/mplfinance

Financial Markets Data Visualization using Matplotlib

Primary LanguagePythonOtherNOASSERTION

mplfinance Checks

mplfinance

matplotlib utilities for the visualization, and visual analysis, of financial data

Installation

pip install --upgrade mplfinance


Contents and Tutorials


The New API

This repository, matplotlib/mplfinance, contains a new matplotlib finance API that makes it easier to create financial plots. It interfaces nicely with Pandas DataFrames.

More importantly, the new API automatically does the extra matplotlib work that the user previously had to do "manually" with the old API. (The old API is still available within this package; see below).

The conventional way to import the new API is as follows:

    import mplfinance as mpf

The most common usage is then to call

    mpf.plot(data)

where data is a Pandas DataFrame object containing Open, High, Low and Close data, with a Pandas DatetimeIndex.

Details on how to call the new API can be found below under Basic Usage, as well as in the jupyter notebooks in the examples folder.

I am very interested to hear from you regarding what you think of the new mplfinance, plus any suggestions you may have for improvement. You can reach me at dgoldfarb.github@gmail.com or, if you prefer, provide feedback or a ask question on our issues page.


Basic Usage

Start with a Pandas DataFrame containing OHLC data. For example,

import pandas as pd
daily = pd.read_csv('examples/data/SP500_NOV2019_Hist.csv',index_col=0,parse_dates=True)
daily.index.name = 'Date'
daily.shape
daily.head(3)
daily.tail(3)
(20, 5)
Open High Low Close Volume
Date
2019-11-01 3050.72 3066.95 3050.72 3066.91 510301237
2019-11-04 3078.96 3085.20 3074.87 3078.27 524848878
2019-11-05 3080.80 3083.95 3072.15 3074.62 585634570

...

Open High Low Close Volume
Date
2019-11-26 3134.85 3142.69 3131.00 3140.52 986041660
2019-11-27 3145.49 3154.26 3143.41 3153.63 421853938
2019-11-29 3147.18 3150.30 3139.34 3140.98 286602291

After importing mplfinance, plotting OHLC data is as simple as calling mpf.plot() on the dataframe

import mplfinance as mpf
mpf.plot(daily)

png


The default plot type, as you can see above, is 'ohlc'. Other plot types can be specified with the keyword argument type, for example, type='candle', type='line', type='renko', or type='pnf'

mpf.plot(daily,type='candle')

png

mpf.plot(daily,type='line')

png

year = pd.read_csv('examples/data/SPY_20110701_20120630_Bollinger.csv',index_col=0,parse_dates=True)
year.index.name = 'Date'
mpf.plot(year,type='renko')

png

mpf.plot(year,type='pnf')

png



We can also plot moving averages with the mav keyword

  • use a scalar for a single moving average
  • use a tuple or list of integers for multiple moving averages
mpf.plot(daily,type='ohlc',mav=4)

png

mpf.plot(daily,type='candle',mav=(3,6,9))

png


We can also display Volume

mpf.plot(daily,type='candle',mav=(3,6,9),volume=True)

png

Notice, in the above chart, there are no gaps along the x-coordinate, even though there are days on which there was no trading. Non-trading days are simply not shown (since there are no prices for those days).

  • However, sometimes people like to see these gaps, so that they can tell, with a quick glance, where the weekends and holidays fall.

  • Non-trading days can be displayed with the show_nontrading keyword.

    • Note that for these purposes non-trading intervals are those that are not represented in the data at all. (There are simply no rows for those dates or datetimes). This is because, when data is retrieved from an exchange or other market data source, that data typically will not include rows for non-trading days (weekends and holidays for example). Thus ...
    • show_nontrading=True will display all dates (all time intervals) between the first time stamp and the last time stamp in the data (regardless of whether rows exist for those dates or datetimes).
    • show_nontrading=False (the default value) will show only dates (or datetimes) that have actual rows in the data. (This means that if there are rows in your DataFrame that exist but contain only NaN values, these rows will still appear on the plot even if show_nontrading=False)
  • For example, in the chart below, you can easily see weekends, as well as a gap at Thursday, November 28th for the U.S. Thanksgiving holiday.

mpf.plot(daily,type='candle',mav=(3,6,9),volume=True,show_nontrading=True)

png


We can also plot intraday data:

intraday = pd.read_csv('examples/data/SP500_NOV2019_IDay.csv',index_col=0,parse_dates=True)
intraday = intraday.drop('Volume',axis=1) # Volume is zero anyway for this intraday data set
intraday.index.name = 'Date'
intraday.shape
intraday.head(3)
intraday.tail(3)
(1563, 4)
Open Close High Low
Date
2019-11-05 09:30:00 3080.80 3080.49 3081.47 3080.30
2019-11-05 09:31:00 3080.33 3079.36 3080.33 3079.15
2019-11-05 09:32:00 3079.43 3079.68 3080.46 3079.43

...

