Read this in other languages: English.
"要在市场上生存，就必须远离聪明，因为，你的聪明在市场面前一钱不值"------缠中说禅
fooltrader是一个利用大数据技术设计的量化分析交易系统,包括数据的抓取,清洗,结构化,计算,展示,回测和交易.
它的目标是提供一个统一的框架来对全市场(股票,期货,债券,外汇,数字货币,宏观经济等)进行研究,回测,预测,交易.
它的适用对象包括:量化交易员,财经类专业师生,对经济数据感兴趣的人,程序员,喜欢自由而有探索精神的人
输入你感兴趣的个股,查看其净利润跟股价的关系.
聚合展示你关心的视图
api输出结果具体字段含义请参考数据协议.
In [1]: from fooltrader.api import finance
In [2]: finance.get_income_statement_items('300027',report_period='2017-06-30')
#试一试
#finance.get_balance_sheet_items('300027',,report_event_date='2017-01-01')
#finance.get_cash_flow_statement_items('300027')
Out[2]:
{'EPS': 0.15,
'ManagingCosts': 257005115.85,
'accumulatedOtherComprehensiveIncome': 471486112.3,
'assetsDevaluation': -21647912.31,
'attributableToMinorityShareholders': 90255906.93,
'attributableToOwnersOfParentCompany': 381230205.37,
'businessTaxesAndSurcharges': 80033207.21,
'code': '300027',
'dilutedEPS': 0.15,
'disposalLossOnNonCurrentLiability': 281050.25,
'exchangeGains': 0.0,
'financingExpenses': 132202866.43,
'id': 'stock_sz_300027_20170630',
'incomeFromChangesInFairValue': 0.0,
'incomeTaxExpense': 111864455.56,
'investmentIncome': 541478955.17,
'investmentIncomeFromRelatedEnterpriseAndJointlyOperating': '45035770.67',
'minorityInterestIncome': 91203287.92,
'netProfit': 521516997.38,
'netProfitAttributedToParentCompanyOwner': 430313709.46,
'nonOperatingExpenditure': 13775609.35,
'nonOperatingIncome': 27864700.17,
'operatingCosts': 679308123.4,
'operatingProfit': 619292362.12,
'operatingRevenue': 1465863805.45,
'operatingTotalCosts': 1388050398.5,
'otherComprehensiveIncome': -50030885.08,
'reportDate': '2017-06-30',
'reportEventDate': '2017-08-29',
'securityId': 'stock_sz_300027',
'sellingExpenses': 261148997.92,
'totalProfits': 633381452.94}
# 营业利润=营业收入-营业成本-营业税金及附加-销售费用-管理费用-财务费用-资产减值损失+公允价值变动收益（损失的话用减）+投资收益
def check_operating_profit(security_item):
income_statement_list = get_income_statement_items(security_item=security_item)
for income_statement in income_statement_list:
operatingProfit = income_statement["operatingRevenue"] \
- income_statement["operatingCosts"] \
- income_statement["businessTaxesAndSurcharges"] \
- income_statement["sellingExpenses"] \
- income_statement["ManagingCosts"] \
- income_statement["financingExpenses"] \
- income_statement["assetsDevaluation"] \
+ income_statement["incomeFromChangesInFairValue"] \
+ income_statement["investmentIncome"]
diff = operatingProfit - income_statement["operatingProfit"]
if abs(diff) >= 1:
print("{} operating profit calculating not pass,calculating result:{},report result:{}".format(
income_statement['id'], operatingProfit, income_statement["operatingProfit"]))
else:
print("{} operating profit calculating pass".format(income_statement['id']))可以用该工具迅速检查财务报表的质量,同时也可以让你对财务报表有更深入的认识.更多例子
In [3]: from fooltrader.datamanager import finance_check
In [4]: finance_check.check_operating_profit('300027')
stock_sz_300027_20061231 operating profit calculating pass
...
