def populate_indicators(self, dataframe: DataFrame,
metadata: dict) -> DataFrame:
# # Commodity Channel Index: values [Oversold:-100, Overbought:100]
dataframe['buy-cci'] = ta.CCI(dataframe,
timeperiod=self.buy_params['cci-period'])
dataframe['sell-cci'] = ta.CCI(
dataframe, timeperiod=self.sell_params['sell-cci-period'])
# RSI
dataframe['buy-rsi'] = ta.RSI(dataframe,
timeperiod=self.buy_params['rsi-period'])
dataframe['sell-rsi'] = ta.RSI(
dataframe, timeperiod=self.sell_params['sell-rsi-period'])
# KAMA - Kaufman Adaptive Moving Average
dataframe['buy-kama-short'] = ta.KAMA(
dataframe, timeperiod=self.buy_params['kama-short-period'])
dataframe['buy-kama-long'] = ta.KAMA(
dataframe, timeperiod=self.buy_params['kama-long-period'])
dataframe['buy-kama-long-slope'] = (dataframe['buy-kama-long'] /
dataframe['buy-kama-long'].shift())
dataframe['sell-kama-short'] = ta.KAMA(
dataframe, timeperiod=self.sell_params['sell-kama-short-period'])
dataframe['sell-kama-long'] = ta.KAMA(
dataframe, timeperiod=self.sell_params['sell-kama-long-period'])
dataframe['sell-kama-long-slope'] = (
dataframe['sell-kama-long'] / dataframe['sell-kama-long'].shift())
return dataframe
def populate_indicators(self, dataframe: DataFrame,
metadata: dict) -> DataFrame:
"""
Dynamic TA indicators
Used so hyperopt can optimized around the period of various indicators
"""
dataframe['kama-short'] = ta.KAMA(dataframe, timeperiod=5)
dataframe['kama-long'] = ta.KAMA(dataframe, timeperiod=20)
dataframe['cci'] = ta.CCI(dataframe, timeperiod=21)
dataframe['rsi'] = ta.RSI(dataframe, timeperiod=14)
return dataframe
def populate_indicators(self, dataframe: DataFrame, metadata: dict) -> DataFrame:
dataframe['sar'] = ta.SAR(dataframe)
dataframe['rmi'] = RMI(dataframe)
dataframe['kama-3'] = ta.KAMA(dataframe, timeperiod=3)
dataframe['kama-21'] = ta.KAMA(dataframe, timeperiod=21)
macd = ta.MACD(dataframe)
dataframe['macd'] = macd['macd']
dataframe['macdsignal'] = macd['macdsignal']
dataframe['macdhist'] = macd['macdhist']
dataframe['volume_ma'] = dataframe['volume'].rolling(window=24).mean()
return dataframe
def KAMA(ohlcv, kw):
""" :return Kaufman Adaptive Moving Average (kama) """
params = {'timeperiod': 30}
timeperiod = _get_params(kw, params, ['timeperiod'])[0]
result = talib.KAMA(ohlcv, timeperiod)
return {
'kama': result
}
def populate_indicators(dataframe: DataFrame, metadata: dict) -> DataFrame:
"""
Dynamic TA indicators
Used so hyperopt can optimized around the period of various indicators
"""
for kshort in range(kshortStart, (kshortEnd + 1)):
dataframe[f'kama-short({kshort})'] = ta.KAMA(dataframe,
timeperiod=kshort)
for klong in range(klongStart, (klongEnd + 1)):
dataframe[f'kama-long({klong})'] = ta.KAMA(dataframe,
timeperiod=klong)
for klong in range(klongStart, (klongEnd + 1)):
dataframe[f'kama-long-slope({klong})'] = (
dataframe[f'kama-long({klong})'] /
dataframe[f'kama-long({klong})'].shift())
for ccip in range(cciStart, (cciEnd + 1)):
dataframe[f'cci({ccip})'] = ta.CCI(dataframe, timeperiod=ccip)
for rsip in range(rsiStart, (rsiEnd + 1)):
dataframe[f'rsi({rsip})'] = ta.RSI(dataframe, timeperiod=rsip)
"""
Static TA indicators.
