| Overall Statistics |
|
Total Orders 733 Average Win 4.27% Average Loss -0.35% Compounding Annual Return 337.077% Drawdown 8.100% Expectancy 6.421 Start Equity 1000000 End Equity 1450048.79 Net Profit 45.005% Sharpe Ratio 6.368 Sortino Ratio 10.22 Probabilistic Sharpe Ratio 96.242% Loss Rate 44% Win Rate 56% Profit-Loss Ratio 12.36 Alpha 1.846 Beta 0.06 Annual Standard Deviation 0.292 Annual Variance 0.085 Information Ratio 5.266 Tracking Error 0.304 Treynor Ratio 30.936 Total Fees $996.86 Estimated Strategy Capacity $940000.00 Lowest Capacity Asset HTZ XTDCDCDRI6HX Portfolio Turnover 5.93% |
#region imports
from AlgorithmImports import *
#endregion
import talib
import numpy as np
weights = {}
weights["TASUKIGAP"] = 1.5
weights["SEPARATINGLINES"] = 1
weights["GAPSIDESIDEWHITE"] = .5
weights["HARAMI"] = 1.5
weights["HIKKAKE"] = 1.5
weights["HOMINGPIGEON"] = 1
weights["HAMMER"] = .5
weights["MARUBOZU"] = .5
weights["DARKCLOUDCOVER"] = -1.5
weights["3LINESTRIKE"] = -1.5
weights["ENGULFING"] = -1
weights["SHOOTINGSTAR"] = -.5
def get_candlestick_score(rolling_window, trend):
rolling_window = [x for x in rolling_window if x is not None]
rolling_window.reverse()
size = len(rolling_window)
O = np.array([rolling_window[i].Open for i in range(size)])
H = np.array([rolling_window[i].High for i in range(size)])
L = np.array([rolling_window[i].Low for i in range(size)])
C = np.array([rolling_window[i].Close for i in range(size)])
continuation_patterns = []
continuation_patterns.append(talib.CDLTASUKIGAP(O, H, L, C))
continuation_patterns.append(talib.CDLSEPARATINGLINES(O, H, L, C))
continuation_patterns.append(talib.CDLGAPSIDESIDEWHITE(O,H,L,C))
reversal_to_bull_patterns = []
reversal_to_bull_patterns.append(talib.CDLHARAMI(O,H,L,C))
reversal_to_bull_patterns.append(talib.CDLHIKKAKE(O,H,L,C))
reversal_to_bull_patterns.append(talib.CDLHOMINGPIGEON(O,H,L,C))
reversal_to_bull_patterns.append(talib.CDLHAMMER(O,H,L,C))
reversal_to_bull_patterns.append(talib.CDLMARUBOZU(O,H,L,C))
reversal_to_bear_patterns = []
reversal_to_bear_patterns.append(talib.CDLDARKCLOUDCOVER(O,H,L,C))
reversal_to_bear_patterns.append(talib.CDL3LINESTRIKE(O,H,L,C))
reversal_to_bear_patterns.append(talib.CDLENGULFING(O,H,L,C))
reversal_to_bear_patterns.append(talib.CDLSHOOTINGSTAR(O,H,L,C))
final_weight = 0
#if trend >.6 or trend < .4:
for i in range(len(continuation_patterns)-1):
if continuation_patterns[i].any() > 0:
if i == 0:
# TASUKI GAP
final_weight += weights["TASUKIGAP"] * trend
elif i == 1:
# SEPARATING LINES
final_weight += weights["SEPARATINGLINES"] * trend
elif i == 2:
# GAP SIDE SIDE WHITE
final_weight += weights["GAPSIDESIDEWHITE"] * trend
#elif trend >=.4 and trend <.5:
for i in range(len(reversal_to_bull_patterns)-1):
if reversal_to_bull_patterns[i].any() > 0:
if i == 0:
# HARAMI
final_weight += weights["HARAMI"]
elif i == 1:
# HIKKAKE
final_weight += weights["HIKKAKE"]
elif i == 2:
# HOMING PIGEON
final_weight += weights["HOMINGPIGEON"]
elif i == 3:
# HAMMER
final_weight += weights["HAMMER"]
elif i == 4:
# MARUBOZU
final_weight += weights["MARUBOZU"]
#elif trend <=.6 and trend >=.5:
for i in range(len(reversal_to_bear_patterns)-1):
if reversal_to_bear_patterns[i].any() > 0:
if i == 0:
# DARK CLOUD COVER
final_weight += weights["DARKCLOUDCOVER"]
elif i == 1:
# 3 LINE STRIKE
final_weight += weights["3LINESTRIKE"]
elif i == 2:
# ENGULFING
final_weight += weights["ENGULFING"]
elif i == 3:
# SHOOTING STAR
final_weight += weights["SHOOTINGSTAR"]
return final_weight# region imports
from AlgorithmImports import *
# endregion
import numpy as np
import pandas as pd
from finta import TA
from typing import Union
def supertrend(
data: pd.DataFrame,
period: int = 10,
ATR_multiplier: float = 3.0,
source: pd.Series = None,
) -> pd.DataFrame:
"""SuperTrend indicator, ported from the SuperTrend indicator by
KivancOzbilgic on TradingView.
Parameters
----------
data : pd.DataFrame
The OHLC data.
period : int, optional
The lookback period. The default is 10.
ATR_multiplier : int, optional
The ATR multiplier. The default is 3.0.
source : pd.Series, optional
The source series to use in calculations. If None, hl/2 will be
used. The default is None.
Returns
-------
supertrend_df : pd.DataFrame
A Pandas DataFrame of containing the SuperTrend indicator, with
columns of 'uptrend' and 'downtrend' containing uptrend/downtrend
support/resistance levels, and 'trend', containing -1/1 to indicate
the current implied trend.
References
----------
https://www.tradingview.com/script/r6dAP7yi/
"""
if source is None:
source = (data["High"].values + data["Low"].values) / 2
# Calculate ATR
atr = TA.ATR(data, period)
up = source - (ATR_multiplier * atr)
up_list = [up[0]]
up_times = [data.index[0]]
N_up = 0
dn = source + (ATR_multiplier * atr)
dn_list = [dn[0]]
dn_times = [data.index[0]]
N_dn = 0
trend = 1
trend_list = [trend]
for i in range(1, len(data)):
if trend == 1:
if data["Close"].values[i] > max(up[N_up:i]):
up_list.append(max(up[N_up:i]))
up_times.append(data.index[i])
dn_list.append(np.nan)
dn_times.append(data.index[i])
else:
trend = -1
N_dn = i
dn_list.append(dn[i])
dn_times.append(data.index[i])
up_list.append(np.nan)
up_times.append(data.index[i])
else:
if data["Close"].values[i] < min(dn[N_dn:i]):
dn_list.append(min(dn[N_dn:i]))
dn_times.append(data.index[i])
up_list.append(np.nan)
up_times.append(data.index[i])
else:
trend = 1
N_up = i
up_list.append(up[i])
up_times.append(data.index[i])
dn_list.append(np.nan)
dn_times.append(data.index[i])
trend_list.append(trend)
supertrend_df = pd.DataFrame(
{"uptrend": up_list, "downtrend": dn_list, "trend": trend_list}, index=up_times
)
return supertrend_df
def halftrend(
data: pd.DataFrame, amplitude: int = 2, channel_deviation: float = 2
) -> pd.DataFrame:
"""HalfTrend indicator, ported from the HalfTrend indicator by
Alex Orekhov (everget) on TradingView.
Parameters
----------
data : pd.DataFrame
OHLC price data.
amplitude : int, optional
The lookback window. The default is 2.
channel_deviation : float, optional
The ATR channel deviation factor. The default is 2.
Returns
-------
htdf : TYPE
DESCRIPTION.
References
----------
https://www.tradingview.com/script/U1SJ8ubc-HalfTrend/
"""
# Initialisation
atr2 = TA.ATR(data, 100) / 2
dev = channel_deviation * atr2
high_price = data["High"].rolling(amplitude).max().fillna(0)
low_price = data["Low"].rolling(amplitude).min().fillna(0)
highma = TA.SMA(data, period=amplitude, column="High")
lowma = TA.SMA(data, period=amplitude, column="Low")
trend = np.zeros(len(data))
next_trend = np.zeros(len(data))
max_low_price = np.zeros(len(data))
max_low_price[0] = data["Low"].iloc[0]
min_high_price = np.zeros(len(data))
min_high_price[0] = data["High"].iloc[0]
for i in range(1, len(data)):
if next_trend[i - 1] == 1:
max_low_price[i] = max(low_price.iloc[i - 1], max_low_price[i - 1])
if (
highma.iloc[i] < max_low_price[i]
and data["Close"].iloc[i] < data["Low"].iloc[i - 1]
):
trend[i] = 1
next_trend[i] = 0
min_high_price[i] = high_price.iloc[i]
else:
# assign previous values again
trend[i] = trend[i - 1]
next_trend[i] = next_trend[i - 1]
min_high_price[i] = min_high_price[i - 1]
else:
min_high_price[i] = min(high_price.iloc[i - 1], min_high_price[i - 1])
if (
lowma.iloc[i] > min_high_price[i]
and data["Close"].iloc[i] > data["High"].iloc[i - 1]
):
trend[i] = 0
next_trend[i] = 1
max_low_price[i] = low_price.iloc[i]
else:
# assign previous values again
trend[i] = trend[i - 1]
next_trend[i] = next_trend[i - 1]
max_low_price[i] = max_low_price[i - 1]
up = np.zeros(len(data))
up[0] = max_low_price[0]
down = np.zeros(len(data))
down[0] = min_high_price[0]
atr_high = np.zeros(len(data))
atr_low = np.zeros(len(data))
for i in range(1, len(data)):
if trend[i] == 0:
if trend[i - 1] != 0:
up[i] = down[i - 1]
else:
up[i] = max(max_low_price[i - 1], up[i - 1])
atr_high[i] = up[i] + dev.iloc[i]
atr_low[i] = up[i] - dev.iloc[i]
else:
if trend[i - 1] != 1:
down[i] = up[i - 1]
else:
down[i] = min(min_high_price[i - 1], down[i - 1])
atr_high[i] = down[i] + dev.iloc[i]
atr_low[i] = down[i] - dev.iloc[i]
halftrend = np.where(trend == 0, up, down)
buy = np.where((trend == 0) & (np.roll(trend, 1) == 1), 1, 0)
sell = np.where((trend == 1) & (np.roll(trend, 1) == 0), 1, 0)
# Construct DataFrame
htdf = pd.DataFrame(
data={
"halftrend": halftrend,
"atrHigh": np.nan_to_num(atr_high),
"atrLow": np.nan_to_num(atr_low),
"buy": buy,
"sell": sell,
},
index=data.index,
)
# Clear false leading signals
htdf["buy"].values[:100] = np.zeros(100)
htdf["sell"].values[:100] = np.zeros(100)
# Replace leading zeroes with nan
htdf["atrHigh"] = htdf.atrHigh.replace(to_replace=0, value=float("nan"))
htdf["atrLow"] = htdf.atrLow.replace(to_replace=0, value=float("nan"))
return htdf
def range_filter(
data: pd.DataFrame,
range_qty: float = 2.618,
range_period: int = 14,
smooth_range: bool = True,
smooth_period: int = 27,
av_vals: bool = False,
av_samples: int = 2,
mov_source: str = "body",
filter_type: int = 1,
) -> pd.DataFrame:
"""Price range filter, ported from the Range Filter [DW] indicator by
DonovanWall on TradingView. The indicator was designed to filter out
minor price action for a clearer view of trends.
