文章目录

1. 代码

一个股票数据（沪深）爬虫和选股策略测试框架，数据基于雅虎YQL和新浪财经。

根据选定的日期范围抓取所有沪深两市股票的行情数据。
根据指定的选股策略和指定的日期进行选股测试。
计算选股测试实际结果（包括与沪深300指数比较）。
保存数据到JSON文件、CSV文件。
支持使用表达式定义选股策略。
支持多线程处理。

代码

项目结构

main.py

from stockholm import Stockholm
import option
import os

def checkFoldPermission(path):
    if(path == 'USER_HOME/tmp/stockholm_export'):
        path = os.path.expanduser('~') + '/tmp/stockholm_export'
    try:
        if not os.path.exists(path):
            os.makedirs(path)
        else:
            txt = open(path + os.sep + "test.txt","w")
            txt.write("test")
            txt.close()
            os.remove(path + os.sep + "test.txt")
            
    except Exception as e:
        print(e)
        return False
    return True

def main():
    args = option.parser.parse_args()
    if not checkFoldPermission(args.store_path):
        print('\nPermission denied: %s' % args.store_path)
        print('Please make sure you have the permission to save the data!\n')
    else:
        print('Stockholm is starting...\n')
        stockh = Stockholm(args)
        stockh.run()
        print('Stockholm is done...\n')

if __name__ == '__main__':
    main()

option.py

import argparse 
import datetime

def get_date_str(offset):
    if(offset is None):
        offset = 0
    date_str = (datetime.datetime.today() + datetime.timedelta(days=offset)).strftime("%Y-%m-%d")
    return date_str

_default = dict(
    reload_data = 'Y',
    gen_portfolio = 'N',
    output_type = 'json',
    charset = 'utf-8',
    test_date_range = 60,
    start_date = get_date_str(-90),
    end_date = get_date_str(None),
    target_date = get_date_str(None),
    store_path = 'USER_HOME/tmp/stockholm_export',
    thread = 10,
    testfile_path = './portfolio_test.txt',
    db_name = 'stockholm',
    methods = ''
    )

parser = argparse.ArgumentParser(description='A stock crawler and portfolio testing framework.') 

parser.add_argument('--reload', type=str, default=_default['reload_data'], dest='reload_data', help='Reload the stock data or not (Y/N), Default: %s' % _default['reload_data'])

parser.add_argument('--portfolio', type=str, default=_default['gen_portfolio'], dest='gen_portfolio', help='Generate the portfolio or not (Y/N), Default: %s' % _default['gen_portfolio'])

parser.add_argument('--output', type=str, default=_default['output_type'], dest='output_type', help='Data output type (json/csv/all), Default: %s' % _default['output_type'])

parser.add_argument('--charset', type=str, default=_default['charset'], dest='charset', help='Data output charset (utf-8/gbk), Default: %s' % _default['charset'])

parser.add_argument('--testrange', type=int, default=_default['test_date_range'], dest='test_date_range', help='Test date range(days): %s' % _default['test_date_range'])

parser.add_argument('--startdate', type=str, default=_default['start_date'], dest='start_date', help='Data loading start date, Default: %s' % _default['start_date'])

parser.add_argument('--enddate', type=str, default=_default['end_date'], dest='end_date', help='Data loading end date, Default: %s' % _default['end_date'])

parser.add_argument('--targetdate', type=str, default=_default['target_date'], dest='target_date', help='Portfolio generating target date, Default: %s' % _default['target_date'])

parser.add_argument('--storepath', type=str, default=_default['store_path'], dest='store_path', help='Data file store path, Default: %s' % _default['store_path'])

parser.add_argument('--thread', type=int, default=_default['thread'], dest='thread', help='Thread number, Default: %s' % _default['thread'])

parser.add_argument('--testfile', type=str, default=_default['testfile_path'], dest='testfile_path', help='Portfolio test file path, Default: %s' % _default['testfile_path'])

parser.add_argument('--dbname', type=str, default=_default['db_name'], dest='db_name', help='MongoDB DB name, Default: %s' % _default['db_name'])

parser.add_argument('--methods', type=str, default=_default['methods'], dest='methods', help='Target methods for back testing, Default: %s' % _default['methods'])

def main():
    args = parser.parse_args()
    print(args)

if __name__ == '__main__':
    main()

stockholm.py

#coding:utf-8
import requests
import json
import datetime
import timeit
import time
import io
import os
import csv
import re
from pymongo import MongoClient
from multiprocessing.dummy import Pool as ThreadPool
from functools import partial

class Stockholm(object):