Only used when --spaces does not include buy or sell
"""
dataframe['cci'] = ta.CCI(dataframe, timeperiod=cciStatic)
# RSI
dataframe['rsi'] = ta.RSI(dataframe, timeperiod=rsiStatic)
# KAMA - Kaufman Adaptive Moving Average
dataframe['kama-short'] = ta.KAMA(dataframe,
timeperiod=kamaShortStatic)
dataframe['kama-long'] = ta.KAMA(dataframe, timeperiod=kamaLongStatic)
dataframe['kama-long-slope'] = (dataframe['kama-long'] /
dataframe['kama-long'].shift())
return dataframe
def populate_indicators(self, dataframe: DataFrame,
metadata: dict) -> DataFrame:
if not metadata['pair'] in self.custom_trade_info:
self.custom_trade_info[metadata['pair']] = {}
## Base Timeframe / Pair
dataframe['kama'] = ta.KAMA(dataframe, length=233)
# RMI: https://www.tradingview.com/script/kwIt9OgQ-Relative-Momentum-Index/
dataframe['rmi'] = cta.RMI(dataframe, length=24, mom=5)
# Momentum Pinball: https://www.tradingview.com/script/fBpVB1ez-Momentum-Pinball-Indicator/
dataframe['roc-mp'] = ta.ROC(dataframe, timeperiod=1)
dataframe['mp'] = ta.RSI(dataframe['roc-mp'], timeperiod=3)
# MA Streak: https://www.tradingview.com/script/Yq1z7cIv-MA-Streak-Can-Show-When-a-Run-Is-Getting-Long-in-the-Tooth/
dataframe['mastreak'] = cta.mastreak(dataframe, period=4)
# Percent Change Channel: https://www.tradingview.com/script/6wwAWXA1-MA-Streak-Change-Channel/
upper, mid, lower = cta.pcc(dataframe, period=40, mult=3)
dataframe['pcc-lowerband'] = lower
dataframe['pcc-upperband'] = upper
lookup_idxs = dataframe.index.values - (
abs(dataframe['mastreak'].values) + 1)
valid_lookups = lookup_idxs >= 0
dataframe['sbc'] = np.nan
dataframe.loc[valid_lookups, 'sbc'] = dataframe['close'].to_numpy()[
lookup_idxs[valid_lookups].astype(int)]
dataframe['streak-roc'] = 100 * (dataframe['close'] -
dataframe['sbc']) / dataframe['sbc']
# Trends, Peaks and Crosses
dataframe['candle-up'] = np.where(
dataframe['close'] >= dataframe['close'].shift(), 1, 0)
dataframe['candle-up-trend'] = np.where(
dataframe['candle-up'].rolling(5).sum() >= 3, 1, 0)
dataframe['rmi-up'] = np.where(
dataframe['rmi'] >= dataframe['rmi'].shift(), 1, 0)
dataframe['rmi-up-trend'] = np.where(
dataframe['rmi-up'].rolling(5).sum() >= 3, 1, 0)
dataframe['rmi-dn'] = np.where(
dataframe['rmi'] <= dataframe['rmi'].shift(), 1, 0)
dataframe['rmi-dn-count'] = dataframe['rmi-dn'].rolling(8).sum()
dataframe['streak-bo'] = np.where(
dataframe['streak-roc'] < dataframe['pcc-lowerband'], 1, 0)
dataframe['streak-bo-count'] = dataframe['streak-bo'].rolling(8).sum()
# Indicators used only for ROI and Custom Stoploss
ssldown, sslup = cta.SSLChannels_ATR(dataframe, length=21)
dataframe['sroc'] = cta.SROC(dataframe,
roclen=21,
emalen=13,
smooth=21)
dataframe['ssl-dir'] = np.where(sslup > ssldown, 'up', 'down')
# Base pair informative timeframe indicators
informative = self.dp.get_pair_dataframe(pair=metadata['pair'],
timeframe=self.inf_timeframe)
# Get the "average day range" between the 1d high and 1d low to set up guards
informative['1d-high'] = informative['close'].rolling(24).max()