Parameters
----------
data : pd.DataFrame
The OHLC price data.
range_qty : float, optional
The range size. The default is 2.618.
range_period : int, optional
The range period. The default is 14.
smooth_range : bool, optional
Smooth the price range. The default is True.
smooth_period : int, optional
The smooting period. The default is 27.
av_vals : bool, optional
Average values. The default is False.
av_samples : int, optional
The number of average samples to use. The default is 2.
mov_source : str, optional
The price movement source ('body' or 'wicks'). The default is 'body'.
filter_type : int, optional
The filter type to use in calculations (1 or 2). The default is 1.
Returns
-------
rfi : pd.DataFrame
A dataframe containing the range filter indicator bounds.
References
----------
https://www.tradingview.com/script/lut7sBgG-Range-Filter-DW/
"""
high_val = 0.0
low_val = 0.0
# Get high and low values
if mov_source == "body":
high_val = data["Close"]
low_val = data["Close"]
elif mov_source == "wicks":
high_val = data["High"]
low_val = data["Low"]
# Get filter values
rng = _range_size(
(high_val + low_val) / 2, "AverageChange", range_qty, range_period
)
rfi = _calculate_range_filter(
high_val,
low_val,
rng,
range_period,
filter_type,
smooth_range,
smooth_period,
av_vals,
av_samples,
)
return rfi
def bullish_engulfing(data: pd.DataFrame, detection: str = None):
"""Bullish engulfing pattern detection."""
if detection == "SMA50":
sma50 = sma(data["Close"].values, 50)
down_trend = np.where(data["Close"].values < sma50, True, False)
elif detection == "SMA50/200":
sma50 = sma(data["Close"].values, 50)
sma200 = sma(data["Close"].values, 200)
down_trend = np.where(
(data["Close"].values < sma50) & (data["Close"].values < sma200),
True,
False,
)
else:
down_trend = np.full(len(data), True)
body_len = 14 # ema depth for bodyAvg
body_high = np.maximum(data["Close"].values, data["Open"].values)
body_low = np.minimum(data["Close"].values, data["Open"].values)
body = body_high - body_low
body_avg = ema(body, body_len)
short_body = body < body_avg
long_body = body > body_avg
white_body = data["Open"].values < data["Close"].values
black_body = data["Open"].values > data["Close"].values
inside_bar = [False]
for i in range(1, len(data)):
val = (body_high[i - 1] > body_high[i]) and (body_low[i - 1] < body_low[i])
inside_bar.append(val)
engulfing_bullish = [False]
for i in range(1, len(data)):
condition = (
down_trend[i]
& white_body[i]
& long_body[i]
& black_body[i - 1]
& short_body[i - 1]
& (data["Close"].values[i] >= data["Open"].values[i - 1])
& (data["Open"].values[i] <= data["Close"].values[i - 1])
& (
(data["Close"].values[i] > data["Open"].values[i - 1])
| (data["Open"].values[i] < data["Close"].values[i - 1])
)
)
engulfing_bullish.append(condition)
return engulfing_bullish
def bearish_engulfing(data: pd.DataFrame, detection: str = None):
"""Bearish engulfing pattern detection."""
if detection == "SMA50":
sma50 = sma(data["Close"].values, 50)
up_trend = np.where(data["Close"].values > sma50, True, False)
elif detection == "SMA50/200":
sma50 = sma(data["Close"].values, 50)
sma200 = sma(data["Close"].values, 200)
up_trend = np.where(
(data["Close"].values > sma50) & (data["Close"].values > sma200),
True,
False,
)
else:
up_trend = np.full(len(data), True)
body_len = 14 # ema depth for bodyAvg
body_high = np.maximum(data["Close"].values, data["Open"].values)
body_low = np.minimum(data["Close"].values, data["Open"].values)
body = body_high - body_low
body_avg = ema(body, body_len)
short_body = body < body_avg
long_body = body > body_avg
white_body = data["Open"].values < data["Close"].values
black_body = data["Open"].values > data["Close"].values
inside_bar = [False]
for i in range(1, len(data)):
val = (body_high[i - 1] > body_high[i]) and (body_low[i - 1] < body_low[i])
inside_bar.append(val)
engulfing_bearish = [False]
for i in range(1, len(data)):
condition = (
up_trend[i]
& black_body[i]
& long_body[i]
& white_body[i - 1]
& short_body[i - 1]
& (data["Close"].values[i] <= data["Open"].values[i - 1])
& (data["Open"].values[i] >= data["Close"].values[i - 1])
& (
(data["Close"].values[i] < data["Open"].values[i - 1])
| (data["Open"].values[i] > data["Close"].values[i - 1])
)
)
engulfing_bearish.append(condition)
return engulfing_bearish
def find_swings(data: pd.DataFrame, n: int = 2) -> pd.DataFrame:
"""Locates swings in the inputted data using a moving average gradient
method.
Parameters
----------
data : pd.DataFrame | pd.Series | list | np.array
An OHLC dataframe of price, or an array/list/Series of data from an
indicator (eg. RSI).
n : int, optional
The moving average period. The default is 2.
Returns
-------
swing_df : pd.DataFrame
A dataframe containing the swing levels detected.
pd.Series(hl2, name="hl2"),
"""
# Prepare data
if isinstance(data, pd.DataFrame):
# OHLC data
hl2 = (data["High"].values + data["Low"].values) / 2
swing_data = pd.Series(ema(hl2, n), index=data.index)
low_data = data["Low"].values
high_data = data["High"].values
elif isinstance(data, pd.Series):
# Pandas series data
swing_data = pd.Series(ema(data.fillna(0), n), index=data.index)
low_data = data
high_data = data
else:
# Find swings in alternative data source
data = pd.Series(data)
# Define swing data
swing_data = pd.Series(ema(data, n), index=data.index)
low_data = data
high_data = data
signed_grad = np.sign((swing_data - swing_data.shift(1)).bfill())
swings = (signed_grad != signed_grad.shift(1).bfill()) * -signed_grad
# Calculate swing extrema
lows = []
highs = []
for i, swing in enumerate(swings):
if swing < 0:
# Down swing, find low price
highs.append(0)
lows.append(min(low_data[i - n + 1 : i + 1]))
elif swing > 0:
# Up swing, find high price
highs.append(max(high_data[i - n + 1 : i + 1]))
lows.append(0)
else:
# Price movement
highs.append(0)
lows.append(0)
# Determine last swing
trend = rolling_signal_list(-swings)
swings_list = merge_signals(lows, highs)
last_swing = rolling_signal_list(swings_list)
# Need to return both a last swing low and last swing high list
last_low = rolling_signal_list(lows)
last_high = rolling_signal_list(highs)
swing_df = pd.DataFrame(
data={"Highs": last_high, "Lows": last_low, "Last": last_swing, "Trend": trend},
index=swing_data.index,
)
return swing_df
def classify_swings(swing_df: pd.DataFrame, tol: int = 0) -> pd.DataFrame:
"""Classifies a dataframe of swings (from find_swings) into higher-highs,
lower-highs, higher-lows and lower-lows.
Parameters
----------
swing_df : pd.DataFrame
The dataframe returned by find_swings.
tol : int, optional
The classification tolerance. The default is 0.
Returns
-------
swing_df : pd.DataFrame
A dataframe containing the classified swings.
"""
# Create copy of swing dataframe
swing_df = swing_df.copy()
new_level = np.where(swing_df.Last != swing_df.Last.shift(), 1, 0)
candles_since_last = candles_between_crosses(new_level, initial_count=1)
# Add column 'candles since last swing' CSLS
swing_df["CSLS"] = candles_since_last
# Find strong Support and Resistance zones
swing_df["Support"] = (swing_df.CSLS > tol) & (swing_df.Trend == 1)
swing_df["Resistance"] = (swing_df.CSLS > tol) & (swing_df.Trend == -1)
# Find higher highs and lower lows
swing_df["Strong_lows"] = (
swing_df["Support"] * swing_df["Lows"]
) # Returns high values when there is a strong support
swing_df["Strong_highs"] = (
swing_df["Resistance"] * swing_df["Highs"]
) # Returns high values when there is a strong support
# Remove duplicates to preserve indexes of new levels
swing_df["FSL"] = unroll_signal_list(
swing_df["Strong_lows"]
) # First of new strong lows
swing_df["FSH"] = unroll_signal_list(
swing_df["Strong_highs"]
) # First of new strong highs
# Now compare each non-zero value to the previous non-zero value.
low_change = np.sign(swing_df.FSL) * (
swing_df.FSL
- swing_df.Strong_lows.replace(to_replace=0, method="ffill").shift()
)
high_change = np.sign(swing_df.FSH) * (
swing_df.FSH
- swing_df.Strong_highs.replace(to_replace=0, method="ffill").shift()
)
# the first low_change > 0.0 is not a HL
r_hl = []
first_valid_idx = -1
for i in low_change.index:
v = low_change[i]
if first_valid_idx == -1 and not np.isnan(v) and v != 0.0:
first_valid_idx = i
if first_valid_idx != -1 and i > first_valid_idx and v > 0.0:
hl = True
else:
hl = False
r_hl.append(hl)
# the first high_change < 0.0 is not a LH
r_lh = []
first_valid_idx = -1
for i in high_change.index:
v = high_change[i]
if first_valid_idx == -1 and not np.isnan(v) and v != 0.0:
first_valid_idx = i
if first_valid_idx != -1 and i > first_valid_idx and v < 0.0:
lh = True
else:
lh = False
r_lh.append(lh)
swing_df["LL"] = np.where(low_change < 0, True, False)
# swing_df["HL"] = np.where(low_change > 0, True, False)
swing_df["HL"] = r_hl
swing_df["HH"] = np.where(high_change > 0, True, False)
# swing_df["LH"] = np.where(high_change < 0, True, False)
swing_df["LH"] = r_lh
return swing_df
def detect_divergence(
classified_price_swings: pd.DataFrame,
classified_indicator_swings: pd.DataFrame,
tol: int = 2,
method: int = 0,
) -> pd.DataFrame:
"""Detects divergence between price swings and swings in an indicator.
Parameters
----------
classified_price_swings : pd.DataFrame
The output from classify_swings using OHLC data.
classified_indicator_swings : pd.DataFrame
The output from classify_swings using indicator data.
tol : int, optional
The number of candles which conditions must be met within. The
default is 2.
method : int, optional
The method to use when detecting divergence (0 or 1). The default is 0.
Raises
------
Exception
When an unrecognised method of divergence detection is requested.
Returns
-------
divergence : pd.DataFrame
A dataframe containing divergence signals.
Notes
-----
Options for the method include:
0: use both price and indicator swings to detect divergence (default)
1: use only indicator swings to detect divergence (more responsive)
"""
regular_bullish = []
regular_bearish = []
hidden_bullish = []
hidden_bearish = []
if method == 0:
for i in range(len(classified_price_swings)):
# Look backwards in each
# REGULAR BULLISH DIVERGENCE
if (
sum(classified_price_swings["LL"][i - tol + 1 : i + 1])
+ sum(classified_indicator_swings["HL"][i - tol + 1 : i + 1])
> 1
):
regular_bullish.append(True)
else:
regular_bullish.append(False)
# REGULAR BEARISH DIVERGENCE
if (
sum(classified_price_swings["HH"][i - tol + 1 : i + 1])
+ sum(classified_indicator_swings["LH"][i - tol + 1 : i + 1])
> 1
):
regular_bearish.append(True)
else:
regular_bearish.append(False)
# HIDDEN BULLISH DIVERGENCE
if (
sum(classified_price_swings["HL"][i - tol + 1 : i + 1])
+ sum(classified_indicator_swings["LL"][i - tol + 1 : i + 1])
> 1
):
hidden_bullish.append(True)
else:
hidden_bullish.append(False)
# HIDDEN BEARISH DIVERGENCE
if (
sum(classified_price_swings["LH"][i - tol + 1 : i + 1])
+ sum(classified_indicator_swings["HH"][i - tol + 1 : i + 1])
> 1
):
hidden_bearish.append(True)
else:
hidden_bearish.append(False)
divergence = pd.DataFrame(
data={
"regularBull": unroll_signal_list(regular_bullish),
"regularBear": unroll_signal_list(regular_bearish),
"hiddenBull": unroll_signal_list(hidden_bullish),
"hiddenBear": unroll_signal_list(hidden_bearish),
},
index=classified_price_swings.index,
)
elif method == 1:
# Use indicator swings only to detect divergence
# for i in range(len(classified_price_swings)):
if True:
price_at_indi_lows = (
classified_indicator_swings["FSL"] != 0
) * classified_price_swings["Lows"]
price_at_indi_highs = (
classified_indicator_swings["FSH"] != 0
) * classified_price_swings["Highs"]
# Determine change in price between indicator lows
price_at_indi_lows_change = np.sign(price_at_indi_lows) * (
price_at_indi_lows
- price_at_indi_lows.replace(to_replace=0, method="ffill").shift()
)
price_at_indi_highs_change = np.sign(price_at_indi_highs) * (
price_at_indi_highs
- price_at_indi_highs.replace(to_replace=0, method="ffill").shift()
)
# DETECT DIVERGENCES
regular_bullish = (classified_indicator_swings["HL"]) & (
price_at_indi_lows_change < 0
)
regular_bearish = (classified_indicator_swings["LH"]) & (
price_at_indi_highs_change > 0
)
hidden_bullish = (classified_indicator_swings["LL"]) & (
price_at_indi_lows_change > 0
)
hidden_bearish = (classified_indicator_swings["HH"]) & (
price_at_indi_highs_change < 0
)
divergence = pd.DataFrame(
data={
"regularBull": regular_bullish,
"regularBear": regular_bearish,
"hiddenBull": hidden_bullish,
"hiddenBear": hidden_bearish,
},
index=classified_price_swings.index,
)
else:
raise Exception("Error: unrecognised method of divergence detection.")
return divergence
def autodetect_divergence(
ohlc: pd.DataFrame,
indicator_data: pd.DataFrame,
tolerance: int = 1,
method: int = 0,
) -> pd.DataFrame:
"""A wrapper method to automatically detect divergence from inputted OHLC price
data and indicator data.
Parameters
----------
ohlc : pd.DataFrame
A dataframe of OHLC price data.
indicator_data : pd.DataFrame
dataframe of indicator data.
tolerance : int, optional
A parameter to control the lookback when detecting divergence.
The default is 1.
method : int, optional
The divergence detection method. Set to 0 to use both price and
indicator swings to detect divergence. Set to 1 to use only indicator
swings to detect divergence. The default is 0.
Returns
-------
divergence : pd.DataFrame
A DataFrame containing columns 'regularBull', 'regularBear',
'hiddenBull' and 'hiddenBear'.
See Also
--------
autotrader.indicators.find_swings
autotrader.indicators.classify_swings
autotrader.indicators.detect_divergence
"""
# Price swings
price_swings = find_swings(ohlc)
price_swings_classified = classify_swings(price_swings)
# Indicator swings
indicator_swings = find_swings(indicator_data)
indicator_classified = classify_swings(indicator_swings)
# Detect divergence
divergence = detect_divergence(
price_swings_classified, indicator_classified, tol=tolerance, method=method
)
return divergence
def heikin_ashi(data: pd.DataFrame):
"""Calculates the Heikin-Ashi candlesticks from Japanese candlestick
data.
"""
# Create copy of data to prevent overwriting
working_data = data.copy()
# Calculate Heikin Ashi candlesticks
ha_close = 0.25 * (
working_data["Open"]
+ working_data["Low"]
+ working_data["High"]
+ working_data["Close"]
)
ha_open = 0.5 * (working_data["Open"] + working_data["Close"])
ha_high = np.maximum(
working_data["High"].values,
working_data["Close"].values,
working_data["Open"].values,
)
ha_low = np.minimum(
working_data["Low"].values,
working_data["Close"].values,
working_data["Open"].values,
)
ha_data = pd.DataFrame(
data={"Open": ha_open, "High": ha_high, "Low": ha_low, "Close": ha_close},
index=working_data.index,
)
return ha_data
def ha_candle_run(ha_data: pd.DataFrame):
"""Returns a list for the number of consecutive green and red
Heikin-Ashi candles.
Parameters
----------
ha_data: pd.DataFrame
The Heikin Ashi OHLC data.
See Also
--------
heikin_ashi
"""
green_candle = np.where(ha_data["Close"] - ha_data["Open"] > 0, 1, 0)
red_candle = np.where(ha_data["Close"] - ha_data["Open"] < 0, 1, 0)
green_run = []
red_run = []
green_sum = 0
red_sum = 0
for i in range(len(ha_data)):
if green_candle[i] == 1:
green_sum += 1
else:
green_sum = 0
if red_candle[i] == 1:
red_sum += 1
else:
red_sum = 0
green_run.append(green_sum)
red_run.append(red_sum)
return green_run, red_run
def N_period_high(data: pd.DataFrame, N: int):
"""Returns the N-period high."""
highs = data["High"].rolling(N).max()
return highs
def N_period_low(data: pd.DataFrame, N: int):
"""Returns the N-period low."""
lows = data["Low"].rolling(N).min()
return lows
def crossover(ts1: pd.Series, ts2: pd.Series) -> pd.Series:
"""Locates where two timeseries crossover each other, returning 1 when
list_1 crosses above list_2, and -1 for when list_1 crosses below list_2.
Parameters
----------
ts1 : pd.Series
The first timeseries.
ts2 : pd.Series
The second timeseries.
Returns
-------
crossovers : pd.Series
The crossover series.
"""
signs = np.sign(ts1 - ts2)
crossovers = pd.Series(data=signs * (signs != signs.shift(1)), name="crossovers")
return crossovers
def cross_values(
ts1: Union[list, pd.Series],
ts2: Union[list, pd.Series],
ts_crossover: Union[list, pd.Series] = None,
) -> Union[list, pd.Series]:
"""Returns the approximate value of the point where the two series cross.
Parameters
----------
ts1 : list | pd.Series
The first timeseries..
ts2 : list | pd.Series
The second timeseries..
ts_crossover : list | pd.Series, optional
The crossovers between timeseries 1 and timeseries 2.
Returns
-------
cross_points : list | pd.Series
The values at which crossovers occur.
"""
if ts_crossover is None:
ts_crossover = crossover(ts1, ts2)
last_cross_point = ts1.iloc[0]
cross_points = [last_cross_point]
for i in range(1, len(ts_crossover)):
if ts_crossover.iloc[i] != 0:
i0 = 0
m_a = ts1.iloc[i] - ts1.iloc[i - 1]
m_b = ts2.iloc[i] - ts2.iloc[i - 1]
ix = (ts2.iloc[i - 1] - ts1.iloc[i - 1]) / (m_a - m_b) + i0
cross_point = m_a * (ix - i0) + ts1.iloc[i - 1]
last_cross_point = cross_point
else:
cross_point = last_cross_point
cross_points.append(cross_point)
# Replace nans with 0
cross_points = [0 if x != x else x for x in cross_points]
if isinstance(ts1, pd.Series):
# Convert to Series
cross_points = pd.Series(data=cross_points, index=ts1.index, name="crossval")
return cross_points
def candles_between_crosses(
crosses: Union[list, pd.Series], initial_count: int = 0
) -> Union[list, pd.Series]:
"""Returns a rolling sum of candles since the last cross/signal occurred.
Parameters
----------
crosses : list | pd.Series
The list or Series containing crossover signals.
Returns
-------
counts : TYPE
The rolling count of bars since the last crossover signal.
See Also
---------
indicators.crossover
"""
count = 0
counts = []
for i in range(len(crosses)):
if crosses[i] == 0:
# Change in signal - reset count
count += 1
else:
count = initial_count
counts.append(count)
if isinstance(crosses, pd.Series):
# Convert to Series
counts = pd.Series(data=counts, index=crosses.index, name="counts")
return counts
def rolling_signal_list(signals: Union[list, pd.Series]) -> list:
"""Returns a list which repeats the previous signal, until a new
signal is given.
Parameters
----------
signals : list | pd.Series
A series of signals. Zero values are treated as 'no signal'.
Returns
-------
list
A list of rolled signals.
Examples
--------
>>> rolling_signal_list([0,1,0,0,0,-1,0,0,1,0,0])
[0, 1, 1, 1, 1, -1, -1, -1, 1, 1, 1]
"""
rolling_signals = [0]
last_signal = rolling_signals[0]
if isinstance(signals, list):
for i in range(1, len(signals)):
if signals[i] != 0:
last_signal = signals[i]
rolling_signals.append(last_signal)
else:
for i in range(1, len(signals)):
if signals.iloc[i] != 0:
last_signal = signals.iloc[i]
rolling_signals.append(last_signal)
if isinstance(signals, pd.Series):
rolling_signals = pd.Series(data=rolling_signals, index=signals.index)
return rolling_signals
def unroll_signal_list(signals: Union[list, pd.Series]) -> np.array:
"""Unrolls a rolled signal list.
Parameters
----------
signals : Union[list, pd.Series]
DESCRIPTION.
Returns
-------
unrolled_signals : np.array
The unrolled signal series.
See Also
--------
This function is the inverse of rolling_signal_list.
Examples
--------
>>> unroll_signal_list([0, 1, 1, 1, 1, -1, -1, -1, 1, 1, 1])
array([ 0., 1., 0., 0., 0., -1., 0., 0., 1., 0., 0.])
"""
unrolled_signals = np.zeros(len(signals))
for i in range(1, len(signals)):
if signals[i] != signals[i - 1]:
unrolled_signals[i] = signals[i]
if isinstance(signals, pd.Series):
unrolled_signals = pd.Series(data=unrolled_signals, index=signals.index)
return unrolled_signals
def merge_signals(signal_1: list, signal_2: list) -> list:
"""Returns a single signal list which has merged two signal lists.
Parameters
----------
signal_1 : list
The first signal list.
signal_2 : list
The second signal list.
Returns
-------
merged_signal_list : list
The merged result of the two inputted signal series.
Examples
--------
>>> s1 = [1,0,0,0,1,0]
>>> s2 = [0,0,-1,0,0,-1]
>>> merge_signals(s1, s2)
[1, 0, -1, 0, 1, -1]
"""
merged_signal_list = signal_1.copy()
for i in range(len(signal_1)):
if signal_2[i] != 0:
merged_signal_list[i] = signal_2[i]
return merged_signal_list
def build_grid_price_levels(
grid_origin: float,
grid_space: float,
grid_levels: int,
grid_price_space: float = None,
pip_value: float = 0.0001,
) -> np.array:
"""Generates grid price levels."""
# Calculate grid spacing in price units
if grid_price_space is None:
grid_price_space = grid_space * pip_value
# Generate order_limit_price list
grid_price_levels = np.linspace(
grid_origin - grid_levels * grid_price_space,
grid_origin + grid_levels * grid_price_space,
2 * grid_levels + 1,
)
return grid_price_levels
def build_grid(
grid_origin: float,
grid_space: float,
grid_levels: int,
order_direction: int,
order_type: str = "stop-limit",
grid_price_space: float = None,
pip_value: float = 0.0001,
take_distance: float = None,
stop_distance: float = None,
stop_type: str = None,
) -> dict:
"""Generates a grid of orders.
Parameters
----------
grid_origin : float
The origin of grid, specified as a price.
grid_space : float
The spacing between grid levels, specified as pip distance.
grid_levels : int
The number of grid levels either side of origin.
order_direction : int
The direction of each grid level order (1 for long, -1 for short).
order_type : str, optional
The order type of each grid level order. The default is 'stop-limit'.
grid_price_space : float, optional
The spacing between grid levels, specified as price units distance.
The default is None.
pip_value : float, optional
The instrument-specific pip value. The default is 0.0001.
take_distance : float, optional
The distance (in pips) of each order's take profit. The default is None.
stop_distance : float, optional
The distance (in pips) of each order's stop loss. The default is None.
stop_type : str, optional
The stop loss type. The default is None.
Returns
-------
grid : dict
A dictionary containing all orders on the grid.
"""
# Check if stop_distance was provided without a stop_type
if stop_distance is not None and stop_type is None:
# set stop_type to 'limit' by default
stop_type = "limit"
# Calculate grid spacing in price units
if grid_price_space is None:
grid_price_space = grid_space * pip_value
# Generate order_limit_price list
order_limit_prices = np.linspace(
grid_origin - grid_levels * grid_price_space,
grid_origin + grid_levels * grid_price_space,
2 * grid_levels + 1,
)
# Construct nominal order
nominal_order = {}
nominal_order["order_type"] = order_type
nominal_order["direction"] = order_direction
nominal_order["stop_distance"] = stop_distance
nominal_order["stop_type"] = stop_type
nominal_order["take_distance"] = take_distance
# Build grid
grid = {}
for order, limit_price in enumerate(order_limit_prices):
grid[order] = nominal_order.copy()
grid[order]["order_stop_price"] = order_limit_prices[order]
grid[order]["order_limit_price"] = order_limit_prices[order]
return grid
def merge_grid_orders(grid_1: np.array, grid_2: np.array) -> np.array:
"""Merges grid dictionaries into one and re-labels order numbers so each
order number is unique.
"""
order_offset = len(grid_1)
grid = grid_1.copy()
for order_no in grid_2:
grid[order_no + order_offset] = grid_2[order_no]
return grid
def last_level_crossed(data: pd.DataFrame, base: float) -> list:
"""Returns a list containing the last grid level touched.
The grid levels are determined by the base input variable,
which corresponds to the pip_space x pip_value.
"""
last_level_crossed = np.nan
levels_crossed = []
for i in range(len(data)):
high = data["High"].values[i]
low = data["Low"].values[i]
upper_prices = []
lower_prices = []
for price in [high, low]:
upper_prices.append(base * np.ceil(price / base))
lower_prices.append(base * np.floor(price / base))
if lower_prices[0] != lower_prices[1]:
# Candle has crossed a level
last_level_crossed = lower_prices[0]
levels_crossed.append(last_level_crossed)
return levels_crossed
def build_multiplier_grid(
origin: float,
direction: int,
multiplier: float,
no_levels: int,
precision: int,
spacing: float,
) -> list:
"""Constructs grid levels with a multiplying grid space.
Parameters
----------
origin : float
The origin of grid as price amount.
direction : int
The direction of grid (1 for long, -1 for short).
multiplier : float
The grid space multiplier when price moves away from the origin
opposite to direction.
no_levels : int
The number of levels to calculate either side of the origin.
precision : int
The instrument precision (eg. 4 for most currencies, 2 for JPY).
spacing : float
The spacing of the grid in price units.
"""
levels = [i for i in range(1, no_levels + 1)]
pos_levels = [round(origin + direction * spacing * i, precision) for i in levels]
neg_spaces = [spacing * multiplier ** (i) for i in levels]
neg_levels = []
prev_neg_level = origin
for i in range(len(levels)):
next_neg_level = prev_neg_level - direction * neg_spaces[i]
prev_neg_level = next_neg_level
neg_levels.append(round(next_neg_level, precision))
grid = neg_levels + [origin] + pos_levels
grid.sort()
return grid
def last_level_touched(data: pd.DataFrame, grid: np.array) -> np.array:
"""Calculates the grid levels touched by price data."""
# initialise with nan
last_level_crossed = np.nan
levels_touched = []
for i in range(len(data)):
high = data["High"].values[i]
low = data["Low"].values[i]
upper_prices = []
lower_prices = []
for price in [high, low]:
# Calculate level above
upper_prices.append(
grid[next(x[0] for x in enumerate(grid) if x[1] > price)]
)
# calculate level below
first_level_below_index = next(
x[0] for x in enumerate(grid[::-1]) if x[1] < price
)
lower_prices.append(grid[-(first_level_below_index + 1)])
if lower_prices[0] != lower_prices[1]:
# Candle has crossed a level, since the level below the candle high
# is different to the level below the candle low.
# This essentially means the grid level is between candle low and high.
last_level_crossed = lower_prices[0]
levels_touched.append(last_level_crossed)
return levels_touched
def stoch_rsi(
data: pd.DataFrame,
K_period: int = 3,
D_period: int = 3,
RSI_length: int = 14,
Stochastic_length: int = 14,
):
"""Stochastic RSI indicator."""
rsi1 = TA.RSI(data, period=RSI_length)
stoch = stochastic(rsi1, rsi1, rsi1, Stochastic_length)
K = sma(stoch, K_period)
D = sma(K, D_period)
return K, D
def stochastic(
high: pd.Series, low: pd.Series, close: pd.Series, period: int = 14
) -> pd.Series:
"""Stochastics indicator."""
K = np.zeros(len(high))
for i in range(period, len(high)):
low_val = min(low[i - period + 1 : i + 1])
high_val = max(high[i - period + 1 : i + 1])
K[i] = 100 * (close[i] - low_val) / (high_val - low_val)
return K
def sma(data: pd.DataFrame, period: int = 14) -> list:
"""Smoothed Moving Average."""
sma_list = []
for i in range(len(data)):
average = sum(data[i - period + 1 : i + 1]) / period
sma_list.append(average)
return sma_list
def ema(data: pd.DataFrame, period: int = 14, smoothing: int = 2) -> list:
"""Exponential Moving Average."""
ema = [sum(data[:period]) / period]
for price in data[period:]:
ema.append(
(price * (smoothing / (1 + period)))
+ ema[-1] * (1 - (smoothing / (1 + period)))
)
for i in range(period - 1):
ema.insert(0, np.nan)
return ema
def true_range(data: pd.DataFrame, period: int = 14):
"""True range."""
high_low = data["High"] - data["Low"]
high_close = np.abs(data["High"] - data["Close"].shift())
low_close = np.abs(data["Low"] - data["Close"].shift())
ranges = pd.concat([high_low, high_close, low_close], axis=1)
true_range = np.max(ranges, axis=1)
return true_range
def atr(data: pd.DataFrame, period: int = 14):
"""Average True Range."""
tr = true_range(data, period)
atr = tr.rolling(period).sum() / period
return atr
def create_bricks(data: pd.DataFrame, brick_size: float = 0.002, column: str = "Close"):
"""Creates a dataframe of price-sized bricks.
Parameters
----------
data : pd.DataFrame
The OHLC price data.
brick_size : float, optional
The brick size in price units. The default is 0.0020.
column : str, optional
The column of the OHLC to use. The default is 'Close'.
Returns
-------
bricks : pd.DataFrame
The Open and Close prices of each brick, indexed by brick close time.
"""
brick_open = data[column][0]
opens = [brick_open]
close_times = [data.index[0]]
for i in range(len(data)):
price = data[column][i]
price_diff = price - brick_open
if abs(price_diff) > brick_size:
# New brick(s)
no_new_bricks = abs(int(price_diff / brick_size))
for b in range(no_new_bricks):
brick_close = brick_open + np.sign(price_diff) * brick_size
brick_open = brick_close
opens.append(brick_open)
close_times.append(data.index[i])
bricks = pd.DataFrame(data={"Open": opens, "Close": opens}, index=close_times)
bricks["Close"] = bricks["Close"].shift(-1)
return bricks
def _conditional_ema(x, condition=1, n=14, s=2):
"Conditional sampling EMA functtion"
if type(condition) == int:
condition = condition * np.ones(len(x))
ema = np.zeros(len(x))
for i in range(1, len(x)):
if condition[i]:
ema[i] = (x[i] - ema[i - 1]) * (s / (1 + n)) + ema[i - 1]
else:
ema[i] = ema[i - 1]
return pd.Series(ema, x.index, name=f"{n} period conditional EMA")
def _conditional_sma(x, condition=1, n=14):
"Conditional sampling SMA functtion"
if type(condition) == int:
condition = condition * np.ones(len(x))
# Calculate SMA
sma = x.rolling(n).mean()
# Filter by condition
sma = sma * condition
return sma
def _stdev(x, n):
"Standard deviation function"
sd = np.sqrt(_conditional_sma(x**2, 1, n) - _conditional_sma(x, 1, n) ** 2)
return sd
def _range_size(x, scale="AverageChange", qty=2.618, n=14):
"Range size function"
rng_size = 0
if scale == "AverageChange":
AC = _conditional_ema(abs(x - x.shift(1)), 1, n)
rng_size = qty * AC
elif scale == "ATR":
tr = TA.TR(x)
atr = _conditional_ema(tr, 1, n)
rng_size = qty * atr
elif scale == "StandardDeviation":
sd = _stdev(x, n)
rng_size = qty * sd
return rng_size
def _calculate_range_filter(h, idx, rng, n, rng_type, smooth, sn, av_rf, av_n):
"""Two type range filter function."""
