IPython 与 RQAlpha

加载 RQAlpha magic

In [3]:

%load_ext rqalpha

The rqalpha extension is already loaded. To reload it, use:
  %reload_ext rqalpha

查看 RQAlpha magic 帮助

我们可以通过 %%rqalpha 直接在 cell 中运行回测代码。%%rqalpha 后面的参数等价于在 CLI 中后面的 rqalpha run 的参数

In [2]:

%%rqalpha -h
""

Usage: ipykernel_launcher.py [OPTIONS]
 
  Start to run a strategy
 
Options:
  -h, --help                      Show this message and exit.
  -d, --data-bundle-path PATH
  -f, --strategy-file PATH
  -s, --start-date DATE
  -e, --end-date DATE
  -bm, --benchmark TEXT
  -mm, --margin-multiplier FLOAT
  -a, --account TEXT...           set account type with starting cash
  -fq, --frequency [1d|1m|tick]
  -rt, --run-type [b|p]
  --resume
  --source-code TEXT
  -l, --log-level [verbose|debug|info|error|none]
  --disable-user-system-log       disable user system log stdout
  --disable-user-log              disable user log stdout
  --locale [cn|en]
  --extra-vars TEXT               override context vars
  --enable-profiler               add line profiler to profile your strategy
  --config TEXT                   config file path
  -mc, --mod-config TEXT...       mod extra config
  --stock-t1 / --no-stock-t1      [sys_accounts] enable/disable stock T+1
  --report PATH                   [sys_analyser] save report
  -o, --output-file PATH          [sys_analyser] output result pickle file
  -p, --plot / --no-plot          [sys_analyser] plot result
  --plot-save TEXT                [sys_analyser] save plot to file
  --progress                      [sys_progress]show progress bar
  --no-short-stock / --short-stock
                                  [sys_risk] enable stock shorting
  --signal                        [sys_simulation] exclude match engine
  -sp, --slippage FLOAT           [sys_simulation] set slippage
  -cm, --commission-multiplier FLOAT
                                  [sys_simulation] set commission multiplier
  -me, --match-engine [current_bar|next_bar|last|best_own|best_counterparty]
                                  [Deprecated][sys_simulation] set matching
                                  type
  -mt, --matching-type [current_bar|next_bar|last|best_own|best_counterparty]
                                  [sys_simulation] set matching type

使用 %%rqalpha 进行回测

In [3]:

%%rqalpha -s 20100101 -e 20170505 -p -bm 000001.XSHG --account stock 100000
 
def init(context):
    context.stocks = ['000300.XSHG', '000905.XSHG', '000012.XSHG']
 
 
def handle_bar(context, bar_dict):
    [hs, zz, gz] = context.stocks
    hs_history20 = history_bars(hs, 20, '1d', 'close')
    zz_history20 = history_bars(zz, 20, '1d', 'close')
 
    hsIncrease = hs_history20[-1] - hs_history20[0]
    zzIncrease = zz_history20[-1] - zz_history20[0]
 
    positions = context.portfolio.positions
    [hsQuality, zzQuality, gzQuality] = [positions[hs].quantity, positions[zz].quantity, positions[gz].quantity]
    if hsIncrease < 0 and zzIncrease < 0:
        if hsQuality > 0: order_target_percent(hs, 0)
        if zzQuality > 0: order_target_percent(zz, 0)
        order_target_percent(gz, 1)
    elif hsIncrease < zzIncrease:
        if hsQuality > 0: order_target_percent(hs, 0)
        if gzQuality > 0: order_target_percent(gz, 0)
        order_target_percent(zz, 1)
    else:
        if zzQuality > 0: order_target_percent(zz, 0)
        if gzQuality > 0: order_target_percent(gz, 0)
        order_target_percent(hs, 1)
        #logger.info("positions hs300: " + str(hsQuality) + ", zz500: " + str(zzQuality) + ", gz: " + str(gzQuality))

获取回测报告

运行完回测后，报告会自动存储到 report 变量中。可以直接通过report 变量获取当次回测的结果。

另外 rqalpha 的 mod 的输出会自动存储在 results 变量中。

In [4]:

results.keys()

Out[4]:

dict_keys(['sys_analyser'])

In [5]:

report.keys()

Out[5]:

dict_keys(['summary', 'trades', 'portfolio', 'benchmark_portfolio', 'stock_account', 'stock_positions'])

In [6]:

report.trades[:5]

Out[6]:

In [7]:

report.portfolio[:5]

Out[7]:

In [8]:

report.stock_positions[:5]

Out[8]:

使用 run_func 运行回测

In [9]:

config = {
  "base": {
    "start_date": "2010-01-01",
    "end_date": "2017-05-05",
    "benchmark": "000001.XSHG",
    "accounts": {
        "stock": 100000
    }
  },
  "extra": {
    "log_level": "info",
  },
  "mod": {
    "sys_analyser": {
      "enabled": True,
      "plot": True,
    },
  }
}
 
 
from rqalpha.api import *
from rqalpha import run_func
 
 
def init(context):
    context.stocks = ['000300.XSHG', '000905.XSHG', '000012.XSHG']
 
 
def handle_bar(context, bar_dict):
    [hs, zz, gz] = context.stocks
    hs_history20 = history_bars(hs, 20, '1d', 'close')
    zz_history20 = history_bars(zz, 20, '1d', 'close')
 
    hsIncrease = hs_history20[-1] - hs_history20[0]
    zzIncrease = zz_history20[-1] - zz_history20[0]
 
    positions = context.portfolio.positions
    [hsQuality, zzQuality, gzQuality] = [positions[hs].quantity, positions[zz].quantity, positions[gz].quantity]
    if hsIncrease < 0 and zzIncrease < 0:
        if hsQuality > 0: order_target_percent(hs, 0)
        if zzQuality > 0: order_target_percent(zz, 0)
        order_target_percent(gz, 1)
    elif hsIncrease < zzIncrease:
        if hsQuality > 0: order_target_percent(hs, 0)
        if gzQuality > 0: order_target_percent(gz, 0)
        order_target_percent(zz, 1)
    else:
        if zzQuality > 0: order_target_percent(zz, 0)
        if gzQuality > 0: order_target_percent(gz, 0)
        order_target_percent(hs, 1)
 
 
results = run_func(init=init, handle_bar=handle_bar, config=config)

In [10]:

report = results["sys_analyser"]

In [11]:

report["trades"][:5]

Out[11]:

In [ ]: