openbb-technical

openbb-technical adds the obb.technical namespace, and obb.technical.signals beneath it. It computes indicators and signals on price data you pass in, from any command's results, a DataFrame, or a list of dictionaries. It fetches nothing.

V5 rewrote the extension. It has 59 commands, up from 27 in V4, and the indicator commands now return only the date and the indicator's own columns rather than the input bars with columns appended. Join results on date when you need both.

Installation

pip install openbb-technical
openbb-build

The package depends on openbb-core[pandas], pandas-ta-openbb, scipy, and statsmodels. Every command is also a POST endpoint under /api/v1/technical/ when served with openbb-api.

Commands

Category	Commands
Overlay	sma, ema, wma, hma, zlma, dema, tema, kama, frama, vwma, bbands, donchian, kc, ichimoku, supertrend
Oscillator	rsi, stoch, cci, fisher, cg, williams_r, mfi, trix, ultimate_oscillator, awesome_oscillator
Trend	macd, adx, di, aroon, choppiness
Volume	obv, ad, adosc, vwap
Volatility	atr, realized_volatility, realized_volatility_compare, cones
Structure	fib, demark, pivot_points
Statistics	returns_stats, drawdown, stationarity, hurst, autocorrelation, clenow
Signal	signals.crossovers, signals.oscillator_signals, signals.breakouts, signals.divergences, signals.candlestick_patterns, signals.regime
Multi	correlation, correlation_matrix, relative_rotation, screen, multi, indicators

Commands new in V5 include dema, tema, kama, frama, vwma, supertrend, williams_r, mfi, trix, ultimate_oscillator, awesome_oscillator, di, choppiness, realized_volatility, realized_volatility_compare, pivot_points, all of the statistics category except clenow, all six signals, and every multi command except relative_rotation.

Most commands take data, index (default "date"), and, for single-column calculations, target (default "close"). Indicators that read high, low, or volume need those columns in data.

Single-symbol indicators

The examples use daily bars from Cboe, which need no credentials.

from openbb import obb

spy = obb.cboe.equity.historical(symbol="SPY", start_date="2022-01-01").results

rsi = obb.technical.rsi(data=spy, length=14)
bands = obb.technical.bbands(data=spy, length=20, std=2.0)
macd = obb.technical.macd(data=spy, fast=12, slow=26, signal=9)
pivots = obb.technical.pivot_points(data=spy, method="camarilla", anchor="week")

bands.to_df()

bbands returns lower, middle, upper, bandwidth, and percent; macd returns macd, signal, and histogram. pivot_points supports the classic, fibonacci, woodie, camarilla, and demark methods over day, week, or month anchors.

realized_volatility estimates annualized volatility over a rolling window with one of six estimators: std, parkinson, garman_klass, hodges_tompkins, rogers_satchell, or yang_zhang, the default. realized_volatility_compare runs the estimators named in models side by side, and cones places current realized volatility within its historical range across several windows.

rv = obb.technical.realized_volatility(data=spy, model="garman_klass", window=21)
compare = obb.technical.realized_volatility_compare(
    data=spy, models=["std", "parkinson", "yang_zhang"], window=30
)
cones = obb.technical.cones(data=spy, model="yang_zhang")

The statistics commands describe the series rather than smooth it: returns_stats for distribution and risk-adjusted return figures at a chosen frequency, drawdown for peak-to-trough losses, stationarity for ADF and KPSS tests with verdicts, hurst for the Hurst exponent by rescaled range or DFA, and autocorrelation for ACF and PACF.

Signals

The signal commands turn indicators into events.