smoothed_range = _conditional_ema(rng, 1, sn)
r = smoothed_range if smooth else rng
r_filt = (h + idx) / 2
if rng_type == 1:
for i in range(1, len(h)):
if h[i] - r[i] > r_filt[i - 1]:
r_filt[i] = h[i] - r[i]
elif idx[i] + r[i] < r_filt[i - 1]:
r_filt[i] = idx[i] + r[i]
else:
r_filt[i] = r_filt[i - 1]
elif rng_type == 2:
for i in range(1, len(h)):
if h[i] >= r_filt[i - 1] + r[i]:
r_filt[i] = (
r_filt[i - 1] + np.floor(abs(h[i] - r_filt[i - 1]) / r[i]) * r[i]
)
elif idx[i] <= r_filt[i - 1] - r[i]:
r_filt[i] = (
r_filt[i - 1] - np.floor(abs(idx[i] - r_filt[i - 1]) / r[i]) * r[i]
)
else:
r_filt[i] = r_filt[i - 1]
# Define nominal values
r_filt1 = r_filt.copy()
hi_band1 = r_filt1 + r
lo_band1 = r_filt1 - r
# Calculate indicator for averaged filter changes
r_filt2 = _conditional_ema(r_filt1, r_filt1 != r_filt1.shift(1), av_n)
hi_band2 = _conditional_ema(hi_band1, r_filt1 != r_filt1.shift(1), av_n)
lo_band2 = _conditional_ema(lo_band1, r_filt1 != r_filt1.shift(1), av_n)
# Assign indicator
rng_filt = r_filt2 if av_rf else r_filt1
hi_band = hi_band2 if av_rf else hi_band1
lo_band = lo_band2 if av_rf else lo_band1
# Construct output
rfi = pd.DataFrame(
data={"upper": hi_band, "lower": lo_band, "rf": rng_filt}, index=rng_filt.index
)
# Classify filter direction
rfi["fdir"] = np.sign(rfi.rf - rfi.rf.shift(1)).fillna(0)
return rfi
def chandelier_exit(
data: pd.DataFrame, length: int = 22, mult: float = 3.0, use_close: bool = False
):
# ohlc4 = (data["Open"] + data["High"] + data["Low"] + data["Close"]) / 4
atr = mult * TA.ATR(data, length)
high_field = "Close" if use_close else "High"
low_field = "Close" if use_close else "Low"
longstop = data[high_field].rolling(length).max() - atr
shortstop = data[low_field].rolling(length).min() + atr
direction = np.where(data["Close"] > shortstop, 1, -1)
chandelier_df = pd.concat(
{
"longstop": longstop,
"shortstop": shortstop,
},
axis=1,
)
chandelier_df["direction"] = direction
chandelier_df["signal"] = np.where(
chandelier_df["direction"] != chandelier_df["direction"].shift(),
chandelier_df["direction"],
0,
)
return chandelier_df# region imports
from AlgorithmImports import *
from trendCalculator import get_trend
import numpy as np
# endregion
class This2024q2wip2(QCAlgorithm):
def Initialize(self):
self.SetStartDate(2024, 1, 1)
self.SetEndDate(2024, 4, 1)
self.SetCash(1000000)
# universe selection
self.rebalanceTime = self.Time
self.AddUniverse(self.CoarseFilter, self.FineFilter)
self.UniverseSettings.Resolution = Resolution.Hour
self.plotting = False
self.logging = False
self.final_universe_size = 1000 #int(self.get_parameter("final_universe_size")) # the number of stocks to hold in the universe
self.portfolio_weight_bias = 300 #int(self.get_parameter("portfolio_weight_bias")) # the bias for the portfolio weight
# indicator parameters
self.sma_slow_length = 29 #self.get_parameter("slow_length") # the length in days of the slow moving average
self.sma_fast_length = 12 #self.get_parameter("fast_length") # the length in days of the fast moving average
self.trend_sma_length = 100 # the length in days of the trend moving average
self.ATR_multiplier = 1 # the multiplier for the ATR for the slow SMA bands
self.price_rolling_length = 35 * 7 #int(self.get_parameter("price_rolling")) * 7 # the length in hours of price data to hold
self.aroon_threshold = 95 #int(self.get_parameter("aroon_threshold"))
# Trend parameters
self.trend_threshold = .08 #float(self.get_parameter("trend_threshold")) # minimum trend either direction to enter trade
self.trend_length = self.price_rolling_length # length of rolling price data to check for trend in hours
self.order = 10 # for trend calculation purposes
self.k = 2 # for trend calculation purposes
self.leading_direction_band_multiply = 1.25 #float(self.get_parameter("leading_band")) # multiplier for leading direction band
# Risk management parameters
self.price_smoothing = 2
self.rebuy_period = [7*7, 10*7] # minimum and maximum period in which to re-enter after trailing stop in hours
self.trailing_stop = .25 #float(self.get_parameter("trailing_stop")) # percent value of trailing stop
self.short_trailing_stop = .25
self.days_breaking_multiplier = 3 #int(self.get_parameter("days_breaking"))
self.days_breaking_before_enter = self.days_breaking_multiplier * 7 # used to determine how many days to wait before entering trade after sma break
self.days_breaking_trend = 30 * 7#int(self.get_parameter("days_breaking_trend")) * 7
self.rsi_high_threshold = 45 # rsi value to stop out long
self.rsi_low_threshold = 55 # rsi value to stop out short
self.rsi_rolling_length = 10 * 7#int(self.get_parameter("rsi_rolling"))# length of rsi rolling window
self.rsi_order = 5 # for rsi trend calculation
self.trend_distance = 0 #float(self.get_parameter("trend_distance")) # distance from trend sma to be considered far enough to enter trade
self.adx_threshold = 50 #int(self.get_parameter("adx_threshold")) # minimum adx value to enter trade
self.sell_rsi_margin = 15 #int(self.get_parameter("sell_rsi_margin")) # margin for rsi to sell out
self.rsi_trend_threshold = 1 #float(self.get_parameter("rsi_threshold")) # rsi trend threshold to sell out\
self.rsi_selling_toggle = True
# dictionaries to hold indicator data etc for each symbol
self.activeStocks = set()
self.sma_slow = {}
self.sma_slow_consolidators = {}
self.sma_fast = {}
self.sma_fast_consolidators = {}
self.sma_trend = {}
self.sma_trend_consolidators = {}
self.price_rolling = {}
self.sell_prices = {}
self.cover_prices = {}
self.ATRS = {}
self.atr_consolidators = {}
self.days_breakings = {}
self.days_breakings_trend = {}
self.above_below_sma = {}
self.peak_while_long = {}
self.peak_while_short = {}
self.aroons = {}
self.aroon_consolidators = {}
self.RSIS = {}
self.rsi_consolidators = {}
self.RSI_last_location = {}
self.rsi_rolling = {}
self.ADX = {}
self.adx_consolidators = {}
# portfolio management
self.trade_list = []
self.max_position_size = .25#float(self.get_parameter("max_position_size")) # max position size as a proportion of total portfolio value
self.bought_dates = {}
self.days_doing_nothing = 35#int(self.get_parameter("days_doing_nothing"))
def CoarseFilter(self, coarse):
# Rebalancing monthly
if self.Time <= self.rebalanceTime:
return self.Universe.Unchanged
# rebalance universe every month
self.rebalanceTime = self.Time + timedelta(days=90)
# filter top 50 by volume
sortedByDollarVolume = sorted(coarse, key=lambda x: x.DollarVolume, reverse=True)
return [x.Symbol for x in sortedByDollarVolume if x.HasFundamentalData][:1000]
def FineFilter(self, fine):
# filter by top volume (again (redundant for now))
sortedByMarketCap = sorted(fine, key=lambda x: x.DollarVolume, reverse=True)
return [x.Symbol for x in sortedByMarketCap if x.price > 10 and x.MarketCap > 2000000000][:self.final_universe_size]
def OnSecuritiesChanged(self, changes):
# region removed securities
for x in changes.RemovedSecurities:
self.Liquidate(x.Symbol)
if x.Symbol in self.activeStocks:
self.activeStocks.remove(x.Symbol)
# initialize indicators and data structures for new securities
for x in changes.AddedSecurities:
self.activeStocks.add(x.Symbol)
self.sma_slow[x.Symbol] = DoubleExponentialMovingAverage(self.sma_slow_length)
self.sma_slow_consolidators[x.Symbol] = TradeBarConsolidator(timedelta(days=1))
self.register_indicator(x.Symbol, self.sma_slow[x.Symbol], self.sma_slow_consolidators[x.Symbol])
self.WarmUpIndicator(x.Symbol, self.sma_slow[x.Symbol])
self.sma_fast[x.Symbol] = DoubleExponentialMovingAverage(self.sma_fast_length)
self.sma_fast_consolidators[x.Symbol] = TradeBarConsolidator(timedelta(days=1))
self.register_indicator(x.Symbol, self.sma_fast[x.Symbol], self.sma_fast_consolidators[x.Symbol])
self.WarmUpIndicator(x.Symbol, self.sma_fast[x.Symbol])
self.aroons[x.Symbol] = AroonOscillator(14, 14)
self.aroon_consolidators[x.Symbol] = TradeBarConsolidator(timedelta(days=1))
self.register_indicator(x.Symbol, self.aroons[x.Symbol], self.aroon_consolidators[x.Symbol])
self.WarmUpIndicator(x.Symbol, self.aroons[x.Symbol])
self.ATRS[x.Symbol] = AverageTrueRange(14)
self.atr_consolidators[x.Symbol] = TradeBarConsolidator(timedelta(days=1))
self.register_indicator(x.Symbol, self.ATRS[x.Symbol], self.atr_consolidators[x.Symbol])
self.WarmUpIndicator(x.Symbol, self.ATRS[x.Symbol])
self.RSIS[x.Symbol] = RelativeStrengthIndex(14)
self.rsi_consolidators[x.Symbol] = TradeBarConsolidator(timedelta(days=1))
self.register_indicator(x.Symbol, self.RSIS[x.Symbol], self.rsi_consolidators[x.Symbol])
self.WarmUpIndicator(x.Symbol, self.RSIS[x.Symbol])
self.ADX[x.Symbol] = AverageDirectionalIndex(14)
self.adx_consolidators[x.Symbol] = TradeBarConsolidator(timedelta(days=1))
self.register_indicator(x.Symbol, self.ADX[x.Symbol], self.adx_consolidators[x.Symbol])
self.WarmUpIndicator(x.Symbol, self.ADX[x.Symbol])
self.sma_trend[x.Symbol] = ExponentialMovingAverage(self.trend_sma_length)
self.sma_trend_consolidators[x.Symbol] = TradeBarConsolidator(timedelta(days=1))
self.register_indicator(x.Symbol, self.sma_trend[x.Symbol], self.sma_trend_consolidators[x.Symbol])
self.WarmUpIndicator(x.Symbol, self.sma_trend[x.Symbol])
self.price_rolling[x.Symbol] = RollingWindow[float](self.price_rolling_length)
self.above_below_sma[x.Symbol] = 0
self.peak_while_long[x.Symbol] = None
self.peak_while_short[x.Symbol] = None
self.sell_prices[x.Symbol] = None
self.cover_prices[x.Symbol] = None
self.days_breakings[x.Symbol] = 0
self.days_breakings_trend[x.Symbol] = 0
self.RSI_last_location[x.Symbol] = None
self.rsi_rolling[x.Symbol] = RollingWindow[float](self.rsi_rolling_length)
history = self.History[TradeBar](x.Symbol, (self.trend_sma_length) + (self.days_breaking_trend), Resolution.Daily)
