sc-backtest 0.1.14

Index future simple stat and time-series test module

Simple Backtest Module (Personal Usage)

Chang Sun | 孙畅

Email

Install and Update

pip install --upgrade sc-backtest

or (if slow)

pip install --upgrade -i https://pypi.tuna.tsinghua.edu.cn/simple sc-backtest

Simple Test

• Check for factor validity
  • Statistical:
    • CDF
    • Markout
    • Hist
    • ...
  • Time-Series:
    • Sign-Trade
    • Value-Trade
    • ...

# x: factors
# y: asset's future ret

import pandas as pd
import numpy as np
from sc_backtest import simpletest, dataset

data = dataset.get_data('adj_close_price', frequency=5)
x = data.pct_change(240).iloc[:, 0]
y = data.pct_change().shift(-1).iloc[:, 0]
st = simpletest()
st.plot_cdf(x, y)
st.plot_composite(x, y)

Backtest (bt)

• Backtest
  • get_report
  • get_pnl_plot
  • round_test
  • ...

# x: factors
# y: asset's future ret

import pandas as pd
import numpy as np
from sc_backtest import simpletest, bt, dataset

data = dataset.get_data('adj_close_price', frequency=5)
x = data.pct_change(240).iloc[:, 0]
y = data.pct_change().shift(-1).iloc[:, 0]
st = simpletest()
data = st.simple_pnl(x, y, data_return=True)
report = bt.get_report(data['delta_med'], y)
bt.get_pnl_plot(data['delta_med'], y)

Technical Analysis (ta)

Reference: ta

import pandas as pd
import numpy as np
from sc_backtest import ta, dataset

data = dataset.get_data('adj_close_price', frequency=5)
macd_diff = ta.trend.macd(data.iloc[:, 0]).macd_diff()

Technical Analysis2 (ta2)

Variou moving average function and stat model

• sma, ema, wma, ...
• rsi, atr, ...
• z_score, div_std, de_mean, ...

import pandas as pd
import numpy as np
from sc_backtest import ta2, dataset

data = dataset.get_data('adj_close_price', frequency=5)
wma = ta2.wma(data.iloc[:, 0], window=5)

DataFrame Function (df_func)

# example

def df_sim_yoy(window):
    def _sim_yoy(x):
        temp = pd.DataFrame(x)
        return (temp - temp.shift(window)) / ((temp.abs() + temp.shift(window).abs()) / 2)

    _sim_yoy.__name__ = f'df_sim_yoy_{int(window)}'
    return _sim_yoy

Example

Input your factor and underlying asset's future return with index type as DatetimeIndex and get the composite factor analysis stat and simple-pnl time-series plots.

# x: factors
# y: asset's future ret

import pandas as pd
import numpy as np
from sc_backtest import simpletest, bt, dataset, ta2

data = dataset.get_data('adj_close_price', frequency=5)
x = data.pct_change().apply(lambda x: ta2.ema(x, window=240))
y = data.pct_change().shift(-1)
st = simpletest()
st.plot_composite_cs(x, y, ic=True, horizon=5)
bt.get_pnl_plot(x, y, alpha=True)