stock_sz_300027_20170630 operating profit calculating pass
stock_sz_300027_20170930 operating profit calculating pass我的博客介绍fooltrader投资之财务指标
K线数据
In [5]: from fooltrader.api import quote
In [6]: quote.get_kdata('300027',start_date='20170630',end_date='20170715')
#试一试
#quote.get_kdata('300027',start_date='20170630',end_date='20170715',fuquan='qfq')
#quote.get_kdata('300027',start_date='20170630',end_date='20170715',fuquan='hfq')
Out[6]:
timestamp code name low open close high volume turnover securityId preClose change changePct turnoverRate tCap mCap factor
timestamp
2017-06-30 2017-06-30 300027 华谊兄弟 8.03 8.11 8.09 8.11 9515735 7.684533e+07 stock_sz_300027 8.11 -0.02 -0.2466 0.3832 2.244575e+10 2.008766e+10 15.055
2017-07-03 2017-07-03 300027 华谊兄弟 8.07 8.09 8.20 8.22 15577742 1.270549e+08 stock_sz_300027 8.09 0.11 1.3597 0.6274 2.275095e+10 2.036079e+10 15.055
2017-07-04 2017-07-04 300027 华谊兄弟 8.12 8.20 8.15 8.22 8672705 7.068477e+07 stock_sz_300027 8.20 -0.05 -0.6098 0.3493 2.261222e+10 2.023664e+10 15.055
2017-07-05 2017-07-05 300027 华谊兄弟 8.11 8.15 8.19 8.22 12458305 1.017991e+08 stock_sz_300027 8.15 0.04 0.4908 0.5017 2.272320e+10 2.033596e+10 15.055
2017-07-06 2017-07-06 300027 华谊兄弟 8.17 8.20 8.20 8.29 18642574 1.533405e+08 stock_sz_300027 8.19 0.01 0.1221 0.7508 2.275095e+10 2.036079e+10 15.055
2017-07-07 2017-07-07 300027 华谊兄弟 8.13 8.18 8.17 8.19 9414275 7.679682e+07 stock_sz_300027 8.20 -0.03 -0.3659 0.3791 2.266771e+10 2.028630e+10 15.055
2017-07-10 2017-07-10 300027 华谊兄弟 8.08 8.18 8.09 8.19 12679949 1.029643e+08 stock_sz_300027 8.17 -0.08 -0.9792 0.5107 2.244575e+10 2.008766e+10 15.055
2017-07-11 2017-07-11 300027 华谊兄弟 8.06 8.10 8.08 8.13 11412820 9.235337e+07 stock_sz_300027 8.09 -0.01 -0.1236 0.4596 2.241801e+10 2.006283e+10 15.055
2017-07-12 2017-07-12 300027 华谊兄弟 7.93 8.07 8.03 8.10 13776145 1.104951e+08 stock_sz_300027 8.08 -0.05 -0.6188 0.5548 2.227928e+10 1.993868e+10 15.055
2017-07-13 2017-07-13 300027 华谊兄弟 8.09 8.10 8.28 8.37 43554957 3.590739e+08 stock_sz_300027 8.03 0.25 3.1133 1.7541 2.297291e+10 2.055944e+10 15.055
2017-07-14 2017-07-14 300027 华谊兄弟 8.23 8.25 8.41 8.55 37967053 3.193673e+08 stock_sz_300027 8.28 0.13 1.5700 1.5291 2.333359e+10 2.088223e+10 15.055tick数据
In [7]: for tick in quote.get_ticks('300027',the_date='2017-07-03'):
...: print(tick)
...:
...:
#试一试
#quote.get_ticks('300027',start='2017-06-30',end='2017-07-10')
timestamp price volume turnover direction code securityId
timestamp
2017-07-03 09:25:03 2017-07-03 09:25:03 8.09 41 33169 1 300027 stock_sz_300027
2017-07-03 09:30:03 2017-07-03 09:30:03 8.09 10 8090 1 300027 stock_sz_300027
... ... ... ... ... ... ... ...
