Source code for backtrader.indicators.contrib.kalman_filter_line
#!/usr/bin/env python
"""Functional-test indicators migrated to contrib.
Generated from a single functional strategy module to preserve file-local
helper functions and constants without cross-test name collisions.
"""
import math
from .. import Indicator
__all__ = [
"KalmanFilterLine",
"KalmanFilterCandleIndicator",
]
def weighted_price(data, ago=0):
"""Return the weighted price ``(high + low + 2*close) / 4`` for a bar.
Args:
data: Data feed exposing high/low/close lines.
ago: Bar offset (0 is the current bar).
Returns:
The weighted price as a float.
"""
return (float(data.high[ago]) + float(data.low[ago]) + 2.0 * float(data.close[ago])) / 4.0
def indicator_source_price(data, ago=0):
"""Return the indicator source price for a bar or scalar line.
Args:
data: Data feed with OHLC lines, or a single value line.
ago: Bar offset (0 is the current bar).
Returns:
The weighted price when OHLC lines are present, otherwise the line's
value at ``ago``.
"""
if all(hasattr(data, attr) for attr in ("high", "low", "close")):
return weighted_price(data, ago)
return float(data[ago])
[docs]
class KalmanFilterLine(Indicator):
"""Single-series Kalman-style adaptive filter with a velocity term."""
lines = ("value", "color")
params = (
("k", 1.0),
("price_shift_points", 0.0),
)
def __init__(self):
"""Set the minimum period and initialize filter state."""
self.addminperiod(2)
self._initialized = False
self._velocity = 0.0
self.sqrt100 = math.sqrt(float(self.p.k) / 100.0)
self.k100 = float(self.p.k) / 100.0
[docs]
def next(self):
"""Advance the filter one bar and emit the value and color lines."""
source_price = indicator_source_price(self.data, 0)
if not self._initialized:
self.lines.value[0] = source_price + float(self.p.price_shift_points)
self.lines.color[0] = 0
self._velocity = 0.0
self._initialized = True
return
prev_value = float(self.lines.value[-1]) - float(self.p.price_shift_points)
distance = source_price - prev_value
error = prev_value + distance * self.sqrt100
self._velocity += distance * self.k100
filtered = error + self._velocity + float(self.p.price_shift_points)
self.lines.value[0] = filtered
self.lines.color[0] = 1 if self._velocity > 0 else 0
[docs]
class KalmanFilterCandleIndicator(Indicator):
"""Builds Kalman-filtered OHLC candles and a bull/bear color line."""
lines = ("k_open", "k_high", "k_low", "k_close", "color")
params = (
("k", 1.0),
("point", 0.01),
("price_shift", 0),
)
def __init__(self):
"""Construct per-OHLC Kalman filter lines and set the minimum period."""
price_shift_points = float(self.p.point) * float(self.p.price_shift)
self.k_open_line = KalmanFilterLine(
self.data.open, k=self.p.k, price_shift_points=price_shift_points
)
self.k_high_line = KalmanFilterLine(
self.data.high, k=self.p.k, price_shift_points=price_shift_points
)
self.k_low_line = KalmanFilterLine(
self.data.low, k=self.p.k, price_shift_points=price_shift_points
)
self.k_close_line = KalmanFilterLine(
self.data.close, k=self.p.k, price_shift_points=price_shift_points
)
self.addminperiod(3)
[docs]
def next(self):
"""Assemble the filtered candle and classify its bull/bear color."""
o = float(self.k_open_line.value[0])
h = max(float(self.k_high_line.value[0]), o)
low_price = min(float(self.k_low_line.value[0]), o)
c = float(self.k_close_line.value[0])
h = max(h, c)
low_price = min(low_price, c)
self.lines.k_open[0] = o
self.lines.k_high[0] = h
self.lines.k_low[0] = low_price
self.lines.k_close[0] = c
if o < c:
color = 2
elif o > c:
color = 0
else:
color = 1
self.lines.color[0] = color