# Import libraries
import pandas as pd
import cufflinks as cf

import PortfolioAnalyser as pa

# Set cufflinks offline
cf.go_offline()

Continuando o anterior post vamos fazer agora o gráfico dos ETFs com o código limpo antes de proceder

# Ler IS3Q, colocar index como datetime e dar o nome do ticker à coluna 
IS3Q = pd.read_csv('IS3Q.csv', index_col='Date', parse_dates=True)[['Price']]
IS3Q.columns = ['IS3Q']

# Ler IWDA, colocar index como datetime e dar o nome do ticker à coluna 
IWDA = pd.read_csv('IWDA.csv', index_col='Date', parse_dates=True)[['Price']]
IWDA.columns = ['IWDA']

# Fazer dataframe ETFs com merge dos dois ETFs
ETFs = pa.merge_time_series(IS3Q, IWDA, how='inner')
ETFs = ETFs.sort_index(ascending=True)

# Fazer o growth index
ETFs_gi = pa.compute_growth_index(ETFs)

# Fazer o gráfico plotly
ETFs_gi.iplot(dimensions=pa.dimensions)

Para fazer os drawdowns apenas temos de fazer a função para tal da biblioteca PortfolioAnalyser. Assim:

# Fazer DataFrame dos Drawdowns e associar a uma variável
DD = pa.compute_drawdowns(ETFs_gi)
DD

	IS3Q	IWDA
Date
2014-10-08	0.00	0.00
2014-10-09	0.00	0.00
2014-10-10	-1.48	-1.49
2014-10-13	-2.29	-2.31
2014-10-14	-2.55	-2.62
...	...	...
2020-08-27	-6.87	-7.07
2020-08-28	-7.50	-7.79
2020-08-31	-7.85	-8.14
2020-09-01	-7.78	-7.83
2020-09-02	-5.93	-6.19

1485 rows × 2 columns

# Fazer o gráfico plotly
DD.iplot(dimensions=pa.dimensions)

Nada mais simples que isto. Apenas 2 linhas de código.

Vamos fazer agora a tabela de performance (não se esqueçam que têm de usar o nome da variável/DataFrame que tem ambos os ETFs):

# Performance since inception
pa.compute_performance_table(ETFs_gi)

Performance from 2014-10-08 to 2020-09-02 (≈ 5.9 years)

	CAGR	Return	StdDev	Sharpe	Max DD	MAR
IS3Q	11.48%	90.12%	17.24%	0.67	-32.30%	0.36
IWDA	10.51%	80.46%	17.54%	0.60	-33.64%	0.31

ETFs_gi

	IS3Q	IWDA
Date
2014-10-08	100.00	100.00
2014-10-09	100.56	100.44
2014-10-10	99.08	98.95
2014-10-13	98.26	98.13
2014-10-14	98.00	97.81
...	...	...
2020-08-27	188.22	178.78
2020-08-28	186.94	177.38
2020-08-31	186.23	176.72
2020-09-01	186.38	177.32
2020-09-02	190.12	180.46

1485 rows × 2 columns

Por defeito a tabela de performanance mostra a performance desde o início da DataFrame. Mas a função tem uma opção que se passarem o número de anos ela vai calcular a tabela de performance desse período.

# Performance a 3 anos
pa.compute_performance_table(ETFs_gi, years=3)

Performance from 2017-09-01 to 2020-09-02 (3 years)

	CAGR	Return	StdDev	Sharpe	Max DD	MAR
IS3Q	12.08%	40.80%	17.67%	0.68	-32.30%	0.37
IWDA	10.23%	33.95%	18.03%	0.57	-33.64%	0.30

# Performance a 10 anos
pa.compute_performance_table(ETFs_gi, years=10)

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-10-e3ea5c17c681> in <module>
      1 # Performance a 10 anos
----> 2 pa.compute_performance_table(ETFs_gi, years=10)

~\Google Drive\_GitHub\Articles_and_studies\Financial_Markets_Programming\PortfolioAnalyser.py in compute_performance_table(dataframe, years, freq)
    165 
    166     else:
--> 167         dataframe = filter_by_date(dataframe, years)
    168         df = pd.DataFrame([compute_cagr(dataframe, years=years), compute_return(dataframe),
    169                            compute_StdDev(dataframe), compute_sharpe(dataframe),

~\Google Drive\_GitHub\Articles_and_studies\Financial_Markets_Programming\PortfolioAnalyser.py in filter_by_date(dataframe, years)
    211         pass
    212     elif start < dataframe.index[0]:
--> 213         raise ValueError("You seem to be selecting a date before the start of the time series")
    214     else:
    215         start = str(dataframe.loc[:start].tail(1).index)[16:26]

ValueError: You seem to be selecting a date before the start of the time series