Command	Output	Key arguments
crossovers	One row per moving-average cross, with direction bullish or bearish	fast_length, slow_length, mamode of sma, ema, wma, hma, or zlma
breakouts	One row per close beyond a channel, with direction upside or downside	method of donchian or bollinger, length, band_std
divergences	One row per price and oscillator divergence	indicator of rsi, macd, stoch, or cci; lookback; min_swing_distance
candlestick_patterns	One row per detected pattern, with direction and confidence	patterns, or None for all
oscillator_signals	One row per bar with a regime of overbought, oversold, or neutral and four boolean crossing flags	indicator of rsi, mfi, stoch, williams_r, or cci; optional thresholds
regime	One row per bar classified as strong_trend, weak_trend, ranging, or transition, with a regime_changed flag	ADX and choppiness lengths and thresholds

golden_crosses = obb.technical.signals.crossovers(
    data=spy, fast_length=50, slow_length=200, mamode="sma"
)
rsi_divergences = obb.technical.signals.divergences(data=spy, indicator="rsi", lookback=90)
regime = obb.technical.signals.regime(data=spy)

Multi-symbol analysis

correlation, correlation_matrix, relative_rotation, and screen take long-format data for several symbols, with one row per symbol and date and a symbol column. A multi-symbol provider request returns that shape directly.

basket = obb.cboe.equity.historical(
    symbol="XLE,XLF,XLK,XLV,XLU,SPY", start_date="2022-01-01"
).results

matrix = obb.technical.correlation_matrix(data=basket, window=60, method="spearman")
pairs = obb.technical.correlation(data=basket, window=60, pairs=[("XLE", "SPY"), ("XLK", "SPY")])
rrg = obb.technical.relative_rotation(data=basket, benchmark="SPY")

correlation_matrix returns the symbol list and a nested-list matrix as of the last date, or as_of_date. correlation returns a rolling series per pair; leave pairs unset to compute every pair. relative_rotation returns JdK RS-Ratio and RS-Momentum series for each symbol against the benchmark, which must be one of the symbols in data. The price and volume studies need more than 252 daily observations, and the volatility study more than 504.

screen computes indicators for each symbol and returns the symbols whose latest values satisfy the conditions. A condition names an indicator command, one of that command's output columns, an operator, and a value, with optional indicator_params. gt, gte, lt, lte, and eq compare against a number; between takes (low, high); crossed_above, crossed_below, made_high, and made_low take (threshold, lookback_bars). combine="or" returns symbols that meet any condition instead of all of them.

oversold_uptrend = obb.technical.screen(
    data=basket,
    conditions=[
        {"indicator": "rsi", "column": "rsi", "operator": "lt", "value": 45},
        {"indicator": "adx", "column": "adx", "operator": "gt", "value": 20},
    ],
)

Composition and discovery

multi runs several indicator commands on one series and merges them by date. Each request names a command and its arguments. Each result row has a date and a values dictionary keyed "<indicator>.<column>"; unknown indicator names are skipped.

features = obb.technical.multi(
    data=spy,
    indicators=[
        {"indicator": "rsi", "params": {"length": 14}},
        {"indicator": "atr", "params": {"length": 14}},
        {"indicator": "bbands", "params": {"length": 20}},
    ],
)

indicators returns a catalog of the registered commands, optionally filtered by category: overlay, oscillator, volatility, volume, trend, signal, structure, stats, multi, or all. Each entry lists the command's name, category, description, required input columns, parameters with types, defaults, choices, and bounds, output columns, and an example call. It is intended for building parameter forms or tool definitions without hard-coding command names.

catalog = obb.technical.indicators(category="oscillator").results
[(entry.name, entry.requires_columns) for entry in catalog.indicators]

Charts

With openbb-charting installed, eleven commands accept chart=True: sma, ema, wma, hma, zlma, rsi, macd, adx, aroon, cones, and relative_rotation.

relative_rotation draws the Relative Rotation Graph, RS-Ratio against RS-Momentum, with a trailing tail for each symbol. cones draws, for each window, the current realized volatility with the historical minimum, lower quantile, median, upper quantile, and maximum.

rrg = obb.technical.relative_rotation(data=basket, benchmark="SPY", chart=True)
rrg.show()

Go to the reference for every parameter and returned field.