OpenBB/openbb_platform/extensions/technical/docs/user-guides/02-signals.md

Signal Endpoints

The /signals/ subrouter exposes six event-detection endpoints. Unlike the indicator endpoints (which return a dense per-bar series), these emit a sparse list of detected events — rows only on bars where the condition fires.

All examples assume:

from openbb import obb

data = obb.equity.price.historical(
    symbol="SPY", start_date="2022-01-01", provider="yfinance"
).results

Crossovers

Detects moving-average crossover events between a fast and slow MA.

events = obb.technical.signals.crossovers(
    data=data,
    fast_length=20,
    slow_length=50,
    mamode="sma",       # or "ema", "wma", "hma", "zlma"
).results

Each event row carries: date, direction ("bullish" / "bearish"), fast_value, slow_value, price, distance (price gap between the two MAs at the crossover bar).

Oscillator Signals

Labels every bar with a regime (overbought / oversold / neutral) for the chosen oscillator and flags the four boundary-crossing events.

sig = obb.technical.signals.oscillator_signals(
    data=data,
    indicator="rsi",                 # or "mfi", "stoch", "williams_r", "cci"
    length=14,
    overbought_threshold=70.0,       # optional override; defaults per indicator
    oversold_threshold=30.0,
).results

This is a dense output (one row per bar) with boolean flags crossed_into_overbought, crossed_into_oversold, crossed_out_of_overbought, crossed_out_of_oversold — convenient for plotting shaded regimes and vertical event markers.

Breakouts

Detects when price closes beyond a rolling-channel envelope.

events = obb.technical.signals.breakouts(
    data=data,
    method="donchian",   # rolling high/low; or "bollinger" (mean +/- stdev band)
    length=20,
    band_std=2.0,        # only used for "bollinger"
).results

Each event row carries: date, direction, price, band (the breached level), magnitude (signed price - band), bars_in_range (consolidation length).

Divergences

Detects classical bullish/bearish/hidden/exaggerated divergences between price and an oscillator across the last lookback bars.

events = obb.technical.signals.divergences(
    data=data,
    indicator="rsi",                 # or "macd", "stoch", "cci"
    indicator_length=14,
    lookback=60,
    swing_window=3,                  # bars on either side that define a swing high/low
).results

Each event row identifies the kind, the two swing dates that form the pattern, the indicator and price values at each swing, and the bar where the divergence was confirmed.

Candlestick Patterns

Single-bar and two-bar pattern detection.

events = obb.technical.signals.candlestick_patterns(
    data=data,
    patterns=["doji", "hammer", "shooting_star", "engulfing"],   # None = all
).results

Each event row carries: date, pattern name, direction (bullish / bearish / neutral), confidence (0-1, based on body/wick ratios meeting textbook thresholds).

Supported patterns: doji, hammer, inverted_hammer, hanging_man, shooting_star, bullish_engulfing, bearish_engulfing, morning_star, evening_star, three_white_soldiers, three_black_crows, harami_bullish, harami_bearish, piercing, dark_cloud_cover. Pass patterns=None to scan for all of them.

Regime

Classifies each bar into one of four regimes based on ADX strength and price trendiness.

out = obb.technical.signals.regime(
    data=data,
    length=14,
    adx_strong_threshold=25.0,
    adx_weak_threshold=20.0,
).results

Regimes: strong_trend, weak_trend, ranging, transition. Each row also flags regime_changed on bars where the label shifts — useful for triggering position-sizing rules tied to regime persistence.

Backtest pattern

The sparse event endpoints (crossovers, breakouts, divergences, candlestick_patterns) are designed to plug directly into a vectorised backtester:

events = obb.technical.signals.crossovers(data=data, fast_length=20, slow_length=50).results
trades = [
    e for e in events
    if e.direction == "bullish" and e.distance > 0.5  # filter weak crossovers
]

The dense endpoints (oscillator_signals, regime) are designed for filter conditions:

sig = obb.technical.signals.oscillator_signals(data=data, indicator="rsi", length=14).results
regime = obb.technical.signals.regime(data=data).results

# Only act on RSI crosses during a strong trend
combined = [
    s for s, r in zip(sig, regime)
    if s.crossed_into_overbought and r.regime == "strong_trend"
]