finlab 0.1.2.dev1

A Financial backtesting framework for stock portfolio selection

FinLab

A Financial backtesting framework for stock portfolio selection

Installation

FinLab is a easy to install python package aims to help developing stock portfolio strategies.

pip install finlab

Getting Start

There are two main functionality of finlab in the alpha development. First, taiwan stock market historical data can be obtain by online.Data interface:

import finlab

close = finlab.data.get('close')
close.head()

date	1101	1203	...	1726	2330
2015-1-6	43.0	23.7		82.0	139.5
...	...	...	...	...	...
2015-1-8	42.25	23.8		82.0	133.5

Stock technical indicator calculation

With the integration of TA-Lib package, it is easily to calculate technical indicator using wrapper function finlab.data.talib:

rsi = finlab.data.talib('RSI', timeperiod=14)
rsi.head()

With the simple code above, finlab.data.talib calculates RSI indicator for all stocks in the pool.

Perform dataFrame methods

It is easy to perform stock selection using methods provided by pandas dataframe

# plot stock price
close['1101'].plot()

# caculate monthly average price
sma20 = close.rolling(20).mean()

# select stocks with the price above sma20
signal = close > sma20

# print the stocks statisfy the condition above
last_signal = signal.iloc[-1]
last_signal[last_signal == True] # simplied: last_signal[last_signal]

Stock Backtesting

It is easy to write a condition using the similar calculation as above to create a backtest strategy: For example, let's select 10 stock that have the smallest price.

def strategy(data):
    close = data.get('close')
    buy = close.smallest(10)
    sell = None
    resample_frequency = '60d'
    return buy, sell, resample_frequency

report = finlab.backtest(strategy)
report

FinLab will NEVER record your strategy in the cloud without your permission.

We value your privacy! For every single backtest, an private container is created (using AWS Lambda). After each backtesting, all local variables during simlation is destroyed.

Authors

• Cheng-Yu Han - Ph.D. FinLab