# populate indicators and rolling windows with historical data
count = 0
sma_trend_initialize = ExponentialMovingAverage(self.trend_sma_length)
for bar in history:
if count <= self.trend_sma_length:
sma_trend_initialize.Update(bar.EndTime, bar.Close)
else:
self.price_rolling[x.Symbol].Add(bar.Close)
self.rsi_rolling[x.Symbol].Add(bar.Close)
sma_trend_initialize.Update(bar.EndTime, bar.Close)
if bar.Close > sma_trend_initialize.Current.Value:
ATR_multiplier_bottom = self.leading_direction_band_multiply * self.ATR_multiplier
ATR_multiplier_top = self.ATR_multiplier
if self.days_breakings_trend[x.Symbol] <= 0:
self.days_breakings_trend[x.Symbol] = 1
else:
self.days_breakings_trend[x.Symbol] += 7
else:
ATR_multiplier_top = self.leading_direction_band_multiply * self.ATR_multiplier
ATR_multiplier_bottom = self.ATR_multiplier
if self.days_breakings_trend[x.Symbol] >= 0:
self.days_breakings_trend[x.Symbol] = -1
else:
self.days_breakings_trend[x.Symbol] -= 7
sma_slow_low = self.sma_slow[x.Symbol].Current.Value - ATR_multiplier_bottom * self.ATRS[x.Symbol].Current.Value
sma_slow_high = self.sma_slow[x.Symbol].Current.Value + ATR_multiplier_top * self.ATRS[x.Symbol].Current.Value
if self.sma_fast[x.Symbol].Current.Value > sma_slow_high:
relation = 1
elif self.sma_fast[x.Symbol].Current.Value < sma_slow_low:
relation = -1
else:
relation = 0
if relation == 1 and self.above_below_sma[x.Symbol] == 0 and self.days_breakings[x.Symbol] <=0 and bar.close > self.sma_trend[x.Symbol].Current.Value:
self.days_breakings[x.Symbol] = 1
#self.Log("in history, setting days breaking to 1")
elif relation == -1 and self.above_below_sma[x.Symbol] == 0 and self.days_breakings[x.Symbol] >=0 and bar.close < self.sma_trend[x.Symbol].Current.Value:
self.days_breakings[x.Symbol] = -1
#self.Log("in history, setting days breaking to -1")
if self.days_breakings[x.Symbol] != 0:
if self.days_breakings[x.Symbol] > 0:
if self.sma_fast[x.Symbol].Current.Value > self.sma_slow[x.Symbol].Current.Value:
#if self.RSIS[x.Symbol].Current.Value > 50:
if count%60 == 0:
self.days_breakings[x.Symbol] += 1
else:
self.days_breakings[x.Symbol] = 0
elif self.days_breakings[x.Symbol] < 0:
if self.sma_fast[x.Symbol].Current.Value < self.sma_slow[x.Symbol].Current.Value:
#if self.RSIS[x.Symbol].Current.Value < 50:
if count%60 == 0:
self.days_breakings[x.Symbol] -= 1
else:
self.days_breakings[x.Symbol] = 0
self.above_below_sma[x.Symbol] = relation
count += 1
self.Log("days_breakigns_trend after initialization: " + str(self.days_breakings_trend[x.Symbol]) + " for symbol: " + str(x.Symbol))
def OnData(self, data: Slice):
smoothed_price = None
for symbol in self.activeStocks:
if not data.ContainsKey(symbol) or data[symbol] is None:
continue
self.price_rolling[symbol].Add(data[symbol].close)
self.rsi_rolling[symbol].Add(data[symbol].close)
if not self.sma_fast[symbol].IsReady or not self.sma_slow[symbol].IsReady:
self.Log("smas not ready")
return
# Adjust sma slow bands based on trend
if data[symbol].close > self.sma_trend[symbol].Current.Value:
self.ATR_multiplier_bottom = self.leading_direction_band_multiply * self.ATR_multiplier
self.ATR_multiplier_top = self.ATR_multiplier
if self.days_breakings_trend[symbol] <= 0:
self.days_breakings_trend[symbol] = 1
else:
self.days_breakings_trend[symbol] += 1
else:
self.ATR_multiplier_top = self.leading_direction_band_multiply * self.ATR_multiplier
self.ATR_multiplier_bottom = self.ATR_multiplier
if self.days_breakings_trend[symbol] >= 0:
self.days_breakings_trend[symbol] = -1
else:
self.days_breakings_trend[symbol] -= 1
if self.plotting:
self.Plot("days_breakings_trend", "days_breakings_trend", self.days_breakings_trend[symbol])
self.Plot("aroon", "aroon_up", self.aroons[symbol].AroonUp.Current.Value)
self.Plot("aroon", "aroon_down", self.aroons[symbol].AroonDown.Current.Value)
self.Plot("rsi", "rsi", self.RSIS[symbol].Current.Value)
self.Plot("adx for " + str(symbol), "adx", self.ADX[symbol].Current.Value)
if self.RSI_last_location[symbol] != None:
self.Plot("rsi_locale", "rsi_last_location", self.RSI_last_location[symbol])
# check if fast sma is above or below slow sma
if self.sma_fast[symbol].Current.Value > self.sma_slow[symbol].Current.Value + self.ATR_multiplier_top * self.ATRS[symbol].Current.Value:
relation = 1
elif self.sma_fast[symbol].Current.Value < self.sma_slow[symbol].Current.Value - self.ATR_multiplier_bottom * self.ATRS[symbol].Current.Value:
relation = -1
else:
relation = 0
if self.plotting:
self.Plot("distance_from_trend", "distance_from_trend", data[symbol].close/self.sma_trend[symbol].Current.Value)
self.Plot("distance_from_trend", "trend_distance_down", 1-self.trend_distance)
self.Plot("distance_from_trend", "trend_distance_up", 1+self.trend_distance)
self.Log("far_enough_above_trend_sma: " + str(far_enough_above_trend_sma) + " far_enough_below_trend_sma: " + str(far_enough_below_trend_sma) + " for symbol: " + str(symbol))
# If recently crossed above slow sma band, initiate count for how many days it has been above
if relation == 1 and self.above_below_sma[symbol] == 0 and self.days_breakings[symbol] <=0 and not self.Portfolio[symbol].IsLong:
self.Liquidate(symbol)
self.trade_list = [x for x in self.trade_list if x[0] != symbol]
if self.logging:
self.Log("Liquidating as relation is 1 and crossed above initiating count for symbol: " + str(symbol))
self.Log("sma_fast: " + str(self.sma_fast[symbol].Current.Value) + " sma_slow top band: " + str(self.sma_slow[symbol].Current.Value + self.ATR_multiplier_top * self.ATRS[symbol].Current.Value))
self.Log("days above trend: " + str(self.days_breakings_trend[symbol]))
self.days_breakings[symbol] = 1
self.sell_prices[symbol] = None
self.cover_prices[symbol] = None
self.peak_while_long[symbol] = None
self.peak_while_short[symbol] = None
# If recently crossed below slow sma band, initiate count for how many days it has been below
elif relation == -1 and self.above_below_sma[symbol] == 0 and self.days_breakings[symbol] >=0 and not self.Portfolio[symbol].IsShort:
self.Liquidate(symbol)
self.trade_list = [x for x in self.trade_list if x[0] != symbol]
if self.logging:
self.Log("Liquidating as relation is -1 and crossed below initiating count for symbol: " + str(symbol))
#self.Log("* setting days_breakings to -1")
self.Log("sma_fast: " + str(self.sma_fast[symbol].Current.Value) + " sma_slow bottom band: " + str(self.sma_slow[symbol].Current.Value - self.ATR_multiplier_bottom * self.ATRS[symbol].Current.Value))
self.Log("days below trend: " + str(self.days_breakings_trend[symbol]))
self.days_breakings[symbol] = -1
self.cover_prices[symbol] = None
self.sell_prices[symbol] = None
self.peak_while_long[symbol] = None
self.peak_while_short[symbol] = None
# if no band crosses...
else:
# if long, check for trailing stop
if self.Portfolio[symbol].IsLong:
# smooth price data to avoid short whipsaws
smoothed_price = []
for j in range(self.price_smoothing):
smoothed_price.append(self.price_rolling[symbol][j])
smoothed_price = sum(smoothed_price)/self.price_smoothing
# update the peak while long for trailing stop logic and check if trailing stop is hit
self.peak_while_long[symbol] = max(self.peak_while_long[symbol], data[symbol].close) if self.peak_while_long[symbol] != None else data[symbol].close
if smoothed_price/self.peak_while_long[symbol] < (1-self.trailing_stop):
self.Liquidate(symbol)
self.trade_list = [x for x in self.trade_list if x[0] != symbol]
if self.plotting:
self.Plot("price", "trail", self.peak_while_long[symbol] * (1-self.trailing_stop))
if self.logging:
self.Log("liquidating long trailing stop, smoothed_price: " + str(smoothed_price) + " peak_while_long: " + str(self.peak_while_long[symbol]) + " for symbol: " + str(symbol))
self.sell_prices[symbol] = [0, data[symbol].close]
position = 0
self.peak_while_long[symbol] = None
# check for overbought
if self.RSIS[symbol].Current.Value < self.rsi_high_threshold and self.RSI_last_location[symbol] == "1":
if self.logging:
self.Log("liquidating long overbought, rsi: " + str(self.RSIS[symbol].Current.Value) + " for symbol: " + str(symbol))
self.Liquidate(symbol)
self.trade_list = [x for x in self.trade_list if x[0] != symbol]
self.peak_while_long[symbol] = None
if self.RSIS[symbol].Current.Value < self.rsi_high_threshold and self.RSIS[symbol].Current.Value > self.rsi_low_threshold:
self.RSI_last_location[symbol] = "0"
elif self.RSIS[symbol].Current.Value >= self.rsi_high_threshold:
self.RSI_last_location[symbol] = "1"
elif self.RSIS[symbol].Current.Value <= self.rsi_low_threshold:
self.RSI_last_location[symbol] = "-1"
# check if rsi trend hints at reversal sell out
# consolidate rsi_rolling to groups of 60
rsi_arr = [x for x in self.rsi_rolling[symbol]]
rsi_trend = get_trend(self, rsi_arr, self.rsi_order, self.k)[1]
if self.plotting:
self.Plot("trend", "rsi_trend", rsi_trend)
self.Log("rsi_trend: " + str(rsi_trend) + " for symbol: " + str(symbol))
if rsi_trend < -self.rsi_trend_threshold and self.RSIS[symbol].Current.Value < 55:
if self.rsi_selling_toggle:
self.Liquidate(symbol)
if self.logging:
self.Log("liquidating long rsi trend, rsi_trend: " + str(rsi_trend) + " for symbol: " + str(symbol))
self.trade_list = [x for x in self.trade_list if x[0] != symbol]
self.peak_while_long[symbol] = None
self.sell_prices[symbol] = None
# check for rsi dip below 55
if self.RSIS[symbol].Current.Value < 50 - self.sell_rsi_margin:
if self.rsi_selling_toggle:
if self.logging:
self.Log("liquidating long rsi dip, rsi: " + str(self.RSIS[symbol].Current.Value) + " for symbol: " + str(symbol))
rsi_arr = [x for x in self.rsi_rolling[symbol]]
self.Log("rsi_arr: " + str(rsi_arr))
self.Liquidate(symbol)
self.trade_list = [x for x in self.trade_list if x[0] != symbol]
