OHLCデータからテクニカル指標を求める

import json

import requests

import datetime

import numpy as np

import matplotlib.dates as mdates

import matplotlib.pyplot as plt

from datetime import datetime, timedelta, timezone

from pandas import DataFrame, Series, to_datetime, concat

from mpl_finance import candlestick_ohlc


if __name__ == '__main__':

    response = json.loads(requests.get("https://api.cryptowat.ch/markets/bitflyer/btcfxjpy/ohlc?periods=60&after=1557172800&before=1557176340").text)

    # 7番目の要素はdocsに記載がないので何なのか不明

    df = pd.DataFrame(response['result']['60'], columns=['date','open','high','low','close','volume','unknown']).dropna().drop(columns=['volume','unknown'])

    # print(df.head(3))

    #         date    open    high     low   close

    # 0  1557432000  664434  665011  663901  664755

    # 1  1557432060  664826  664826  663800  663834

    # 2  1557432120  663834  664117  663723  664069

    # レスポンスのタイムスタンプがUnixTimestamp形式なのでdatetime型に変換し、pandas.DatetimeIndexを設定する

    df['date'] = pd.to_datetime(df['date'], unit='s', utc=True)

    df.set_index('date', inplace=True)

    df.index = df.index.tz_convert('Asia/Tokyo')

    # print(df.head(3))

    #                             open    high     low   close

    # date                                                     

    # 2019-05-10 05:00:00+09:00  664434  665011  663901  664755

    # 2019-05-10 05:01:00+09:00  664826  664826  663800  663834

    # 2019-05-10 05:02:00+09:00  663834  664117  663723  664069

    plt.style.use('ggplot')

    ax = plt.subplot()

    ax.xaxis.set_major_locator(mdates.AutoDateLocator())

    ax.xaxis.set_major_formatter(mdates.DateFormatter('%H:%M', tz=timezone(timedelta(hours=9))))

    # candlestick_ohlc の第二引数に渡すタプルイテレータを生成

    # @see https://github.com/matplotlib/mpl_finance/blob/master/mpl_finance.py

    quotes = zip(mdates.date2num(df.index), df['open'], df['high'], df['low'], df['close'])

    candlestick_ohlc(ax, quotes, width=(1/24/len(df))*0.7, colorup='g', colordown='r')

    plt.show()

def sma(ohlc: DataFrame, period=21) -> Series:

    return Series(

        ohlc['close'].rolling(period).mean(),

        name=f'SMA {period}',

    )

sma(df, 5).plot.line(color='y', legend=True)

sma(df, 10).plot.line(color='c', legend=True)

sma(df, 21).plot.line(color='m', legend=True)

def ema(ohlc: DataFrame, expo=21) -> Series:

    return Series(

        ohlc['close'].ewm(span=expo).mean(),

        name=f'EMA {expo}',

    )

ema(df, 5).plot.line(color='y', legend=True)

ema(df, 10).plot.line(color='c', legend=True)

ema(df, 21).plot.line(color='m', legend=True)

def wma(ohlc: DataFrame, period: int = 9) -> Series:

    # WMA = ( 価格 * n + 価格(1) * n-1 + ... 価格(n-1) * 1) / ( n * (n+1) / 2 )

    denominator = (period * (period + 1)) / 2

    weights = Series(np.arange(1, period + 1)).iloc[::-1]


    return Series(

        ohlc['close'].rolling(period, min_periods=period).apply(lambda x: np.sum(weights * x) / denominator, raw=True),

        name=f'WMA {period}'

    )

wma(df, 5).plot.line(color='y', legend=True)

wma(df, 10).plot.line(color='c', legend=True)

wma(df, 21).plot.line(color='m', legend=True)

def sd(ohlc: DataFrame, period=10, ddof=0) -> float:

    return ohlc['close'].tail(period).std(ddof=ddof)

>>> sd(df, ddof=0)

232.7845570479279

>>> sd(df, ddof=1)

245.37646812828467