Open Close High Low
Date
2019-11-08 15:57:00 3090.73 3090.70 3091.02 3090.52
2019-11-08 15:58:00 3090.73 3091.04 3091.13 3090.58
2019-11-08 15:59:00 3091.16 3092.91 3092.91 3090.96

The above dataframe contains Open,High,Low,Close data at 1 minute intervals for the S&P 500 stock index for November 5, 6, 7 and 8, 2019. Let's look at the last hour of trading on November 6th, with a 7 minute and 12 minute moving average.

iday = intraday.loc['2019-11-06 15:00':'2019-11-06 16:00',:]
mpf.plot(iday,type='candle',mav=(7,12))

png

The "time-interpretation" of the mav integers depends on the frequency of the data, because the mav integers are the number of data points used in the Moving Average (not the number of days or minutes, etc). Notice above that for intraday data the x-axis automatically displays TIME instead of date. Below we see that if the intraday data spans into two (or more) trading days the x-axis automatically displays BOTH TIME and DATE

iday = intraday.loc['2019-11-05':'2019-11-06',:]
mpf.plot(iday,type='candle')

png


In the plot below, we see what an intraday plot looks like when we display non-trading time periods with show_nontrading=True for intraday data spanning into two or more days.

mpf.plot(iday,type='candle',show_nontrading=True)

png


Below: 4 days of intraday data with show_nontrading=True

mpf.plot(intraday,type='ohlc',show_nontrading=True)

png


Below: the same 4 days of intraday data with show_nontrading defaulted to False.

mpf.plot(intraday,type='line') 

png


Below: Daily data spanning across a year boundary automatically adds the YEAR to the DATE format

df = pd.read_csv('examples/data/yahoofinance-SPY-20080101-20180101.csv',index_col=0,parse_dates=True)
df.shape
df.head(3)
df.tail(3)
(2519, 6)
Open High Low Close Adj Close Volume
Date
2007-12-31 147.100006 147.610001 146.059998 146.210007 118.624741 108126800
2008-01-02 146.529999 146.990005 143.880005 144.929993 117.586205 204935600
2008-01-03 144.910004 145.490005 144.070007 144.860001 117.529449 125133300

...

Open High Low Close Adj Close Volume
Date
2017-12-27 267.380005 267.730011 267.010010 267.320007 267.320007 57751000
2017-12-28 267.890015 267.920013 267.450012 267.869995 267.869995 45116100
2017-12-29 268.529999 268.549988 266.640015 266.859985 266.859985 96007400
mpf.plot(df[700:850],type='bars',volume=True,mav=(20,40))

png

For more examples of using mplfinance, please see the jupyter notebooks in the examples directory.


Some History

My name is Daniel Goldfarb. In November 2019, I became the maintainer of matplotlib/mpl-finance. That module is being deprecated in favor of the current matplotlib/mplfinance. The old mpl-finance consisted of code extracted from the deprecated matplotlib.finance module along with a few examples of usage. It has been mostly un-maintained for the past three years.

It is my intention to archive the matplotlib/mpl-finance repository soon, and direct everyone to matplotlib/mplfinance. The main reason for the rename is to avoid confusion with the hyphen and the underscore: As it was, mpl-finance was installed with the hyphen, but imported with an underscore mpl_finance. Going forward it will be a simple matter of both installing and importing mplfinance.


Old API availability

With this new mplfinance package installed, in addition to the new API, users can still access the old API.
The old API may be removed someday, but for the foreseeable future we will keep it ... at least until we are very confident that users of the old API can accomplish the same things with the new API.

To access the old API with the new mplfinance package installed, change the old import statements

from:

    from mpl_finance import <method>

to:

    from mplfinance.original_flavor import <method>

where <method> indicates the method you want to import, for example:

    from mplfinance.original_flavor import candlestick_ohlc