self.peak_while_long[symbol] = None
self.sell_prices[symbol] = None
# if short...
elif self.Portfolio[symbol].IsShort:
# smooth price data to avoid short whipsaws
smoothed_price = []
for j in range(self.price_smoothing):
smoothed_price.append(self.price_rolling[symbol][j])
smoothed_price = sum(smoothed_price)/self.price_smoothing
# update the peak while short for trailing stop logic and check if trailing stop is hit
self.peak_while_short[symbol] = min(self.peak_while_short[symbol], data[symbol].close) if self.peak_while_short[symbol] != None else data[symbol].close
if smoothed_price/self.peak_while_short[symbol] > (1+ self.short_trailing_stop):
self.Liquidate(symbol)
self.trade_list = [x for x in self.trade_list if x[0] != symbol]
if self.plotting:
self.Plot("price", "trail", self.peak_while_short[symbol] * (1+self.short_trailing_stop))
if self.logging:
self.Log("liquidating short trailing stop, smoothed_price: " + str(smoothed_price) + " peak_while_short: " + str(self.peak_while_short[symbol]) + " for symbol: " + str(symbol))
self.cover_prices[symbol] = [0, data[symbol].close]
position = 0
self.peak_while_short[symbol] = None
# check for oversold
if self.RSIS[symbol].Current.Value > self.rsi_low_threshold and self.RSI_last_location[symbol] == "-1":
if self.rsi_selling_toggle:
self.Liquidate(symbol)
if self.logging:
self.Log("liquidating short oversold, rsi: " + str(self.RSIS[symbol].Current.Value) + " for symbol: " + str(symbol))
self.trade_list = [x for x in self.trade_list if x[0] != symbol]
self.peak_while_long[symbol] = None
if self.RSIS[symbol].Current.Value < self.rsi_high_threshold and self.RSIS[symbol].Current.Value > self.rsi_low_threshold:
self.RSI_last_location[symbol] = "0"
elif self.RSIS[symbol].Current.Value >= self.rsi_high_threshold:
self.RSI_last_location[symbol] = "1"
elif self.RSIS[symbol].Current.Value <= self.rsi_low_threshold:
self.RSI_last_location[symbol] = "-1"
# check if rsi trend hints at reversal sell out
# consolidate rsi_rolling to groups of 60
rsi_arr = [x for x in self.rsi_rolling[symbol]]
rsi_trend = get_trend(self, rsi_arr, self.rsi_order, self.k)[1]
if self.plotting:
self.Plot("trend", "rsi_trend", rsi_trend)
if rsi_trend > self.rsi_trend_threshold and self.RSIS[symbol].Current.Value > 45:
if self.rsi_selling_toggle:
if self.logging:
self.Log("liquidating short rsi trend, rsi_trend: " + str(rsi_trend) + " for symbol: " + str(symbol))
self.Liquidate(symbol)
self.trade_list = [x for x in self.trade_list if x[0] != symbol]
self.peak_while_short[symbol] = None
self.cover_prices[symbol] = None
# check for rsi spike above 45
if self.RSIS[symbol].Current.Value > 50 + self.sell_rsi_margin:
if self.rsi_selling_toggle:
if self.logging:
self.Log("liquidating short rsi spike, rsi: " + str(self.RSIS[symbol].Current.Value) + " for symbol: " + str(symbol))
self.Liquidate(symbol)
self.trade_list = [x for x in self.trade_list if x[0] != symbol]
self.peak_while_short[symbol] = None
self.cover_prices[symbol] = None
# if band crosses, check if enough days have passed to enter trade
if self.days_breakings[symbol] != 0:
if self.days_breakings[symbol] > 0:
if self.sma_fast[symbol].Current.Value > self.sma_slow[symbol].Current.Value:
if self.RSIS[symbol].Current.Value > 50:
self.days_breakings[symbol] += 1
#self.Log("*incrementing days_breakings")
if self.days_breakings[symbol] >= self.days_breaking_before_enter and self.days_breakings_trend[symbol] >= self.days_breaking_trend:
price_arr = [x for x in self.price_rolling[symbol]]
trend = get_trend(self, price_arr, self.order, self.k)[1]/data[symbol].close
# consolidate rsi_rolling to groups of 60
rsi_arr = [x for x in self.rsi_rolling[symbol]]
rsi_trend = get_trend(self, rsi_arr, self.rsi_order, self.k)[1]
if self.plotting:
self.Plot("trend", "trend", trend)
self.Plot("trend", "rsi_trend", rsi_trend)
#if self.logging:
#self.Log("enough days have passed, going long " + str(symbol))
if trend > self.trend_threshold and self.aroons[symbol].AroonUp.Current.Value > self.aroon_threshold and self.aroons[symbol].AroonUp.Current.Value > self.aroons[symbol].AroonDown.Current.Value and self.ADX[symbol].Current.Value >= self.adx_threshold and self.days_breakings_trend[symbol] >= self.days_breaking_trend:
self.days_breakings[symbol] = 0
self.Log("*buy, aroonup: " + str(self.aroons[symbol].AroonUp.Current.Value) + " aroondown: " + str(self.aroons[symbol].AroonDown.Current.Value) + " adx: " + str(self.ADX[symbol].Current.Value) + "days breaking trend: " + str(self.days_breakings_trend[symbol]) + " for: " + str(symbol))
amount_above = data[symbol].close - self.sma_slow[symbol].Current.Value
self.trade_list.append([symbol, trend/data[symbol].price, amount_above/data[symbol].price, self.aroons[symbol].AroonUp.Current.Value, self.ADX[symbol].Current.Value, 1])
self.peak_while_long[symbol] = data[symbol].close
else:
self.days_breakings[symbol] = 0
#self.Log("* reset to 0 as below slow sma")
elif self.days_breakings[symbol] < 0:
if self.sma_fast[symbol].Current.Value < self.sma_slow[symbol].Current.Value:
if self.RSIS[symbol].Current.Value < 50:
self.days_breakings[symbol] -= 1
#self.Log("not shorting yet, current days breaking: " + str(self.days_breakings[symbol]) + " needs to reach: " + str(-self.days_breaking_before_enter))
if self.days_breakings[symbol] <= -self.days_breaking_before_enter and self.days_breakings_trend[symbol] <= -self.days_breaking_trend:
price_arr = [x for x in self.price_rolling[symbol]]
trend = get_trend(self, price_arr, self.order, self.k)[1]/data[symbol].close
# consolidate rsi_rolling to groups of 60
rsi_arr = [x for x in self.rsi_rolling[symbol]]
rsi_trend = get_trend(self, rsi_arr, self.rsi_order, self.k)[1]
if self.plotting:
self.Plot("trend", "trend", trend)
self.Plot("trend", "rsi_trend", rsi_trend)
#if self.logging:
#self.Log("enough days have passed, going short " + str(symbol))
if trend < -self.trend_threshold and self.aroons[symbol].AroonDown.Current.Value > self.aroon_threshold and self.aroons[symbol].AroonDown.Current.Value > self.aroons[symbol].AroonUp.Current.Value and self.ADX[symbol].Current.Value >= self.adx_threshold and self.days_breakings_trend[symbol] <= -self.days_breaking_trend:
self.days_breakings[symbol] = 0
self.Log("*short, aroondown: " + str(self.aroons[symbol].AroonDown.Current.Value) + " aroonup: " + str(self.aroons[symbol].AroonUp.Current.Value) + " adx: " + str(self.ADX[symbol].Current.Value) + "days breaking trend: " + str(self.days_breakings_trend[symbol]) + " for: " + str(symbol))
amount_below = data[symbol].close - self.sma_slow[symbol].Current.Value
self.trade_list.append([symbol, trend/data[symbol].price, amount_below/data[symbol].price, self.aroons[symbol].AroonDown.Current.Value, self.ADX[symbol].Current.Value, -1])
self.peak_while_short[symbol] = data[symbol].close
else:
self.days_breakings[symbol] = 0
# if stopped out, check if price is above slow sma and fast sma and reenter trade
if self.sell_prices[symbol] is not None and self.sell_prices[symbol][0] > self.rebuy_period[0] and self.sell_prices[symbol][0] < self.rebuy_period[1] and not self.Portfolio[symbol].IsShort and not self.Portfolio[symbol].IsLong:
if data[symbol].close > self.sell_prices[symbol][1] and data[symbol].close > self.sma_fast[symbol].Current.Value and data[symbol].close > self.sma_slow[symbol].Current.Value + self.ATR_multiplier_top * self.ATRS[symbol].Current.Value:
# buy back in\
price_arr = [x for x in self.price_rolling[symbol]]
trend = get_trend(self, price_arr, self.order, self.k)[1]/data[symbol].close
if self.plotting:
self.Plot("trend", "trend", trend)
if trend > self.trend_threshold and self.aroons[symbol].AroonUp.Current.Value > self.aroon_threshold and self.aroons[symbol].AroonUp.Current.Value > self.aroons[symbol].AroonDown.Current.Value and self.ADX[symbol].Current.Value >= self.adx_threshold and self.RSIS[symbol].Current.Value >= 55:
amount_above = data[symbol].close - self.sma_slow[symbol].Current.Value
self.trade_list.append([symbol, trend/data[symbol].price, amount_above/data[symbol].price, self.aroons[symbol].AroonUp.Current.Value, self.ADX[symbol].Current.Value, 1])
self.peak_while_long[symbol] = data[symbol].close
self.sell_prices[symbol] = None
# if stopped out, check if price is below slow sma and fast sma and reenter trade
if self.cover_prices[symbol] is not None and self.cover_prices[symbol][0] > self.rebuy_period[0] and self.cover_prices[symbol][0] < self.rebuy_period[1] and not self.Portfolio[symbol].IsShort and not self.Portfolio[symbol].IsLong:
if data[symbol].close < self.cover_prices[symbol][1] and data[symbol].close < self.sma_fast[symbol].Current.Value and data[symbol].close < self.sma_slow[symbol].Current.Value - self.ATR_multiplier_bottom * self.ATRS[symbol].Current.Value:
# buy back in
price_arr = [x for x in self.price_rolling[symbol]]
trend = get_trend(self, price_arr, self.order, self.k)[1]/data[symbol].close
if self.plotting:
self.Plot("trend", "trend", trend)
if trend < -self.trend_threshold and self.aroons[symbol].AroonDown.Current.Value > self.aroon_threshold and self.aroons[symbol].AroonDown.Current.Value > self.aroons[symbol].AroonUp.Current.Value and self.ADX[symbol].Current.Value >= self.adx_threshold and self.RSIS[symbol].Current.Value <= 45:
amount_below = data[symbol].close - self.sma_slow[symbol].Current.Value
self.trade_list.append([symbol, trend/data[symbol].price, amount_below/data[symbol].price, self.aroons[symbol].AroonDown.Current.Value, self.ADX[symbol].current.Value, -1])
if self.logging:
self.Log("reentering after stopping out short " + str(symbol))
self.peak_while_short[symbol] = data[symbol].close
self.cover_prices[symbol] = None
self.above_below_sma[symbol] = relation
if self.sell_prices[symbol] != None:
self.sell_prices[symbol][0] += 1
if self.cover_prices[symbol] != None:
self.cover_prices[symbol][0] += 1
if self.plotting:
# plot if long, short, or neutral:
self.Plot("quantity", "quantity", self.Portfolio[symbol].Quantity)
self.Plot("above/below sma", "above/below sma", self.above_below_sma[symbol])
self.Plot("sma", "Short", self.sma_fast[symbol].Current.Value)
self.Plot("sma", "Long", self.sma_slow[symbol].Current.Value)
#self.Plot("sma", "day_long sma", self.day_sma_slow.Current.Value)
self.Plot("sma", "slow_lower_band", self.sma_slow[symbol].Current.Value - self.ATR_multiplier_bottom * self.ATRS[symbol].Current.Value)
self.Plot("sma", "slow_upper_band", self.sma_slow[symbol].Current.Value + self.ATR_multiplier_top * self.ATRS[symbol].Current.Value)