    def __init__(self, args):
        ## flag of if need to reload all stock data
        self.reload_data = args.reload_data
        ## flag of if need to generate portfolio
        self.gen_portfolio = args.gen_portfolio
        ## type of output file json/csv or both
        self.output_type = args.output_type
        ## charset of output file utf-8/gbk
        self.charset = args.charset
        ## portfolio testing date range(# of days)
        self.test_date_range = args.test_date_range
        ## stock data loading start date(e.g. 2014-09-14)
        self.start_date = args.start_date
        ## stock data loading end date
        self.end_date = args.end_date
        ## portfolio generating target date
        self.target_date = args.target_date
        ## thread number
        self.thread = args.thread
        ## data file store path
        if(args.store_path == 'USER_HOME/tmp/stockholm_export'):
            self.export_folder = os.path.expanduser('~') + '/tmp/stockholm_export'
        else:
            self.export_folder = args.store_path
        ## portfolio testing file path
        self.testfile_path = args.testfile_path
        ## methods for back testing
        self.methods = args.methods

        ## for getting quote symbols
        self.all_quotes_url = 'http://money.finance.sina.com.cn/d/api/openapi_proxy.php'
        ## for loading quote data
        self.yql_url = 'http://query.yahooapis.com/v1/public/yql'
        ## export file name
        self.export_file_name = 'stockholm_export'

        self.index_array = ['000001.SS', '399001.SZ', '000300.SS']
        self.sh000001 = {'Symbol': '000001.SS', 'Name': '上证指数'}
        self.sz399001 = {'Symbol': '399001.SZ', 'Name': '深证成指'}
        self.sh000300 = {'Symbol': '000300.SS', 'Name': '沪深300'}
        ## self.sz399005 = {'Symbol': '399005.SZ', 'Name': '中小板指'}
        ## self.sz399006 = {'Symbol': '399006.SZ', 'Name': '创业板指'}

        ## mongodb info
        self.mongo_url = 'localhost'
        self.mongo_port = 27017
        self.database_name = args.db_name
        self.collection_name = 'testing_method'
        
    def get_columns(self, quote):
        columns = []
        if(quote is not None):
            for key in quote.keys():
                if(key == 'Data'):
                    for data_key in quote['Data'][-1]:
                        columns.append("data." + data_key)
                else:
                    columns.append(key)
            columns.sort()
        return columns

    def get_profit_rate(self, price1, price2):
        if(price1 == 0):
            return None
        else:
            return round((price2-price1)/price1, 5)

    def get_MA(self, number_array):
        total = 0
        n = 0
        for num in number_array:
            if num is not None and num != 0:
                n += 1
                total += num
        return round(total/n, 3)

    def convert_value_check(self, exp):
        val = exp.replace('day', 'quote[\'Data\']').replace('(0)', '(-0)')
        val = re.sub(r'\(((-)?\d+)\)', r'[target_idx\g<1>]', val)
        val = re.sub(r'\.\{((-)?\w+)\}', r"['\g<1>']", val)
        return val

    def convert_null_check(self, exp):
        p = re.compile('\((-)?\d+...\w+\}')
        iterator = p.finditer(exp.replace('(0)', '(-0)'))
        array = []
        for match in iterator:
            v = 'quote[\'Data\']' + match.group()
            v = re.sub(r'\(((-)?\d+)\)', r'[target_idx\g<1>]', v)
            v = re.sub(r'\.\{((-)?\w+)\}', r"['\g<1>']", v)
            v += ' is not None'
            array.append(v)
        val = ' and '.join(array)
        return val

    class KDJ():
        def _avg(self, array):
            length = len(array)
            return sum(array)/length
        
        def _getMA(self, values, window):
            array = []
            x = window
            while x <= len(values):
                curmb = 50
                if(x-window == 0):
                    curmb = self._avg(values[x-window:x])
                else:
                    curmb = (array[-1]*2+values[x-1])/3
                array.append(round(curmb,3))
                x += 1
            return array
        
        def _getRSV(self, arrays):
            rsv = []
            x = 9
            while x <= len(arrays):
                high = max(map(lambda x: x['High'], arrays[x-9:x]))
                low = min(map(lambda x: x['Low'], arrays[x-9:x]))
                close = arrays[x-1]['Close']
                rsv.append((close-low)/(high-low)*100)
                t = arrays[x-1]['Date']
                x += 1
            return rsv
        
        def getKDJ(self, quote_data):
            if(len(quote_data) > 12):
                rsv = self._getRSV(quote_data)
                k = self._getMA(rsv,3)
                d = self._getMA(k,3)
                j = list(map(lambda x: round(3*x[0]-2*x[1],3), zip(k[2:], d)))
                
