# QAAccount与MongoDB

在使用QAAccoount类进行回测时，MongoDB数据库quantaxis中，会涉及到的数据表有 5 个，分别是：

portfolio
user
account
risk
strategy

回测中的逻辑关系

示例

创建资产用户/资产组合/帐户

Quantaxis v1.3.0以后, QA_Account需要由组合来进行创建(推荐)。为避免出过多因使用未来版本会降级的函数导致的警告信息。

1 2	import warnings warnings.filterwarnings('ignore')

创建用户

用户名为qaacc,密码为qaacc

1
2
3

import QUANTAXIS as QA

user = QA.QA_User(username='qaacc', password='qaacc')

### 创建资产组合

1	portfolio = user.new_portfolio('stock')

创建帐户

可以创建多市场的帐户，这里以创建股票市场帐户为例。

1	account = portfolio.new_account(account_cookie='lvjun')

账户的初始资金/初始仓位

默认账户是无仓位, 默认现金 1,000,000 RMB

1	account.init_assets

输出：

1	{'cash': 1000000, 'hold': {}}

1	account.init_cash

输出：

1	account.init_hold

输出：

1	Series([], Name: amount, dtype: object)

简单回测

用帐户account_cookie='lvjun'进行简单回测。Quantaxis版本为 1.10.19。

新建策略

MACD_JCSC策略思想是MACD指标出现金叉（DIF向上突破DEA)买入，出现死叉（DIF向下跌破DEA）卖出。

import QUANTAXIS as QA
import numpy as np
import pandas as pd
import datetime

# define the MACD strategy


def MACD_JCSC(dataframe, SHORT=12, LONG=26, M=9):
    """
    1.DIF向上突破DEA，买入信号参考。
    2.DIF向下跌破DEA，卖出信号参考。
    """
    CLOSE = dataframe.close
    DIFF = QA.EMA(CLOSE, SHORT) - QA.EMA(CLOSE, LONG)
    DEA = QA.EMA(DIFF, M)
    MACD = 2*(DIFF-DEA)

    CROSS_JC = QA.CROSS(DIFF, DEA)
    CROSS_SC = QA.CROSS(DEA, DIFF)
    ZERO = 0
    return pd.DataFrame({'DIFF': DIFF, 'DEA': DEA, 'MACD': MACD, 'CROSS_JC': CROSS_JC, 'CROSS_SC': CROSS_SC, 'ZERO': ZERO})

初始化回测broker及保存策略

策略源码保存可以保存到网络侧(127.0.0.1:8010)，也可以保存到本地。

1
2
3

Broker = QA.QA_BacktestBroker()
# 策略保存到本地, 当含参`if_save=True`时, 会保存到本地。 在数据表·strategy`中可以看到具体信息。
QA.QA_SU_save_strategy('MACD_JCSC', 'Indicator', account.account_cookie)

选择测试标的物数据

# get data from mongodb
codelist = ['000001']
data = QA.QA_fetch_stock_day_adv(codelist, '2017-09-01', '2018-05-20')
data = data.to_qfq()

计算指标信号

指标的计算可以在回测前，也可以在回测中进行。区别在于回测前的计算则是批量计算，效率较高。而回测中进行计算，效率略低，但代码量较小，易于理解。

1
2
3

ind = data.add_func(MACD_JCSC)
print("金叉信号出现：%d 次" % ind['CROSS_JC'].value_counts()[1])
print("死叉信号出现：%d 次" % ind['CROSS_SC'].value_counts()[1])

1 2	金叉信号出现：8 次死叉信号出现：7 次

列出出现信号时的数据。

1	ind.loc[(ind['CROSS_JC'] == 1) \| (ind['CROSS_SC'] == 1)]

输出：

回测

通过迭代产生Dataframe截面数据的方式，对产生信号的标的物进行交易回测。

for items in data_forbacktest.panel_gen:
    for item in items.security_gen:
        daily_ind = ind.loc[item.index]
        if daily_ind.CROSS_JC.iloc[0] > 0:
            order = account.send_order(
                code=item.code[0],
                time=item.date[0],
                amount=1000,
                towards=QA.ORDER_DIRECTION.BUY,
                price=0,
                order_model=QA.ORDER_MODEL.CLOSE,
                amount_model=QA.AMOUNT_MODEL.BY_AMOUNT
            )

            Broker.receive_order(QA.QA_Event(order=order, market_data=item))
            trade_mes = Broker.query_orders(account.account_cookie, 'filled')
            res = trade_mes.loc[order.account_cookie, order.realorder_id]
            order.trade(res.trade_id, res.trade_price,res.trade_amount, res.trade_time)
        elif daily_ind.CROSS_SC.iloc[0] > 0:
            if account.sell_available.get(item.code[0], 0) > 0:
                order = account.send_order(
                    code=item.code[0],
                    time=item.date[0],
                    amount=account.sell_available.get(item.code[0], 0),
                    towards=QA.ORDER_DIRECTION.SELL,
                    price=0,
                    order_model=QA.ORDER_MODEL.MARKET,
                    amount_model=QA.AMOUNT_MODEL.BY_AMOUNT
                )
                Broker.receive_order(QA.QA_Event(order=order, market_data=item))
                trade_mes = Broker.query_orders(account.account_cookie, 'filled')
                res = trade_mes.loc[order.account_cookie, order.realorder_id]
                order.trade(res.trade_id, res.trade_price,res.trade_amount, res.trade_time)
    account.settle()