2017-07-03 14:57:00 2017-07-03 14:57:00 8.19 4 3276 -1 300027 stock_sz_300027
2017-07-03 14:57:03 2017-07-03 14:57:03 8.19 0 0 -1 300027 stock_sz_300027
2017-07-03 15:00:03 2017-07-03 15:00:03 8.20 4394 3603079 1 300027 stock_sz_300027
[2313 rows x 7 columns]
In [8]: from fooltrader.api import event
In [9]: for item in event.get_forecast_items('000338'):
...: print(item)
...:
{'changeStart': None, 'reportDate': '2008-01-28', 'id': 'stock_sz_000338_2008-01-28', 'preEPS': None, 'securityId': 'stock_sz_000338', 'reportPeriod': '2007-12-31', 'description': '潍柴动力预计2007年1-12月净利润较2006年度备考合并净利润增长约140%左右。', 'change': 1.4, 'type': '预增'}
{'changeStart': None, 'reportDate': '2008-07-24', 'id': 'stock_sz_000338_2008-07-24', 'preEPS': 1.87, 'securityId': 'stock_sz_000338', 'reportPeriod': '2008-06-30', 'description': '预计本公司2008年1-6月归属于母公司所有者净利润与2007年同期调整前及调整后的归属于母公司所有者净利润相比增长50%-100%之间。', 'change': 1.0, 'type': '预增'}
{'changeStart': None, 'reportDate': '2009-08-19', 'id': 'stock_sz_000338_2009-08-19', 'preEPS': 3.19, 'securityId': 'stock_sz_000338', 'reportPeriod': '2009-06-30', 'description': '预计本公司2009年半年度营业收入约为人民币158亿元，营业利润约在人民币15-18亿元之间，利润总额约在人民币15-18亿元之间，归属于上市公司股东的净利润约在人民币10.0-12.5亿元之间。', 'change': 0.0, 'type': '预降'}
...
{'changeStart': 1.4, 'reportDate': '2017-04-14', 'id': 'stock_sz_000338_2017-04-14', 'preEPS': 0.11, 'securityId': 'stock_sz_000338', 'reportPeriod': '2017-03-31', 'description': '预计2017年1-3月归属于上市公司股东的净利润为：109,600.00万元至123,300.00万元，较上年同期相比变动幅度：140.00%至170.00%。', 'change': 1.7, 'type': '预增'}
{'changeStart': 0.7, 'reportDate': '2017-04-28', 'id': 'stock_sz_000338_2017-04-28', 'preEPS': 0.26, 'securityId': 'stock_sz_000338', 'reportPeriod': '2017-06-30', 'description': '预计2017年1-6月归属于上市公司股东的净利润为：183,000.00万元至215,000.00万元，较上年同期相比变动幅度：70.00%至100.00%。', 'change': 1.0, 'type': '预增'}
{'changeStart': 1.5, 'reportDate': '2017-08-31', 'id': 'stock_sz_000338_2017-08-31', 'preEPS': 0.38, 'securityId': 'stock_sz_000338', 'reportPeriod': '2017-09-30', 'description': '预计2017年1-9月归属于上市公司股东的净利润为：385,000.00万元至431,400.00万元，较上年同期相比变动幅度：150.00%至180.00%。', 'change': 1.8, 'type': '预增'}In [10]: from fooltrader.api import technical
In [11]: technical.macd('000778',start_date='20170101',end_date='20170301')
Out[11]:
close close_ema12 close_ema26 diff dea macd
timestamp
2017-01-03 5.21 NaN NaN NaN NaN NaN
2017-01-04 5.24 NaN NaN NaN NaN NaN
2017-01-05 5.31 NaN NaN NaN NaN NaN
2017-01-06 5.28 NaN NaN NaN NaN NaN
2017-01-09 5.33 NaN NaN NaN NaN NaN
2017-01-10 5.30 NaN NaN NaN NaN NaN
2017-01-11 5.34 NaN NaN NaN NaN NaN
2017-01-12 5.21 NaN NaN NaN NaN NaN
2017-01-13 5.11 NaN NaN NaN NaN NaN
2017-01-16 4.95 NaN NaN NaN NaN NaN
2017-01-17 5.00 NaN NaN NaN NaN NaN
2017-01-18 5.05 5.146697 NaN NaN NaN NaN
2017-01-19 4.96 5.117975 NaN NaN NaN NaN
2017-01-20 5.00 5.099825 NaN NaN NaN NaN
