Python ggplot issues plotting >8 stocks and legend is cutoff

Question

First let me say ggplot for Python is the beginning of something great and kudos to the developers for putting in the work. Currently I'm having two major issues with the same plot. If I plot 8 stocks or less the image looks good except that the legend runs off the figure area (Problem 1). If I plot more than 8 stocks the plot triggers some erratic line(s) that is clearly not representative of the data. Additionally the legend does not resize and instead leaves off the additional stock tickers (Problem 2). Any help is appreciated. Thanks!

Decent Plot Code:

import datetime
from ggplot import *
import pandas.io.data as web
import pandas as pd
import numpy as np

start = datetime.datetime(2014,1,1)
end = datetime.datetime(2014, 3,19)

stocks = ['APO','AVG','FI','ANIK','CELG','PACW','CBOE','BIIB']

stockData = {}
for ticker in stocks:
    stockData[ticker] = web.get_data_yahoo(ticker, start, end)

price = pd.DataFrame({tic: data['Adj Close'] for tic, data in stockData.iteritems()})

returns = price.pct_change()
returns = returns.apply(cumsum)

rt = returns.index
returns['Date'] = rt


# plotting the cum performance for each security

ret = pd.melt(returns, id_vars='Date') 
plot = ggplot(aes(x='Date', y='value', color='variable'),data=ret) +geom_line()

# plotting the equity curve of the theoretical portfolio

zt = returns
del zt['Date']
zt = zt.apply(np.sum, axis=1)
z = pd.DataFrame(zt, index=zt.index)
z['Date'] = rt
z.columns = ['equity curve', 'Date']


ret2 = pd.melt(z, id_vars='Date') 
plot2 = ggplot(aes(x='Date', y='value'),data=ret2) +geom_line()

print plot
print plot2

BAD Plot Code:

import datetime
from ggplot import *
import pandas.io.data as web
import pandas as pd
import numpy as np

start = datetime.datetime(2014,1,1)
end = datetime.datetime(2014, 3,19)

stocks = ['APO','AVG','FI','ANIK','CELG','PACW','CBOE','BIIB','ISIS', 'SDRL'] # <-- notice two additional tickers

stockData = {}
for ticker in stocks:
    stockData[ticker] = web.get_data_yahoo(ticker, start, end)

price = pd.DataFrame({tic: data['Adj Close'] for tic, data in stockData.iteritems()})

returns = price.pct_change()
returns = returns.apply(cumsum)

rt = returns.index
returns['Date'] = rt


# plotting the cum performance for each security

ret = pd.melt(returns, id_vars='Date') 
plot = ggplot(aes(x='Date', y='value', color='variable'),data=ret) +geom_line()

# plotting the equity curve of the theoretical portfolio

zt = returns
del zt['Date']
zt = zt.apply(np.sum, axis=1)
z = pd.DataFrame(zt, index=zt.index)
z['Date'] = rt
z.columns = ['equity curve', 'Date']


ret2 = pd.melt(z, id_vars='Date') 
plot2 = ggplot(aes(x='Date', y='value'),data=ret2) +geom_line()

print plot
print plot2

Answer 1

For the Problem 2, this is because ggplot run out of colors, you can add more colors to fix the problem, just add following code to the beginning of your code:

import ggplot as gg
gg.colors.COLORS.extend(["#ff0000", "#00ff00", "#0000ff"]) 

For the Problem 1, it seems that we need to place the legend after create the figure:

ret = pd.melt(returns, id_vars='Date').dropna()
plot = ggplot(aes(x='Date', y='value', color='variable'), data=ret) +geom_line()
fig = plot.draw()
ax = fig.axes[0]
offbox = ax.artists[0]
offbox.set_bbox_to_anchor((1, 0.5), ax.transAxes)

here is the result: