Skip to content

New mplfinance package (to replace mpl-finance sometime in 2020).

License

Notifications You must be signed in to change notification settings

biffbifford/mplfinance

 
 

Repository files navigation

Build Status

mplfinance

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


Installation

   pip install mplfinance


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' or type='line'

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

png

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

png



We can also plot moving averages with the mav keyword

  • use a scaler 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 gaps along the x-coordinate corresponding to days on which there was no trading.

  • Many people like to see these gaps so that they can tell, with a quick glance, where the weekends and holidays fall.
  • For example, in the above chart you can see a gap at Thursday, November 28th for the U.S. Thanksgiving holiday.
  • Gaps along the x-axis can be eliminated with the no_xgaps keyword
mpf.plot(daily,type='candle',mav=(3,6,9),volume=True,no_xgaps=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 intervervals 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 number of data points used in the Moving Average. Notice above that for intraday data the x-axis automatically displays TIME instead of date. Below we see that if the intraday data spans two (or more) trading days then two things happen:

  • The x-axis displays BOTH TIME and DATE
  • no-xgaps defaults to True FOR INTRADAY DATA INVOLVING TWO OR MORE TRADING DAYS
iday = intraday.loc['2019-11-05':'2019-11-06',:]
mpf.plot(iday,type='candle')

png


In the plot below, we see what would happen if no_xgaps did NOT default to True for intraday data involving two or more days.

mpf.plot(iday,type='candle',no_xgaps=False)

png


Below: 4 days of intraday data with no_xgaps=False

mpf.plot(intraday,type='ohlc',no_xgaps=False)  # 4 day of intraday with no_xgaps=False

png


Below: 4 days of intraday data with no_xgaps defaulted to True for intraday data spanning more than one day.

mpf.plot(intraday,type='line')  # intraday spanning more than one day defaults to no_xgaps=True

png


Below: Daily data spanning more than a year 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 146.995005 143.880005 144.929993 117.586205 204935600
2008-01-03 144.910004 145.495005 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,no_xgaps=True,mav=(20,40))

png

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


COMING SOON:

  • customize appearance of plot (colors, date format, etc)
  • show trading signals on plot
  • technical studies, such as:
    • Trading Envelope, Bollinger Bands
    • MACD
  • custom studies and/or additional data on plot
    • Ability to plot specified additional columns from DataFrame either within the main ohlc plot, or only the lower axis where volume may be displayed.
  • save plot to file

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.

The new API

At present (Dec 2019) this repository, matplotlib/mplfinance, contains an initial 'alpha', version of the new API for people to play with and provide feedback or pull requests for enhancements.

My own take on the old mpl-finance API is that the methods were too low-level, and too cumbersome to use. The new API in this current package automatically does the extra matplotlib work that the caller previously had to do "manually, on their own" with the old API.

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

    import mplfinance as mpf

The most common usage is to then call mpf.plot(data) where data is a Pandas DataFrame object containing Open, High, Low and Close data, with a Pandas DatetimeIndex.


For details on how to call the new API, see the jupyter notebook(s) in the examples folder:


I am very interested to hear from you regarding how you were using the old mpl-finance (if you were), what you think of the new mplfinance, plus any suggestions you may have for improvement. You can reach me at [email protected]


old API availability

With this new mplfinance package installed, in addition to the new API, users can still access the old API (at least for the next several months) by changing their import statments
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

About

New mplfinance package (to replace mpl-finance sometime in 2020).

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%