2017-01-23 5.05 5.092159 NaN NaN NaN NaN
2017-01-24 5.06 5.087212 NaN NaN NaN NaN
2017-01-25 5.06 5.083025 NaN NaN NaN NaN
2017-01-26 5.07 5.081022 NaN NaN NaN NaN
2017-02-03 5.03 5.073172 NaN NaN NaN NaN
2017-02-06 5.03 5.066530 NaN NaN NaN NaN
2017-02-07 5.01 5.057833 NaN NaN NaN NaN
2017-02-08 5.05 5.056628 NaN NaN NaN NaN
2017-02-09 5.12 5.066378 NaN NaN NaN NaN
2017-02-10 5.27 5.097704 NaN NaN NaN NaN
2017-02-13 5.31 5.130365 NaN NaN NaN NaN
2017-02-14 5.84 5.239540 5.184121 0.055419 0.055419 0.000000
2017-02-15 6.09 5.370380 5.251223 0.119157 0.068166 0.101981
2017-02-16 5.98 5.464167 5.305206 0.158961 0.086325 0.145271
2017-02-17 5.70 5.500449 5.334450 0.165999 0.102260 0.127478
2017-02-20 5.78 5.543457 5.367454 0.176003 0.117009 0.117989
2017-02-21 5.81 5.584464 5.400235 0.184229 0.130453 0.107552
2017-02-22 5.95 5.640700 5.440959 0.199742 0.144310 0.110862
2017-02-23 5.81 5.666746 5.468295 0.198451 0.155139 0.086625
2017-02-24 5.69 5.670324 5.484718 0.185606 0.161232 0.048748
2017-02-27 5.59 5.657966 5.492516 0.165450 0.162076 0.006749
2017-02-28 5.66 5.658279 5.504922 0.153357 0.160332 -0.013950
2017-03-01 5.63 5.653928 5.514187 0.139741 0.156214 -0.032945
我们不需要那么多技术指标,但一定要知道所使用指标的内涵,所以,我们选择自己计算;没错,由于数据的统一性,理所当然地,计算的统一性也有了. 不管是A股,港股,还是数字货币,不管是1分钟级别,还是日线,使用的都是统一的api.
更多用法请查看api文档.
策略的编写,可以采用事件驱动或者时间漫步的方式,查看设计文档
注意:回测框架目前还处于非常初期的阶段
class EventTrader(Trader):
def on_init(self):
self.trader_id = 'aa'
self.only_event_mode = True
self.universe = ['stock_sz_000338']
self.df_map = {}
def on_day_bar(self, bar_item):
current_security = bar_item['securityId']
current_df = self.df_map.get(current_security, pd.DataFrame())
if current_df.empty:
self.df_map[current_security] = current_df
current_df = current_df.append(bar_item, ignore_index=True)
self.df_map[current_security] = current_df
if len(current_df.index) == 10:
ma5 = np.mean(current_df.loc[5:, 'close'])
ma10 = np.mean(current_df.loc[:, 'close'])
# 5日线在10日线上,并且没有持仓,就买入
if ma5 > ma10 and not self.account_service.get_position(current_security):
self.buy(security_id=current_security, current_price=bar_item['close'])
# 5日线在10日线下,并且有持仓,就卖出
elif ma5 < ma10 and self.account_service.get_position(current_security):
self.sell(security_id=current_security, current_price=bar_item['close'])
current_df = current_df.loc[1:, ]
self.df_map[current_security] = current_df运行策略可以实时查看效果,并做进一步的评估
fooltrader是一个层次清晰的系统,你可以在不同的层次对其进行使用,也可以扩展,改造或替换里面的模块.
使用的层次跟架构图里面的模块是一一对应的, 你可以在任何step停下来,进行扩展或者对接你自己熟悉的系统.
当然,还是希望你全部跑通,因为这里的每个模块的技术选型都是经过精心考虑的,并且后续会不停完善.
操作系统:Ubuntu 16.04.3 LTS
原则上,其他也可以,系统使用的组件都是跨平台的,但我只在ubuntu和mac运行过
内存:>=8G
硬盘:越大越好
如果你想直接使用,只需要:
#在python3.5环境下,或者使用virtualenv
pip install fooltrader下载打包好的历史数据data.zip.
看一下数据协议,设置好FOOLTRADER_STORE_PATH,解压下载的文件到该目录.
然后使用定时脚本1和定时脚本2每天抓取增量数据.
该项目的目的之一是方便大家共享数据,不需要每个人都去抓历史数据而导致被屏蔽.