                for idx, data in enumerate(quote_data[0:12]):
                    data['KDJ_K'] = None
                    data['KDJ_D'] = None
                    data['KDJ_J'] = None
                for idx, data in enumerate(quote_data[12:]):
                    data['KDJ_K'] = k[2:][idx]
                    data['KDJ_D'] = d[idx]
                    if(j[idx] > 100):
                        data['KDJ_J'] = 100
                    elif(j[idx] < 0):
                        data['KDJ_J'] = 0
                    else:
                        data['KDJ_J'] = j[idx]
                
            return quote_data

    def load_all_quote_symbol(self):
        print("load_all_quote_symbol start..." + "\n")
        
        start = timeit.default_timer()

        all_quotes = []
        
        all_quotes.append(self.sh000001)
        all_quotes.append(self.sz399001)
        all_quotes.append(self.sh000300)
        ## all_quotes.append(self.sz399005)
        ## all_quotes.append(self.sz399006)
        
        try:
            count = 1
            while (count < 100):
                para_val = '[["hq","hs_a","",0,' + str(count) + ',500]]'
                r_params = {'__s': para_val}
                r = requests.get(self.all_quotes_url, params=r_params)
                if(len(r.json()[0]['items']) == 0):
                    break
                for item in r.json()[0]['items']:
                    quote = {}
                    code = item[0]
                    name = item[2]
                    ## convert quote code
                    if(code.find('sh') > -1):
                        code = code[2:] + '.SS'
                    elif(code.find('sz') > -1):
                        code = code[2:] + '.SZ'
                    ## convert quote code end
                    quote['Symbol'] = code
                    quote['Name'] = name
                    all_quotes.append(quote)
                count += 1
        except Exception as e:
            print("Error: Failed to load all stock symbol..." + "\n")
            print(e)
        
        print("load_all_quote_symbol end... time cost: " + str(round(timeit.default_timer() - start)) + "s" + "\n")
        return all_quotes

    def load_quote_info(self, quote, is_retry):
        print("load_quote_info start..." + "\n")
        
        start = timeit.default_timer()

        if(quote is not None and quote['Symbol'] is not None):
            yquery = 'select * from yahoo.finance.quotes where symbol = "' + quote['Symbol'].lower() + '"'
            r_params = {'q': yquery, 'format': 'json', 'env': 'http://datatables.org/alltables.env'}
            r = requests.get(self.yql_url, params=r_params)
            ## print(r.url)
            ## print(r.text)
            rjson = r.json()
            try:
                quote_info = rjson['query']['results']['quote']
                quote['LastTradeDate'] = quote_info['LastTradeDate']
                quote['LastTradePrice'] = quote_info['LastTradePriceOnly']
                quote['PreviousClose'] = quote_info['PreviousClose']
                quote['Open'] = quote_info['Open']
                quote['DaysLow'] = quote_info['DaysLow']
                quote['DaysHigh'] = quote_info['DaysHigh']
                quote['Change'] = quote_info['Change']
                quote['ChangeinPercent'] = quote_info['ChangeinPercent']
                quote['Volume'] = quote_info['Volume']
                quote['MarketCap'] = quote_info['MarketCapitalization']
                quote['StockExchange'] = quote_info['StockExchange']
                
            except Exception as e:
                print("Error: Failed to load stock info... " + quote['Symbol'] + "/" + quote['Name'] + "\n")
                print(e + "\n")
                if(not is_retry):
                    time.sleep(1)
                    load_quote_info(quote, True) ## retry once for network issue
            
        ## print(quote)
        print("load_quote_info end... time cost: " + str(round(timeit.default_timer() - start)) + "s" + "\n")
        return quote

    def load_all_quote_info(self, all_quotes):
        print("load_all_quote_info start...")
        
        start = timeit.default_timer()
        for idx, quote in enumerate(all_quotes):
            print("#" + str(idx + 1))
            load_quote_info(quote, False)

        print("load_all_quote_info end... time cost: " + str(round(timeit.default_timer() - start)) + "s")
        return all_quotes

    def load_quote_data(self, quote, start_date, end_date, is_retry, counter):
        ## print("load_quote_data start..." + "\n")
        
        start = timeit.default_timer()

        if(quote is not None and quote['Symbol'] is not None):        
            yquery = 'select * from yahoo.finance.historicaldata where symbol = "' + quote['Symbol'].upper() + '" and startDate = "' + start_date + '" and endDate = "' + end_date + '"'
            r_params = {'q': yquery, 'format': 'json', 'env': 'http://datatables.org/alltables.env'}
            try:
                r = requests.get(self.yql_url, params=r_params)
                ## print(r.url)
                ## print(r.text)
                rjson = r.json()
                quote_data = rjson['query']['results']['quote']
                quote_data.reverse()
                quote['Data'] = quote_data
                if(not is_retry):
                    counter.append(1)          
                