informative['1d-low'] = informative['close'].rolling(24).min()
informative['adr'] = informative['1d-high'] - informative['1d-low']
dataframe = merge_informative_pair(dataframe,
informative,
self.timeframe,
self.inf_timeframe,
ffill=True)
# Other stake specific informative indicators
# e.g if stake is BTC and current coin is XLM (pair: XLM/BTC)
if self.config['stake_currency'] in ('BTC', 'ETH'):
coin, stake = metadata['pair'].split('/')
fiat = self.custom_fiat
coin_fiat = f"{coin}/{fiat}"
stake_fiat = f"{stake}/{fiat}"
# Informative COIN/FIAT e.g. XLM/USD - Base Timeframe
coin_fiat_tf = self.dp.get_pair_dataframe(pair=coin_fiat,
timeframe=self.timeframe)
dataframe[f"{fiat}_rmi"] = cta.RMI(coin_fiat_tf, length=55, mom=5)
# Informative STAKE/FIAT e.g. BTC/USD - Base Timeframe
stake_fiat_tf = self.dp.get_pair_dataframe(
pair=stake_fiat, timeframe=self.timeframe)
dataframe[f"{stake}_rmi"] = cta.RMI(stake_fiat_tf,
length=55,
mom=5)
# Informatives for BTC/STAKE if not in whitelist
else:
pairs = self.dp.current_whitelist()
btc_stake = f"BTC/{self.config['stake_currency']}"
if not btc_stake in pairs:
self.custom_btc_inf = True
# BTC/STAKE - Base Timeframe
btc_stake_tf = self.dp.get_pair_dataframe(
pair=btc_stake, timeframe=self.timeframe)
dataframe['BTC_rmi'] = cta.RMI(btc_stake_tf, length=55, mom=5)
dataframe['BTC_close'] = btc_stake_tf['close']
dataframe['BTC_kama'] = ta.KAMA(btc_stake_tf, length=144)
# Slam some indicators into the trade_info dict so we can dynamic roi and custom stoploss in backtest
if self.dp.runmode.value in ('backtest', 'hyperopt'):
self.custom_trade_info[metadata['pair']]['sroc'] = dataframe[[
'date', 'sroc'
]].copy().set_index('date')
self.custom_trade_info[metadata['pair']]['ssl-dir'] = dataframe[[
'date', 'ssl-dir'
]].copy().set_index('date')
self.custom_trade_info[
metadata['pair']]['rmi-up-trend'] = dataframe[[
'date', 'rmi-up-trend'
]].copy().set_index('date')
self.custom_trade_info[
metadata['pair']]['candle-up-trend'] = dataframe[[
'date', 'candle-up-trend'
]].copy().set_index('date')
return dataframe
],
axis=1)
input_df.columns = map(lambda x: x.lower(), data.columns.tolist())
input_array = dict(zip(input_df.columns, input_df.as_matrix().T))
#####################################
# Overlap
# SMA (for MA's, by default: timeperiod==30, price='close'
log_close_sma = abstract.SMA(input_df, timeperiod=30, price='close')
# EMA
log_close_ema = abstract.EMA(input_df)
# WMA
log_close_wma = abstract.EMA(input_df)
# KAMA - Kaufman Adaptive Moving Average
log_close_kama = abstract.KAMA(input_df)
# MIDPRICE - avg(highest high - lowest low) within the lookback period (default 14)
log_midpr = abstract.MIDPRICE(input_df)
# MIDPOINT - avg(highest close - lowest close) within the lookback period (default 14)
log_midpo = abstract.MIDPOINT(input_df)
# Parabolic Stop and Reverse (SAR) - calculate trailing stop points for long and short positions. SAR = P + A(H-P)
log_sar = abstract.SAR(input_df)