也可以用该脚本对数据进行打包共享
不过由于该项目还在快速更新中,最好还是按下面的步骤直接用源码来搞
clone或者fork代码
$ git clone https://github.com/foolcage/fooltrader.git$ cd fooltrader
$ ./init_env.sh如果你最后看到:
Requirements installed.
env ok那么恭喜你,你可以以各种姿势去玩耍了.
$ source ve/bin/activate
$ ./ve/bin/ipython
In [1]: from fooltrader.datamanager import datamanager
#抓取股票元数据
In [2]: datamanager.crawl_stock_meta()
#抓取指数数据
In [3]: datamanager.crawl_index_quote()
#抓取个股K线和tick数据
In [4]: datamanager.crawl_stock_quote(start_code=002797,end_code=002798,crawl_tick=False)
#抓取财务数据
In [5]: datamanager.crawl_finance_data(start_code=002797,end_code=002798)这里把抓取数据作为一个单独的模块,而不是像某些开源项目那样api和爬虫耦合在一起,主要是为了:
爬虫只干爬虫的事:专注抓取的速度,更好的数据分类,数据补全,防屏蔽等
api设计只依赖数据协议,从而具有更好的速度和灵活性
抓取每天的增量数据只需要:
$ ./sched_finance.sh$ ./sched_quote.sh该脚本会定时去抓取"缺少"的数据,在历史数据完整性检查通过后,其实就是只是抓取当天的数据,这样我们就有了一个自动化自我维护的完整数据源.
可在sched_quote.py文件中进行对定时任务进行配置:
#每天17:00运行
@sched.scheduled_job('cron', hour=17, minute=00)
def scheduled_job1():
crawl_stock_quote('000001', '002999')
crawl_index_quote()
@sched.scheduled_job('cron', hour=17, minute=20)
def scheduled_job2():
crawl_stock_quote('300000', '300999')
@sched.scheduled_job('cron', hour=17, minute=40)
def scheduled_job3():
crawl_stock_quote('600000', '666666')最后强调一下,数据抓下来了,怎么使用?请参考数据协议
到这里,如果你不想使用elastic-search,也不想使用python,你就是想用java,mysql,或者你superset,redash,hadoop啥的玩得很熟,没问题,根据数据协议你应该很容易的把数据放到你需要的地方进行研究.
当然,我更希望你把代码贡献到connector里面,pr给我,既提高自己的代码水平,又方便了需要使用的人,岂不快哉?
仅仅只是把数据换一个存储,系统就发生了不可思议的变化.
可以参考官方文档进行安装:https://www.elastic.co/guide/en/elastic-stack/current/installing-elastic-stack.html
也可以用以下命令来完成:
$ #下载xpack
$ wget https://artifacts.elastic.co/downloads/packs/x-pack/x-pack-6.1.1.zip
$ #下载es
$ wget https://artifacts.elastic.co/downloads/elasticsearch/elasticsearch-6.1.1.zip
$ unzip elasticsearch-6.1.1.zip
$ cd elasticsearch-6.1.1/
$ #为es安装xpcck插件,就是刚刚下载的那个x-pack-6.1.1.zip,格式为file://+其路径
$ bin/elasticsearch-plugin install file:///path/to/file/x-pack-6.1.1.zip
$ #用fooltrader中的elasticsearch.yml覆盖es默认配置
$ cp ../fooltrader/config/elasticsearch.yml config/
$ #启动es,可根据自己的情况更改heap大小,<=32g
$ ES_JAVA_OPTS="-Xms8g -Xmx8g" ./bin/elasticsearch
$
$ #下载kibana
$ wget https://artifacts.elastic.co/downloads/kibana/kibana-6.1.1-linux-x86_64.tar.gz
$ tar -xzf kibana-6.1.1-linux-x86_64.tar.gz
$ cd kibana-6.1.1-linux-x86_64/
$ #为kibana安装xpcck插件,就是刚刚下载的那个x-pack-6.1.1.zip,格式为file://+其路径
$ bin/kibana-plugin install file:///path/to/file/x-pack-6.1.1.zip
$ #用fooltrader中的kibana.yml覆盖kibana默认配置
$ cp ../fooltrader/config/kibana.yml config/
$ ./bin/kibana到这里,我还是默认你在fooltrader的ipython环境下.