            except:
                print("Error: Failed to load stock data... " + quote['Symbol'] + "/" + quote['Name'] + "\n")
                if(not is_retry):
                    time.sleep(2)
                    self.load_quote_data(quote, start_date, end_date, True, counter) ## retry once for network issue
        
            print("load_quote_data " + quote['Symbol'] + "/" + quote['Name'] + " end..." + "\n")
            ## print("time cost: " + str(round(timeit.default_timer() - start)) + "s." + "\n")
            ## print("total count: " + str(len(counter)) + "\n")
        return quote

    def load_all_quote_data(self, all_quotes, start_date, end_date):
        print("load_all_quote_data start..." + "\n")
        
        start = timeit.default_timer()

        counter = []
        mapfunc = partial(self.load_quote_data, start_date=start_date, end_date=end_date, is_retry=False, counter=counter)
        pool = ThreadPool(self.thread)
        pool.map(mapfunc, all_quotes) ## multi-threads executing
        pool.close() 
        pool.join()

        print("load_all_quote_data end... time cost: " + str(round(timeit.default_timer() - start)) + "s" + "\n")
        return all_quotes

    def data_process(self, all_quotes):
        print("data_process start..." + "\n")
        
        kdj = self.KDJ()
        start = timeit.default_timer()
        
        for quote in all_quotes:

            if(quote['Symbol'].startswith('300')):
                quote['Type'] = '创业板'
            elif(quote['Symbol'].startswith('002')):
                quote['Type'] = '中小板'
            else:
                quote['Type'] = '主板'
            
            if('Data' in quote):
                try:
                    temp_data = []
                    for quote_data in quote['Data']:
                        if(quote_data['Volume'] != '000' or quote_data['Symbol'] in self.index_array):
                            d = {}
                            d['Open'] = float(quote_data['Open'])
                            ## d['Adj_Close'] = float(quote_data['Adj_Close'])
                            d['Close'] = float(quote_data['Close'])
                            d['High'] = float(quote_data['High'])
                            d['Low'] = float(quote_data['Low'])
                            d['Volume'] = int(quote_data['Volume'])
                            d['Date'] = quote_data['Date']
                            temp_data.append(d)
                    quote['Data'] = temp_data
                except KeyError as e:
                    print("Data Process: Key Error")
                    print(e)
                    print(quote)

        ## calculate Change / 5 10 20 30 Day MA
        for quote in all_quotes:
            if('Data' in quote):
                try:
                    for i, quote_data in enumerate(quote['Data']):
                        if(i > 0):
                            quote_data['Change'] = self.get_profit_rate(quote['Data'][i-1]['Close'], quote_data['Close'])
                            quote_data['Vol_Change'] = self.get_profit_rate(quote['Data'][i-1]['Volume'], quote_data['Volume'])                        
                        else:
                            quote_data['Change'] = None
                            quote_data['Vol_Change'] = None
                            
                    last_5_array = []
                    last_10_array = []
                    last_20_array = []
                    last_30_array = []
                    for i, quote_data in enumerate(quote['Data']):
                        last_5_array.append(quote_data['Close'])
                        last_10_array.append(quote_data['Close'])
                        last_20_array.append(quote_data['Close'])
                        last_30_array.append(quote_data['Close'])
                        quote_data['MA_5'] = None
                        quote_data['MA_10'] = None
                        quote_data['MA_20'] = None
                        quote_data['MA_30'] = None
                        
                        if(i < 4):
                            continue
                        if(len(last_5_array) == 5):
                            last_5_array.pop(0)
                        quote_data['MA_5'] = self.get_MA(last_5_array)
                        
                        if(i < 9):
                            continue
                        if(len(last_10_array) == 10):
                            last_10_array.pop(0)
                        quote_data['MA_10'] = self.get_MA(last_10_array)
                        
                        if(i < 19):
                            continue
                        if(len(last_20_array) == 20):
                            last_20_array.pop(0)
                        quote_data['MA_20'] = self.get_MA(last_20_array)
                        
                        if(i < 29):
                            continue
                        if(len(last_30_array) == 30):
                            last_30_array.pop(0)
                        quote_data['MA_30'] = self.get_MA(last_30_array)
                        
                        
                except KeyError as e:
                    print("Key Error")
                    print(e)
                    print(quote)

