plot(changeXAUUSD,'r',label='XAU USD') plt.plot(shfeXAU,'g',label='SHFE XAU') plt.title('Correlation: ' + str(correlation),font1) plt.grid(True) plt.axis('tight') plt.legend(loc=0,prop=font1) plt.ylabel('Price',font1) plt.show() # plt.plot() # 对数一阶差分处理 def logdiff(self,data): logData = np.log(data) logDiffData = np.diff(logData) return logDiffData # 数据稳定性检验 def adftest(self,data,maxlags): adfResult = sm.tsa.stattools.adfuller(data,maxlags) output = pd.DataFrame(index=['Test Statistic Value', "p-value", "Lags Used", "Number of Observations Used", "Critical V