# Bollinger Bands
log_bband_upper, log_bband_middle, log_bband_lower = abstract.BBANDS(
input_array, 20, 3, 3)
log_bband_upper = pd.DataFrame(log_bband_upper, index=input_df.index)
log_bband_middle = pd.DataFrame(log_bband_middle, index=input_df.index)
log_bband_lower = pd.DataFrame(log_bband_lower, index=input_df.index)
# plot
fig1 = plt.figure(figsize=(12, 9))
def populate_indicators(self, dataframe: DataFrame,
metadata: dict) -> DataFrame:
# Populate/update the trade data if there is any, set trades to false if not live/dry
self.custom_trade_info[metadata['pair']] = self.populate_trades(
metadata['pair'])
# Set up primary indicators
dataframe['rmi-slow'] = RMI(dataframe, length=20, mom=5)
dataframe['rmi-fast'] = RMI(dataframe, length=9, mom=3)
dataframe['vidya'] = VIDYA(dataframe)
macd = ta.MACD(dataframe, fastperiod=12, slowperiod=26, signalperiod=9)
dataframe['macd'] = macd['macd']
dataframe['macdsignal'] = macd['macdsignal']
dataframe['macdhist'] = macd['macdhist']
dataframe['kama-fast'] = ta.KAMA(dataframe, timeperiod=5)
dataframe['kama-slow'] = ta.KAMA(dataframe, timeperiod=13)
# Trend Calculations
dataframe['rmi-up'] = np.where(
dataframe['rmi-slow'] >= dataframe['rmi-slow'].shift(), 1, 0)
dataframe['rmi-dn'] = np.where(
dataframe['rmi-slow'] <= dataframe['rmi-slow'].shift(), 1, 0)
dataframe['rmi-up-trend'] = np.where(
dataframe['rmi-up'].rolling(3, min_periods=1).sum() >= 2, 1, 0)
dataframe['rmi-dn-trend'] = np.where(
dataframe['rmi-dn'].rolling(3, min_periods=1).sum() >= 2, 1, 0)
dataframe['vdy-up'] = np.where(
dataframe['vidya'] >= dataframe['vidya'].shift(), 1, 0)
dataframe['vdy-dn'] = np.where(
dataframe['vidya'] <= dataframe['vidya'].shift(), 1, 0)
dataframe['vdy-up-trend'] = np.where(
dataframe['vdy-up'].rolling(3, min_periods=1).sum() >= 2, 1, 0)
dataframe['vdy-dn-trend'] = np.where(
dataframe['vdy-dn'].rolling(3, min_periods=1).sum() >= 2, 1, 0)
# Informative for STAKE/FIAT and COIN/FIAT on default timeframe, only relevant if stake currency is BTC or ETH
if self.config['stake_currency'] in ('BTC', 'ETH'):
coin, stake = metadata['pair'].split('/')
coin_fiat = f"{coin}/{self.custom_fiat}"
stake_fiat = f"{self.config['stake_currency']}/{self.custom_fiat}"
coin_fiat_inf = self.dp.get_pair_dataframe(
pair=coin_fiat, timeframe=self.timeframe)
stake_fiat_inf = self.dp.get_pair_dataframe(
pair=stake_fiat, timeframe=self.timeframe)
dataframe[f"{self.custom_fiat}_rmi-slow"] = RMI(coin_fiat_inf,
length=20,
mom=5)
dataframe[f"{self.config['stake_currency']}_rmi-slow"] = RMI(
stake_fiat_inf, length=20, mom=5)
# Informative for current pair on inf_timeframe
informative = self.dp.get_pair_dataframe(pair=metadata['pair'],
timeframe=self.inf_timeframe)
inf_macd = ta.MACD(informative,
fastperiod=12,
slowperiod=26,
signalperiod=9)
informative['macd'] = inf_macd['macd']
informative['macdsignal'] = inf_macd['macdsignal']
informative['macdhist'] = inf_macd['macdhist']
dataframe = merge_informative_pair(dataframe,
informative,
self.timeframe,
self.inf_timeframe,
ffill=True)
return dataframe