        ## calculate KDJ
        for quote in all_quotes:
            if('Data' in quote):
                try:
                    kdj.getKDJ(quote['Data'])
                except KeyError as e:
                    print("Key Error")
                    print(e)
                    print(quote)

        print("data_process end... time cost: " + str(round(timeit.default_timer() - start)) + "s" + "\n")

    def data_export(self, all_quotes, export_type_array, file_name):
        
        start = timeit.default_timer()
        directory = self.export_folder
        if(file_name is None):
            file_name = self.export_file_name
        if not os.path.exists(directory):
            os.makedirs(directory)

        if(all_quotes is None or len(all_quotes) == 0):
            print("no data to export...\n")
        
        if('json' in export_type_array):
            print("start export to JSON file...\n")
            f = io.open(directory + '/' + file_name + '.json', 'w', encoding=self.charset)
            json.dump(all_quotes, f, ensure_ascii=False)
            
        if('csv' in export_type_array):
            print("start export to CSV file...\n")
            columns = []
            if(all_quotes is not None and len(all_quotes) > 0):
                columns = self.get_columns(all_quotes[0])
            writer = csv.writer(open(directory + '/' + file_name + '.csv', 'w', encoding=self.charset))
            writer.writerow(columns)

            for quote in all_quotes:
                if('Data' in quote):
                    for quote_data in quote['Data']:
                        try:
                            line = []
                            for column in columns:
                                if(column.find('data.') > -1):
                                    if(column[5:] in quote_data):
                                        line.append(quote_data[column[5:]])
                                else:
                                    line.append(quote[column])
                            writer.writerow(line)
                        except Exception as e:
                            print(e)
                            print("write csv error: " + quote)
            
        if('mongo' in export_type_array):
            print("start export to MongoDB...\n")
            
        print("export is complete... time cost: " + str(round(timeit.default_timer() - start)) + "s" + "\n")

    def file_data_load(self):
        print("file_data_load start..." + "\n")
        
        start = timeit.default_timer()
        directory = self.export_folder
        file_name = self.export_file_name
        
        all_quotes_data = []
        f = io.open(directory + '/' + file_name + '.json', 'r', encoding='utf-8')
        json_str = f.readline()
        all_quotes_data = json.loads(json_str)
        
        print("file_data_load end... time cost: " + str(round(timeit.default_timer() - start)) + "s" + "\n")
        return all_quotes_data

    def check_date(self, all_quotes, date):
        
        is_date_valid = False
        for quote in all_quotes:
            if(quote['Symbol'] in self.index_array):
                for quote_data in quote['Data']:
                    if(quote_data['Date'] == date):
                        is_date_valid = True
        if not is_date_valid:
            print(date + " is not valid...\n")
        return is_date_valid

    def quote_pick(self, all_quotes, target_date, methods):
        print("quote_pick start..." + "\n")
        
        start = timeit.default_timer()

        results = []
        data_issue_count = 0
        
        for quote in all_quotes:
            try:
                if(quote['Symbol'] in self.index_array):
                    results.append(quote)
                    continue
                
                target_idx = None
                for idx, quote_data in enumerate(quote['Data']):
                    if(quote_data['Date'] == target_date):
                        target_idx = idx
                if(target_idx is None):
                    ## print(quote['Name'] + " data is not available at this date..." + "\n")
                    data_issue_count+=1
                    continue
                
                ## pick logic ##
                valid = False
                for method in methods:
                    ## print(method['name'])
                    ## null_check = eval(method['null_check'])
                    try:
                        value_check = eval(method['value_check'])
                        if(value_check):
                            quote['Method'] = method['name']
                            results.append(quote)
                            valid = True
                            break
                    except:
                        valid = False
                if(valid):
                    continue
                                    
                ## pick logic end ##
                
            except KeyError as e:
                ## print("KeyError: " + quote['Name'] + " data is not available..." + "\n")
                data_issue_count+=1
                
        print("quote_pick end... time cost: " + str(round(timeit.default_timer() - start)) + "s" + "\n")
        print(str(data_issue_count) + " quotes of data is not available...\n")
        return results

    def profit_test(self, selected_quotes, target_date):
        print("profit_test start..." + "\n")
        
        start = timeit.default_timer()
        
        results = []
        INDEX = None
        INDEX_idx = 0

        for quote in selected_quotes:
            if(quote['Symbol'] == self.sh000300['Symbol']):
                INDEX = quote
                for idx, quote_data in enumerate(quote['Data']):
                    if(quote_data['Date'] == target_date):
                        INDEX_idx = idx
                break
        