In [1]: from fooltrader.connector import es_connector
#股票元信息->es
In [2]: es_connector.stock_meta_to_es()
#指数数据->es
In [3]: es_connector.index_kdata_to_es()
#个股k线->es
In [4]: es_connector.stock_kdata_to_es()
#你也可以多开几个窗口,指定范围,提高索引速度
In [4]: es_connector.stock_kdata_to_es(start='002000',end='002999')
#财务数据->es
In [5]: es_connector.balance_sheet_to_es()
In [5]: es_connector.income_statement_to_es()
In [5]: es_connector.cash_flow_statement_to_es()然后,我们简单的来领略一下它的威力
查询2017年中报净利润top 5
curl -XPOST 'localhost:9200/income_statement/doc/_search?pretty&filter_path=hits.hits._source' -H 'Content-Type: application/json' -d'
{
"query": {
"range": {
"reportDate": {
"gte": "20170630",
"lte": "20170630"
}
}
},
"size": 5,
"sort": [
{
"netProfit": {
"order": "desc"
}
}
]
}
'
{
"hits": {
"hits": [
{
"_source": {
"exchangeGains": 1.3242E10,
"netProfit": 1.827E9,
"securityId": "stock_sh_601318",
"investmentIncome": 2.0523E10,
"operatingProfit": 7.8107E10,
"accumulatedOtherComprehensiveIncome": 2.0E8,
"attributableToMinorityShareholders": 6.5548E10,
"sellingExpenses": 1.0777E10,
"investmentIncomeFromRelatedEnterpriseAndJointlyOperating": "398259000000.00",
"id": "stock_sh_601318_20170630",
"minorityInterestIncome": 6.238E10,
"code": "601318",
"otherComprehensiveIncome": 6.5506E10,
"nonOperatingIncome": 4.006E9,
"financingExpenses": 0.0,
"reportEventDate": "2017-08-18",
"netProfitAttributedToParentCompanyOwner": 5.778E10,
"disposalLossOnNonCurrentLiability": 9.01E8,
"incomeFromChangesInFairValue": -2.56E8,
"incomeTaxExpense": 2.2E7,
"operatingTotalCosts": 3.4139E11,
"assetsDevaluation": 8.75E8,
"EPS": 1.9449E10,
"operatingCosts": 9.4E7,
"attributableToOwnersOfParentCompany": 1.58E8,
"ManagingCosts": 6.402E10,
"totalProfits": 8.403E9,
"dilutedEPS": 2.4575E10,
"reportDate": "20170630",
"businessTaxesAndSurcharges": 9.442E9,
"operatingRevenue": 4.63765E11,
"nonOperatingExpenditure": 1.35892E11
}
]
}
}
}实际上REST接口天然就有了,做跨平台接口非常方便,根据数据协议 和ES DSL可以非常方便的进行查询和聚合计算.
(文档待完善)
(文档待完善)
(文档待完善)
- 爬虫代理框架
可配置代理服务器池和并发数,从而提高爬虫的健壮性
- 数据抓取
- A股标的信息抓取
- A股tick数据抓取
- A股日线数据抓取
- A股财务数据抓取
- A股事件抓取
数据的处理方式是,先定义数据协议,再寻找数据源,这样做的好处是:数据协议的稳定为整个系统的稳定打下坚实的基础,多数据源比较提高数据准确性,多数据源聚合提高数据完整性.
-
常用技术指标计算(ma,ema,macd等)
-
回测框架
小金属涨疯了,但相关的上市公司股价还在历史新低,我是不是可以买一点?
金叉买,死叉卖,在不同级别上表现如何?在不同标的上表现如何?
相同的策略,如何快速的在所有标的上回测,并进行对比?
利润增长,股价也增长?或者提前反映?滞后反映?各种表现的比例如何?
各个策略之间如何通信,从而形成合力?
- 交易DSL设计
- WEB管理界面,向导式生成策略
- 实时行情及kafka实时计算
- 集成vnpy的交易接口
- 期货数据抓取
- 港股数据抓取
QQ群:300911873
如果你喜欢该项目,请加星支持一下,并在申请入群时告知github user name.