self.Plot("price", "trend", self.sma_trend[symbol].Current.Value)
self.Plot("price", "price", data[symbol].Close)
self.Plot("atr", "atr", self.ATRS[symbol].Current.Value)
self.Plot("days_breaking", "days_breaking",self.days_breakings[symbol])
if self.peak_while_short[symbol] != None:
self.Plot("price", "peak_while_short", self.peak_while_short[symbol])
# add up score for uninvested trades
total_score = 0
for trade in self.trade_list:
total_score += self.portfolio_weight_bias + abs(trade[2] * trade[1] * trade[3] * trade[4])
trade.append(abs(trade[2] * trade[1] * trade[3] * trade[4]))
# sort trade list by score
self.trade_list = sorted(self.trade_list, key=lambda x: x[-1], reverse=True)
# invest in trades
for i in range(len(self.trade_list)):
trade = self.trade_list[i]
proportion = (self.portfolio_weight_bias + (trade[2] * trade[1])) / total_score
margin_left = self.Portfolio.MarginRemaining - 5000
holdings = self.Portfolio.TotalAbsoluteHoldingsCost
cash_left = self.Portfolio.Cash
total_potential = holdings + margin_left + cash_left
if self.logging:
self.Log("proportion: " + str(proportion) + "holdings " + str(holdings) + " margin_left: " + str(margin_left) + " cash_left: " + str(cash_left) + " total_potential: " + str(total_potential) + " for symbol: " + str(trade[0]))
if self.plotting:
self.Plot("buy/sell", "holdings", holdings)
self.Plot("buy/sell", "cash", cash_left)
self.Plot("buy/sell", "margin", margin_left)
self.Plot("buy/sell", "potential", total_potential)
quantity = proportion * (max(margin_left, 0) + max(cash_left, 0)) / self.price_rolling[trade[0]][0]
if self.logging:
self.Log("first quantity: " + str(quantity) + " for symbol: " + str(trade[0]))
# if currently invested and amount we want to invest more makes total position size too large, adjust
if self.Portfolio[trade[0]].Invested:
if abs(((quantity * self.price_rolling[trade[0]][0]) + abs(self.Portfolio[symbol].quantity) * self.price_rolling[trade[0][0]])/total_potential) > self.max_position_size:
#self.Log("quantity value too high, adjusting")
quantity = (self.max_position_size * total_potential) / self.price_rolling[trade[0]][0] - abs(self.Portfolio[symbol].quantity)
if abs((quantity * self.price_rolling[trade[0]][0])/total_potential) > self.max_position_size:
#self.Log("quantity value too high, adjusting")
quantity = (self.max_position_size * total_potential) / self.price_rolling[trade[0]][0]
quantity = quantity
if self.logging:
self.Log("final quantity: " + str(quantity) + " for symbol: " + str(trade[0]))
if self.plotting:
self.Plot("buy/sell", "quantity_value", quantity * self.price_rolling[trade[0]][0])
if trade[5] == -1:
order_properties = OrderProperties()
order_properties.time_in_force = TimeInForce.GOOD_TIL_DATE(self.Time + timedelta(days=2))
self.limit_order(trade[0], -quantity/2, self.price_rolling[trade[0]][0] *.95, order_properties=order_properties)
self.bought_dates[trade[0]] = self.Time
elif trade[5] == 1:
order_properties = OrderProperties()
order_properties.time_in_force = TimeInForce.GOOD_TIL_DATE(self.Time + timedelta(days=2))
self.limit_order(trade[0], quantity, self.price_rolling[trade[0]][0] * 1.05, order_properties=order_properties)
self.bought_dates[trade[0]] = self.Time
self.trade_list = []
keys = list(self.Portfolio.keys())
sortedByProfit = sorted(keys, key=lambda x: self.Portfolio[x].UnrealizedProfitPercent, reverse=True)
for i in range(len(sortedByProfit)):
symbol = sortedByProfit[i]
if self.Portfolio[symbol].UnrealizedProfit != 0.0:
if self.Portfolio[symbol].UnrealizedProfitPercent > .05 or self.Portfolio[symbol].UnrealizedProfitPercent < -.05:
if not self.plotting:
self.Plot("profit", symbol, self.Portfolio[symbol].UnrealizedProfitPercent)
margin_left = self.Portfolio.MarginRemaining - 5000
holdings = self.Portfolio.TotalAbsoluteHoldingsCost
cash_left = self.Portfolio.Cash
if cash_left < 0: cash_left = 0
total_potential = margin_left + cash_left + holdings
# if less than 15% of total portfolio, and profit is more than 5%, increase position size
if abs(self.Portfolio[symbol].UnrealizedProfitPercent) > .05 and self.Portfolio[symbol].holdings_cost/total_potential < self.max_position_size * 1.5:
quantity = (self.max_position_size * 1.5 * total_potential - self.Portfolio[symbol].holdings_cost) / self.price_rolling[symbol][0]
if quantity * self.price_rolling[symbol][0] > margin_left + cash_left:
quantity = (margin_left + cash_left) / self.price_rolling[symbol][0]
if quantity < 0:
continue
if self.Portfolio[symbol].IsShort:
quantity = -quantity/2
order_properties = OrderProperties()
order_properties.time_in_force = TimeInForce.GOOD_TIL_DATE(self.Time + timedelta(days=2))
self.limit_order(symbol, quantity, self.price_rolling[symbol][0] * 1.05, order_properties=order_properties)
if self.logging:
self.Log("increasing position size for symbol: " + str(symbol))
# if 5 days have passed and profit is less than 1$, liquidate
if self.bought_dates[symbol] != None:
if self.Time - self.bought_dates[symbol] > timedelta(days=self.days_doing_nothing) and self.Portfolio[symbol].UnrealizedProfitPercent < .01:
self.Liquidate(symbol)
self.bought_dates[symbol] = None
self.peak_while_long[symbol] = None
self.peak_while_short[symbol] = None
self.sell_prices[symbol] = None
self.cover_prices[symbol] = None
self.trade_list = [x for x in self.trade_list if x[0] != symbol]
if self.logging:
self.Log("liquidating due to low profit for symbol: " + str(symbol))
#region imports
from AlgorithmImports import *
import numpy as np
import matplotlib.pyplot as plt
import pandas as pd
from scipy.signal import argrelextrema
from collections import deque
from matplotlib.lines import Line2D
from datetime import timedelta
#endregion
'''
Much of this code is sourced at the following link: https://raposa.trade/blog/higher-highs-lower-lows-and-calculating-price-trends-in-python/
'''
def getHigherLows(data: np.array, order, K):
'''
Finds consecutive higher lows in price pattern.
Must not be exceeded within the number of periods indicated by the width
parameter for the value to be confirmed.
K determines how many consecutive lows need to be higher.
'''
# Get lows
low_idx = argrelextrema(data, np.less, order=order)[0]
lows = data[low_idx]
# Ensure consecutive lows are higher than previous lows
extrsma = []
ex_deque = deque(maxlen=K)
for i, idx in enumerate(low_idx):
if i == 0:
ex_deque.append(idx)
continue
if lows[i] < lows[i-1]:
ex_deque.clear()
ex_deque.append(idx)
if len(ex_deque) == K:
extrsma.append(ex_deque.copy())
return extrsma
def getLowerHighs(data: np.array, order=5, K=2):
'''
Finds consecutive lower highs in price pattern.
Must not be exceeded within the number of periods indicated by the width
parameter for the value to be confirmed.
K determines how many consecutive highs need to be lower.
'''
# Get highs
high_idx = argrelextrema(data, np.greater, order=order)[0]
highs = data[high_idx]
# Ensure consecutive highs are lower than previous highs
extrsma = []
ex_deque = deque(maxlen=K)
for i, idx in enumerate(high_idx):
if i == 0:
ex_deque.append(idx)
continue
if highs[i] > highs[i-1]:
ex_deque.clear()
ex_deque.append(idx)
if len(ex_deque) == K:
extrsma.append(ex_deque.copy())
return extrsma
def getHigherHighs(data: np.array, order, K):
'''
Finds consecutive higher highs in price pattern.
Must not be exceeded within the number of periods indicated by the width
parameter for the value to be confirmed.
K determines how many consecutive highs need to be higher.
'''
# Get highs
high_idx = argrelextrema(data, np.greater, order = order)[0]
highs = data[high_idx]
# Ensure consecutive highs are higher than previous highs
extrsma = []
ex_deque = deque(maxlen=K)
for i, idx in enumerate(high_idx):
if i == 0:
ex_deque.append(idx)
continue
if highs[i] < highs[i-1]:
ex_deque.clear()
ex_deque.append(idx)
if len(ex_deque) == K:
extrsma.append(ex_deque.copy())
return extrsma
def getLowerLows(data: np.array, order, K):
'''
Finds consecutive lower lows in price pattern.
Must not be exceeded within the number of periods indicated by the width
parameter for the value to be confirmed.
K determines how many consecutive lows need to be lower.
'''
# Get lows
low_idx = argrelextrema(data, np.less, order=order)[0]
lows = data[low_idx]
# Ensure consecutive lows are lower than previous lows
extrsma = []
ex_deque = deque(maxlen=K)
for i, idx in enumerate(low_idx):
if i == 0:
ex_deque.append(idx)
continue
if lows[i] > lows[i-1]:
ex_deque.clear()
ex_deque.append(idx)
if len(ex_deque) == K:
extrsma.append(ex_deque.copy())
return extrsma
def get_trend(QCalgo, close_data, order, K):
'''
Get the trend of the stock
'''
#close_data = [x for x in close_data]
close_data.reverse()
# data set to dataframe empty
data = pd.DataFrame()
data['Close'] = close_data
close = data['Close'].values
hh = getHigherHighs(close, order, K)
hl = getHigherLows(close, order, K)
ll = getLowerLows(close, order, K)
lh = getLowerHighs(close, order, K)
# format for tuples inside patterns: [type, location first price, location second price, first price, second price]
patterns = []
for pattern in hh:
# append a tuple with date and "hh"
patterns.append(('hh', pattern[0], pattern[1], close[pattern[0]], close[pattern[1]]))
for pattern in hl:
patterns.append(('hl', pattern[0], pattern[1], close[pattern[0]], close[pattern[1]]))
for pattern in ll:
patterns.append(('ll', pattern[0], pattern[1], close[pattern[0]], close[pattern[1]]))
for pattern in lh:
patterns.append(('lh', pattern[0], pattern[1], close[pattern[0]], close[pattern[1]]))
# sort by the second date
patterns.sort(key=lambda x: x[2], reverse=True)
trend = 0
recent_movements_length = 15
recent_movements = patterns[:recent_movements_length]
recent_swing_up = 0
recent_swing_down = 0
for x in recent_movements:
if x[0] == 'hh' or x[0] == 'hl':
recent_swing_up += (x[4] - x[3])
print("hh or hl, adding: ", (x[4] - x[3]))
else:
recent_swing_down += (x[4] - x[3])
print("ll or lh, adding: ", (x[4] - x[3]))
recent_swing = recent_swing_up + recent_swing_down
#QCalgo.Log("recent_swing: " + str(recent_swing))
total_movements_length = 45
total_movements = patterns[:total_movements_length]
total_swing_up = 0
total_swing_down = 0
for x in total_movements:
if x[0] == 'hh' or x[0] == 'hl':
total_swing_up += (x[4] - x[3])
else:
total_swing_down += (x[4] - x[3])
total_swing = total_swing_up + total_swing_down
#QCalgo.Log("total_swing: " + str(total_swing))
return (trend, recent_swing, total_swing)