        for quote in selected_quotes:
            target_idx = None
            
            if(quote['Symbol'] in self.index_array):
                continue
            
            for idx, quote_data in enumerate(quote['Data']):
                if(quote_data['Date'] == target_date):
                    target_idx = idx
            if(target_idx is None):
                print(quote['Name'] + " data is not available for testing..." + "\n")
                continue
            
            test = {}
            test['Name'] = quote['Name']
            test['Symbol'] = quote['Symbol']
            test['Method'] = quote['Method']
            test['Type'] = quote['Type']
            if('KDJ_K' in quote['Data'][target_idx]):
                test['KDJ_K'] = quote['Data'][target_idx]['KDJ_K']
                test['KDJ_D'] = quote['Data'][target_idx]['KDJ_D']
                test['KDJ_J'] = quote['Data'][target_idx]['KDJ_J']
            test['Close'] = quote['Data'][target_idx]['Close']
            test['Change'] = quote['Data'][target_idx]['Change']
            test['Vol_Change'] = quote['Data'][target_idx]['Vol_Change']
            test['MA_5'] = quote['Data'][target_idx]['MA_5']
            test['MA_10'] = quote['Data'][target_idx]['MA_10']
            test['MA_20'] = quote['Data'][target_idx]['MA_20']
            test['MA_30'] = quote['Data'][target_idx]['MA_30']
            test['Data'] = [{}]

            for i in range(1,11):
                if(target_idx+i >= len(quote['Data'])):
                    print(quote['Name'] + " data is not available for " + str(i) + " day testing..." + "\n")
                    break

                day2day_profit = self.get_profit_rate(quote['Data'][target_idx]['Close'], quote['Data'][target_idx+i]['Close'])
                test['Data'][0]['Day_' + str(i) + '_Profit'] = day2day_profit
                if(INDEX_idx+i < len(INDEX['Data'])):
                    day2day_INDEX_change = self.get_profit_rate(INDEX['Data'][INDEX_idx]['Close'], INDEX['Data'][INDEX_idx+i]['Close'])
                    test['Data'][0]['Day_' + str(i) + '_INDEX_Change'] = day2day_INDEX_change
                    test['Data'][0]['Day_' + str(i) + '_Differ'] = day2day_profit-day2day_INDEX_change
            
            results.append(test)
            
        print("profit_test end... time cost: " + str(round(timeit.default_timer() - start)) + "s" + "\n")
        return results

    def data_load(self, start_date, end_date, output_types):
        all_quotes = self.load_all_quote_symbol()
        print("total " + str(len(all_quotes)) + " quotes are loaded..." + "\n")
        all_quotes = all_quotes
        ## self.load_all_quote_info(all_quotes)
        self.load_all_quote_data(all_quotes, start_date, end_date)
        self.data_process(all_quotes)
        
        self.data_export(all_quotes, output_types, None)

    def data_test(self, target_date, test_range, output_types):
        ## loading test methods
        methods = []
        path = self.testfile_path
        
        ## from mongodb
        if(path == 'mongodb'):
            print("Load testing methods from Mongodb...\n")
            client = MongoClient(self.mongo_url, self.mongo_port)
            db = client[self.database_name]
            col = db[self.collection_name]
            q = None
            if(len(self.methods) > 0):
                applied_methods = list(map(int, self.methods.split(',')))
                q = {"method_id": {"$in": applied_methods}}
            for doc in col.find(q, ['name','desc','method']):
                print(doc)
                m = {'name': doc['name'], 'value_check': self.convert_value_check(doc['method'])}
                methods.append(m)
                
        ## from test file
        else:
            if not os.path.exists(path):
                print("Portfolio test file is not existed, testing is aborted...\n")
                return
            f = io.open(path, 'r', encoding='utf-8')
            for line in f:
                if(line.startswith('##') or len(line.strip()) == 0):
                    continue
                line = line.strip().strip('\n')
                name = line[line.find('[')+1:line.find(']:')]
                value = line[line.find(']:')+2:]
                m = {'name': name, 'value_check': self.convert_value_check(value)}
                methods.append(m)
                
        if(len(methods) == 0):
            print("No method is loaded, testing is aborted...\n")
            return

        ## portfolio testing 
        all_quotes = self.file_data_load()
        target_date_time = datetime.datetime.strptime(target_date, "%Y-%m-%d")
        for i in range(test_range):
            date = (target_date_time - datetime.timedelta(days=i)).strftime("%Y-%m-%d")
            is_date_valid = self.check_date(all_quotes, date)
            if is_date_valid:
                selected_quotes = self.quote_pick(all_quotes, date, methods)
                res = self.profit_test(selected_quotes, date)
                self.data_export(res, output_types, 'result_' + date)

    def run(self):
        ## output types
        output_types = []
        if(self.output_type == "json"):
            output_types.append("json")
        elif(self.output_type == "csv"):
            output_types.append("csv")
        elif(self.output_type == "all"):
            output_types = ["json", "csv"]
            
