Simulate a smooth timeseries in Python
Question:
I want to use Python to generate some data that will simulate a fairly smooth wandering timeseries – similar to the following plot.
I originally started with a random walk, but if I made my standard deviation small, the data did not wander enough, and if I made the standard deviation too large, the plot is not smooth at all.
Is there a better way to approach this?
Answers:
Just apply a rolling moving average to your results:
from numpy import sqrt
annualized_vol = .30 # 30%
lag = 30
random_normals = np.random.randn(1000) # 1,000 trading days.
daily_vol = sqrt(annualized_vol) * sqrt(1 / 252.) # 252 trading days in a year.
random_daily_log_returns = random_normals * daily_vol
df = pd.DataFrame(random_daily_log_returns).cumsum()
df.rolling(lag).mean().plot()
The bigger the lag and the smaller the vol, the smoother the series
I want to use Python to generate some data that will simulate a fairly smooth wandering timeseries – similar to the following plot.
I originally started with a random walk, but if I made my standard deviation small, the data did not wander enough, and if I made the standard deviation too large, the plot is not smooth at all.
Is there a better way to approach this?
Just apply a rolling moving average to your results:
from numpy import sqrt
annualized_vol = .30 # 30%
lag = 30
random_normals = np.random.randn(1000) # 1,000 trading days.
daily_vol = sqrt(annualized_vol) * sqrt(1 / 252.) # 252 trading days in a year.
random_daily_log_returns = random_normals * daily_vol
df = pd.DataFrame(random_daily_log_returns).cumsum()
df.rolling(lag).mean().plot()
The bigger the lag and the smaller the vol, the smoother the series