Volume Overlay with Pandas

1
votes

I often see this in the context of matplotlib and open-high-low-close, but I'm wondering if you can add a volume overlay within the pandas framework. The final graph we would want would be close to the first one here: (Matplotlib - Finance volume overlay )

Say we have a DataFrame like such:

                       num  rolling_30  rolling_10  rolling_60  Volume
Date
2015-06-23        0.000219    0.000149    0.000168    0.000183       2
2015-06-25        0.000489    0.000162    0.000200    0.000188       3
2015-07-01        0.000164    0.000163    0.000190    0.000186       1
2015-07-02        0.000190    0.000166    0.000190    0.000187       1
2015-07-03        0.000269    0.000171    0.000198    0.000180       1
2015-07-04        0.000935    0.000196    0.000282    0.000193       2
2015-07-08        0.000154    0.000196    0.000288    0.000188       1
2015-07-11        0.000274    0.000202    0.000305    0.000190       1
2015-07-13        0.000872    0.000228    0.000380    0.000201       9

How can we get the ['num','rolling_30','rolling_10','rolling_60'] line chart with the bottom of the chart listing the daily volume? I can do a secondary_y to get volume on the right, but honestly that looks terrible. Need it to be the traditional volume bar-graph at the bottom of the chart.

2
pythonpandasmatplotlib

2 Answers

1
votes

The basic idea is to use .twinx to create a secondary y axis. Below is a short sample to do it. From the graph, you see that the left y axis is for price and moving averages, whereas the right y axis is for volumn.

import pandas as pd
import matplotlib.pyplot as plt

# your data
# ============================
print(df)


               num  rolling_30  rolling_10,  rolling_60  Volume
Date                                                           
2015-06-23  0.0002      0.0001       0.0002      0.0002       2
2015-06-25  0.0005      0.0002       0.0002      0.0002       3
2015-07-01  0.0002      0.0002       0.0002      0.0002       1
2015-07-02  0.0002      0.0002       0.0002      0.0002       1
2015-07-03  0.0003      0.0002       0.0002      0.0002       1
2015-07-04  0.0009      0.0002       0.0003      0.0002       2
2015-07-08  0.0002      0.0002       0.0003      0.0002       1
2015-07-11  0.0003      0.0002       0.0003      0.0002       1
2015-07-13  0.0009      0.0002       0.0004      0.0002       9

# plotting
# ===========================
fig, ax = plt.subplots(figsize=(10,8))
df.drop('Volume', axis=1).plot(ax=ax)
ax.legend(loc='best')
ax2 = ax.twinx()
df['Volume'].plot(kind='bar', ax=ax2, color='g', alpha=0.1)
ax2.set_ylim([0, ax2.get_ylim()[1] * 10])
ax2.legend(loc='best')

enter image description here

1
votes

So while Jianxun's answer is probably completely correct, it won't work on my system. I am on Pandas .17 and the newest matplotlib. I'm also on a Macbook. Basically if I try to share an X axis with two graphs, a line and bar graph, the first instantiated graph disappears.

The best way I can think to fix this is simply by doing two graphs on one (subplots) and hiding the X axis of the top graph. Here's what I did:

        fig = plt.figure(figsize=(10,6))
        ax1 = plt.subplot2grid((3,3), (0,0), colspan=3, rowspan=2)
        ax2 = plt.subplot2grid((3,3), (2,0), colspan=3)
        ax2.yaxis.tick_right()
        ax1.axes.get_xaxis().set_visible(False)
        fig.subplots_adjust(hspace=0)
        pct.drop('Volume', axis=1).plot(ax=ax1)
        pct['Volume'].plot(kind='bar',ax=ax2, rot=75, legend=True)
        #Set-up the X axis to not show so many labels
        n = 6
        ticks = ax2.xaxis.get_ticklocs()
        ticklabels = [l.get_text() for l in ax2.xaxis.get_ticklabels()]
        ax2.xaxis.set_ticks(ticks[::n])
        ax2.xaxis.set_ticklabels(ticklabels[::n])
        plt.gcf().subplots_adjust(bottom=0.15)