        ## loading stock data
        if(self.reload_data == 'Y'):
            print("Start loading stock data...\n")
            self.data_load(self.start_date, self.end_date, output_types)
            
        ## test & generate portfolio
        if(self.gen_portfolio == 'Y'):
            print("Start portfolio testing...\n")
            self.data_test(self.target_date, self.test_date_range, output_types)

mongo_scripts.txt

use stockholm

db.counters.insert(
   {
      _id: "method_id",
      seq: 0
   }
)

function getNextSequence(name) {
   var ret = db.counters.findAndModify(
          {
            query: { _id: name },
            update: { $inc: { seq: 1 } },
            new: true
          }
   );

   return ret.seq;
}

db.testing_method.insert({"method_id": getNextSequence("method_id"), "name":"测试方法1", "desc":"这是一个测试方法。", "user_name":"Stockholm", "user_id":"dtnium@gmail.com", "creation_date": new Date(), "modification_date": new Date(), "method":"day(-2).{KDJ_J}<20 and day(-1).{KDJ_J}<20 and day(0).{KDJ_J}-day(-1).{KDJ_J}>=40 and day(0).{Vol_Change}>=1 and day(0).{MA_10}*1.05>day(0).{Close}"})

db.testing_method.insert({"method_id": getNextSequence("method_id"), "name":"测试方法2", "desc":"这是一个测试方法。", "user_name":"Stockholm", "user_id":"dtnium@gmail.com", "creation_date": new Date(), "modification_date": new Date(), "method":"day(-2).{KDJ_J}-day(-1).{KDJ_J}>20 and day(0).{KDJ_J}-day(-1).{KDJ_J}>20 and day(-1).{KDJ_J}<50 and day(0).{Vol_Change}<=1"})

portfolio_test.txt

## Portfolio selection methodology sample file

[测试方法1]:day(-2).{KDJ_J}<20 and day(-1).{KDJ_J}<20 and day(0).{KDJ_J}-day(-1).{KDJ_J}>=40 and day(0).{Vol_Change}>=1 and day(0).{MA_10}*1.05>day(0).{Close}

[测试方法2]:day(-2).{KDJ_J}-day(-1).{KDJ_J}>20 and day(0).{KDJ_J}-day(-1).{KDJ_J}>20 and day(-1).{KDJ_J}<50 and day(0).{Vol_Change}<=1

##[测试方法3]:50<day(-1).{KDJ_J}<80 and day(-2).{KDJ_J}<day(-1).{KDJ_J} and day(0).{KDJ_J}<day(-1).{KDJ_J}

运行时参数

--storepath c://test --output csv --charset utf-8

运行示例

能干什么

如果你想基于沪深股市行情数据进行一些工作，它可以帮助你导出指定时间范围内所有沪深A股的行情数据和一些技术指标，包括代码、名称、开盘、收盘、最高、最低、成交量、均线、KDJ等。

如果你对于技术分析有兴趣，它可以帮你根据你自定义的规则在所有沪深A股的范围内进行选股，并回测选股策略的收益情况（详细说明后面会有）。这样就能够非常方便快捷地测试和调整基于技术分析的选股策略。

还有些什么问题

行情数据目前来源于雅虎YQL，每日数据的更新时间不太稳定（一般在中国时间午夜左右）。

目前支持的技术指标还不多，还有一些指标如MACD和BOLL后续会增加。

在回测中，如果有在选定时间内发生过除权的股票，收益计算会有问题。

导出格式目前只支持CSV和JSON文本。MongoDB和MySQL会考虑后续加入。

环境

Python 3.4以上

Requests

PyMongo

OSX和CentOS已测。Windows尚未测试，输出路径可能有问题。

pip install requests
pip install pymongo

使用

python main.py [-h] [--reload {Y,N}] [--portfolio {Y,N}] 
               [--output {json,csv,all}] [--storepath PATH] [--thread NUM] 
               [--startdate yyyy-MM-dd] [--enddate yyyy-MM-dd] 
               [--targetdate yyyy-MM-dd] [--testrange NUM] [--testfile PATH]

