Indicators: calculate, compose, and read values¶
Use self.ta inside Strategy.init() for built-in indicators, self.ind for a custom function, and indicators.compute for a calculation outside a backtest. Browse the indicator catalog for every formula, its parameters, inputs, and outputs.
compute · catalog · describe · IndicatorHandle · IndicatorResult · is_cross · is_crossover · is_crossunder
Single and multiple outputs¶
from openbacktesting import Backtest, Strategy
from openbacktesting.test import GOOG
class IndicatorExample(Strategy):
def init(self):
self.ema = self.ta.ema(self.close, length=20)
self.macd = self.ta.macd(fast=12, slow=26, signal=9)
self.smooth = self.ta.sma(self.macd.histogram, length=3)
def next(self):
if self.ema.valid[-1] and self.smooth.valid[-1]:
if self.close[-1] > self.ema[-1] and self.smooth[-1] > 0:
if not self.position:
self.buy(qty=1)
elif self.position:
self.position.close()
result = Backtest(GOOG, IndicatorExample).run()
print(result["Return [%]"])
An EMA returns an IndicatorHandle. A multi-output formula such as MACD returns an IndicatorResult; select .macd, .signal, or .histogram before passing a channel to another indicator. Bracket access such as result["histogram"] also works. Built-in calculations are registered only in init().
Bind inputs and read outputs¶
| Argument or operation | Meaning |
|---|---|
self.ta.ema(length=20) |
Use that indicator's default market column (Close here). |
self.ta.ema(self.hl2, length=20) |
Use a same-run single-channel handle. A column name such as "Close" is also accepted. |
inputs={"high": "High", "low": "Low", "close": "Close"} |
Explicit mapping of formula inputs to data columns or same-run channels; do not combine with positional source arguments. |
Formula keywords such as length, fast, slow |
See each indicator's parameter table for defaults, allowed values, and meaning. |
backend=None |
Use the formula's default implementation. Only documented alternatives are accepted. |
required_outputs=None |
Use default outputs for readiness diagnostics; an explicit channel list changes readiness, not whether next() runs. |
handle[-1], handle[-2] |
Current and previous visible values; check history length. |
handle.valid[-1] |
Whether this channel has a consumable current value. Check it before trading. |
handle.ready[-1] |
Minimum-history/readiness diagnostic. It does not skip callbacks. |
handle.s, handle.df, numpy.asarray(handle) |
Independent snapshots of the visible prefix. |
During next(), saved handles and their slices follow the current bar. A saved array/Series or arithmetic result does not. False and zero may be valid signals; NaN is not a usable signal. Do not use bool(NaN) as a validity check. For a custom calculation, see Strategy.ind.
Crosses and confirmed events¶
self.ta.is_crossover(a, b) returns a boolean at the current bar; is_crossunder reverses direction and is_cross accepts either direction. Both operands need two valid visible observations. These predicates can be called in next().
self.ta.crossover(a, b) instead registers a whole signal history in init(). Use it when composing signals with valuewhen or barssince. A pivot becomes available when its right-hand confirmation window finishes; it does not backfill a tradable signal at the earlier extreme. Check the pivot and session-level pages before using those results.
Calculate outside a strategy¶
from openbacktesting import indicators
from openbacktesting.test import GOOG
ema = indicators.compute("ta.ema", GOOG, parameters={"length": 20})
print(ema.s.tail())
description = indicators.describe("ta.ema")
catalog = indicators.catalog()
compute returns an IndicatorHandle covering the full input. Multi-output offline results select channels by name, for example macd["histogram"]. parameters supplies formula options; inputs maps input names to frame columns. Ordinary offline calls can omit cache and cursor. describe returns one formula's specification and catalog lists the available definitions.
compute¶
indicators.compute(
indicator,
frame,
*,
parameters=None,
backend=None,
inputs=None,
cache=None,
cursor=None,
required_outputs=None,
)
Compute reviewed indicators in bulk on explicit market data and return read-only result handles.
catalog¶
Return a tuple of all reviewed definitions, including deferred entries that cannot be called.
describe¶
Look up an indicator definition by stable ID without computing it.
IndicatorHandle¶
Time is the final dimension; all derived views apply cursors and selectors only when read.
Attributes¶
IndicatorHandle.astype¶
Convert the current prefix to a NumPy array snapshot with dtype; do not retain a dynamic cursor.
IndicatorHandle.copy¶
Return a writable NumPy copy of the current prefix that no longer follows the cursor.
IndicatorHandle.first_ready¶
Return the earliest ready position for required outputs, or total length if never ready; does not control callbacks.
IndicatorHandle.ready_at¶
Query readiness of required outputs and minimum-history conditions; does not control strategy callbacks.
IndicatorResult¶
Read-only attribute container for multiple outputs; channel views are created only during initialization.
Attributes¶
is_cross¶
Check whether either-direction crossing occurs on the last currently visible bar.
is_crossover¶
Check whether the current left value is strictly above the right, with the preceding left value at or below the right.
is_crossunder¶
Check whether the current left value is strictly below the right, with the preceding left value at or above the right.