输出：

QAACCOUNT ==> receive deal  Time 2018-01-02 00:00:00/ Code:000001/ Price:12.48/ TOWARDS:1/ Amounts:1000
QAACCOUNT ==> receive deal  Time 2018-01-03 00:00:00/ Code:000001/ Price:12.32/ TOWARDS:-1/ Amounts:1000
QAACCOUNT ==> receive deal  Time 2018-01-12 00:00:00/ Code:000001/ Price:12.34/ TOWARDS:1/ Amounts:1000
QAACCOUNT ==> receive deal  Time 2018-01-29 00:00:00/ Code:000001/ Price:12.68/ TOWARDS:-1/ Amounts:1000
QAACCOUNT ==> receive deal  Time 2018-03-08 00:00:00/ Code:000001/ Price:11.03/ TOWARDS:1/ Amounts:1000
QAACCOUNT ==> receive deal  Time 2018-03-26 00:00:00/ Code:000001/ Price:10.05/ TOWARDS:-1/ Amounts:1000
QAACCOUNT ==> receive deal  Time 2018-04-10 00:00:00/ Code:000001/ Price:10.4/ TOWARDS:1/ Amounts:1000

风险与绩效分析

可以手工完成分析，或是进入 Web 界面调用数据分析。

风险分析

1 2	Risk = QA.QA_Risk(account) Risk.plot_assets_curve()

输出：

绩效分析

1 2	Performance = QA.QA_Performance(account) Performance.base_message(Performance.pnl)

输出：

{
    "status": 200,
    "result": {
        "total_profit": 340.0,
        "total_loss": -1140.0,
        "total_pnl": 0.3,
        "trading_amounts": 3,
        "profit_amounts": 1,
        "loss_amounts": 2,
        "even_amounts": 0,
        "profit_precentage": 0.33,
        "loss_precentage": 0.67,
        "even_precentage": 0.0,
        "average_profit": 340.0,
        "average_loss": -570.0,
        "average_pnl": 0.6,
        "max_profit": 340.0,
        "max_loss": -980.0,
        "max_pnl": 0.35,
        "netprofio_maxloss_ratio": 0.82,
        "continue_profit_amount": 1,
        "continue_loss_amount": 1,
        "average_holdgap": "12 days 00:00:00",
        "average_profitholdgap": "17 days 00:00:00",
        "average_losssholdgap": "9 days 12:00:00",
        "buyopen": {
            "total_profit": 340.0,
            "total_loss": -1140.0,
            "total_pnl": 0.3,
            "trading_amounts": 3,
            "profit_amounts": 1,
            "loss_amounts": 2,
            "even_amounts": 0,
            "profit_precentage": 0.33,
            "loss_precentage": 0.67,
            "even_precentage": 0.0,
            "average_profit": 340.0,
            "average_loss": -570.0,
            "average_pnl": 0.6,
            "max_profit": 340.0,
            "max_loss": -980.0,
            "max_pnl": 0.35,
            "netprofio_maxloss_ratio": 0.82,
            "continue_profit_amount": 1,
            "continue_loss_amount": 1,
            "average_holdgap": "12 days 00:00:00",
            "average_profitholdgap": "17 days 00:00:00",
            "average_losssholdgap": "9 days 12:00:00"
        },
        "sellopen": {
            "total_profit": 0,
            "total_loss": 0,
            "total_pnl": 0,
            "trading_amounts": 0,
            "profit_amounts": 0,
            "loss_amounts": 0,
            "even_amounts": 0,
            "profit_precentage": 0,
            "loss_precentage": 0,
            "even_precentage": 0,
            "average_profit": 0,
            "average_loss": 0,
            "average_pnl": 0,
            "max_profit": 0,
            "max_loss": 0,
            "max_pnl": 0,
            "netprofio_maxloss_ratio": 0,
            "continue_profit_amount": 0,
            "continue_loss_amount": 0,
            "average_holdgap": "no trade",
            "average_profitholdgap": "no trade",
            "average_losssholdgap": "no trade"
        }
    }
}

Web浏览方式

浏览器输入：http://127.0.0.1:81 ,使用回测时创建用户名/密码。

风险与绩效分析
买卖信号记录

备注：

计算 Performance.continue_profit_amount 等指标时时，原代码中使用了 for _, item in pnl.pnl_money.items():, 在 Pandas ≥ 1.5 时，Series.iteritems(), DataFrame.iteritems() 及 HDFStore.iteritems() 均使用 obj.items 进行替代。¹

### What’s new in 1.5.0 (September 19, 2022)
- Deprecated `DataFrame.iteritems()`, `Series.iteritems()`, `HDFStore.iteritems()` in favor of [`DataFrame.items()`](https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.items.html#pandas.DataFrame.items "pandas.DataFrame.items"), [`Series.items()`](https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.Series.items.html#pandas.Series.items "pandas.Series.items"), `HDFStore.items()` ([GH 45321](https://github.com/pandas-dev/pandas/issues/45321))

### What’s new in 2.0.0 (April 3, 2023)
- Removed deprecated `Series.iteritems()`, `DataFrame.iteritems()`, use `obj.items` instead ([GH 45321](https://github.com/pandas-dev/pandas/issues/45321))

参考

Search - pandas 2.2.2 documentation (pydata.org)↩︎

橘子树下

【Quantaxis】QAAccount简单回测