QuantLET example: Moving Average Cross Over

http://quantlet.net

This example shows the arrangement of Quantlet's building blocks, 'QLets', into directional graphs called 'streams'. Other features related to concurrency, reactive event processing and the simulation meta-model will be shown on future examples.

This is the 'hello world' of algorithmic strategies. A simplification of MAC-O described in Algorithmic Trading Strategies by M. Schoeffel.

Generation of buy and sell signals whenever the spot price of a simulated asset crosses up (buy signal) or down (sell signal) a historical moving average. Please bare in mind this is a very simple and unrealistic example of quantlet, an on-going research work.

Pre-requisites

Ubuntu:

sudo apt-get install python-numpy python-scipy python-matplotlib ipython ipython-notebook python-pandas python-sympy python-nose pandoc

sudo apt-get install 'freetype*'

Pip:

ipython[all] numpy pipe scipy matplotlib quantlet PySide

First things first...

Environment setup.. inline plots, making plots a bit bigger, etc..



In [13]:

    
%pylab inline









    



Populating the interactive namespace from numpy and matplotlib






    



WARNING: pylab import has clobbered these variables: ['plot', 'ma']
`%matplotlib` prevents importing * from pylab and numpy

Sanity check



In [14]:

    
x = np.linspace(0, 2 * np.pi, 100)
y1 = np.sin(x)
y2 = np.sin(3 * x)
plt.fill(x, y1, 'b', x, y2, 'r', alpha=0.3)
plt.show()

Looks good. Ready to start. Quantlet imports... going verbose for the sake of transparency



In [15]:

    
from quantlet import describe, clear
from quantlet.ds.rw import random_walk
from quantlet.ds.plot import plot
from quantlet.function.ma import ma
from quantlet.strats.momentum import maco
from quantlet.strats.portfolio import fixed_sized_delayed

I like my quantlet plots a bit large, so let's adjust accordingly



In [16]:

    
pylab.rcParams['figure.figsize'] = (15.0, 12.0)

Generating a 'random walk'

Plotting just a random walk. Two QLets: random_walk and plot. A different series is produced every time a random_walk dataset is chained



In [17]:

    
random_walk() >> plot

What does random_walk do?



In [18]:

    
random_walk?



In [19]:

    
clear()

Generation of buys and sells signals

Now adding all pieces, each explained:

Datasets:

random_walk(): a geometric brownian motion input dataset for ilustration
plot: plot output dataset

QLets:

ma: generates a moving average of the randowm walk
maco: a 'moving average cross over' strategy as a QLet, generates buys and sells signals based on crossover of price and historical moving average
fixed_sized_delayed: portfolio and execution strategy, given an initial cash value, buys a fixed lot on a buy signal and sells the same lot on a sell signal

Literally one line, what gives...



In [20]:

    
random_walk() \
>> ma \
>> maco \
>> fixed_sized_delayed \
>> plot

In which:

- blue: random walk simulation of an asset price
- green: smoothing as a moving average
- red triangles: sell signals
- green triangles: buy signals
- yellow: porfolio value (porfolio value = cash value + stock value)
- pinkish: portfolio value associated to asset holdings and price

what is the current setup?



In [21]:

    
describe()



In [22]:

    
clear()

Trying a different shock: any differences in reversing buy and sell signals?



In [23]:

    
%time \
random_walk() \
>> ma \
>> maco(reverse=True) \
>> fixed_sized_delayed \
>> plot









    



CPU times: user 164 ms, sys: 0 ns, total: 164 ms
Wall time: 186 ms

Not much...