可选参数

-h, --help                  查看帮助并退出
--reload {Y,N}              是否重新抓取股票数据，默认值：Y
--portfolio {Y,N}           是否生成选股测试结果，默认值：N
--output {json,csv,all}     输出文件格式，默认值：json
--charset {utf-8,gbk}       输出文件编码，默认值：utf-8
--storepath PATH            输出文件路径，默认值：~/tmp/stockholm_export
--thread NUM                线程数，默认值：10
--startdate yyyy-MM-dd      抓取数据的开始日期，默认值：当前系统日期-100天（例如2015-01-01）
--enddate yyyy-MM-dd        抓取数据的结束日期，默认值：当前系统日期
--targetdate yyyy-MM-dd     测试选股策略的目标日期，默认值：当前系统日期
--testrange NUM             测试日期范围天数，默认值：50
--testfile PATH             测试文件路径，默认值：./portfolio_test.txt

可用数据/格式

行情数据:

[
	{"Symbol": "600000.SS", 
	 "Name": "浦发银行"，
	 "Data": [
				 {"Vol_Change": null, "MA_10": null, "Date": "2015-03-26", "High": 15.58, "Open": 15.15, "Volume": 282340700, "Close": 15.36, "Change": null, "Low": 15.04}, 
				 {"Vol_Change": -0.22726, "MA_10": null, "Date": "2015-03-27", "High": 15.55, "Open": 15.32, "Volume": 218174900, "Close": 15.36, "Change": 0.0, "Low": 15.17}
			 ]
	}
]

Date(日期); Open(开盘价); Close(收盘价); High(当日最高); Low(当日最低); Change(价格变化%); Volume(成交量); Vol_Change(成交量较前日变化); MA_5(5日均线); MA_10(10日均线); MA_20(20日均线); MA_30(30日均线); KDJ_K(KDJ指标K); KDJ_D(KDJ指标D); KDJ_J(KDJ指标J);

以上数据都可以用于制定选股策略，后面会介绍具体方法。

选股策略测试数据:

[
	{
		"Symbol": "600000.SS", 
		"Name": "浦发银行", 
		"Close": 14.51, 
		"Change": 0.06456,
		"Vol_Change": 2.39592, 
		"MA_10": 14.171, 
		"KDJ_K": 37.65, 
		"KDJ_D": 33.427, 
		"KDJ_J": 46.096, 
		"Data": [
					{"Day_5_Differ": 0.01869, "Day_9_Profit": 0.08546, "Day_1_Profit": -0.02826, "Day_1_INDEX_Change": -0.00484, "Day_3_INDEX_Change": 0.01557, "Day_5_INDEX_Change": 0.04747, "Day_3_Differ": 0.02647, "Day_9_INDEX_Change": 0.1003, "Day_5_Profit": 0.06616, "Day_3_Profit": 0.04204, "Day_1_Differ": -0.02342, "Day_9_Differ": -0.014840000000000006}
				]
	}
]

Close(收盘价); Change(价格变化%); Vol_Change(成交量较前日变化); MA_10(十天均价); KDJ_K(KDJ指标K); KDJ_D(KDJ指标D); KDJ_J(KDJ指标J); Day_1_Profit(后一天利润率%); Day_1_INDEX_Change(后一天沪深300变化率%); Day_1_Differ(后一天相对利润率%——即利润率-沪深300变化率); Day_n_Profit(后n天利润率%); Day_n_INDEX_Change(后n天沪深300变化率%); Day_n_Differ(后n天相对利润率%——即利润率-沪深300变化率);

行情数据抓取范例

获取从当前日期倒推100天(不是100个交易日)的所有沪深股票行情数据。

执行完成后，数据在当前用户文件夹下./tmp/stockholm_export/stockholm_export.json

python main.py

如果想导出csv文件

python main.py --output=csv

选股策略测试范例

选股策略范例文件内容如下(包括在源码中)

选股策略”method 1”是:前前个交易日的KDJ指标的J值小于20+前个交易日的KDJ指标J值小于20+当前交易日的KDJ指标J值比上个交易日大40+当前交易日成交量变化大于100%

## Portfolio selection methodology sample file

[method 1]:day(-2).{KDJ_J}<20 and day(-1).{KDJ_J}<20 and day(0).{KDJ_J}-day(-1).{KDJ_J}>=40 and day(0).{Vol_Change}>=1

以当前系统日期为目标日期进行倒推60天得选股策略测试。

不重新抓取行情数据并执行测试命令。

执行完毕后，会将测试结果按照每天一个文件的方式保存在./tmp/stockholm_export/。

文件名格式为result_yyyy-MM-dd.json(例如result_2015-03-24.json)。

python main.py --reload=N --portfolio=Y

通过更改测试文件中的选股策略公式，可以随意测试指定时间范围内的选股效果。

XingLiu's Blog

一个用Python编写的股票数据（沪深）爬虫和选股策略测试框架

代码