Using Indicators¶

The mintalib.indicators module provides composable indicator objects that bind a calculation with its parameters (pandas only — for polars, use mintalib.expressions).

Indicators are named in upper case (e.g. SMA, EMA, MACD). An indicator instance is callable and can be passed directly to prices.assign() or invoked as SMA(50)(prices). The | operator chains indicators: EMA(20) | ROC(1) means ROC applied after EMA.

In [1]:

Copied!

import numpy as np
import pandas as pd

from mintalib.samples import sample_prices
from mintalib.indicators import EMA, SMA, ROC, RSI, EVAL, LOG, BBANDS
import numpy as np
import pandas as pd

from mintalib.samples import sample_prices
from mintalib.indicators import EMA, SMA, ROC, RSI, EVAL, LOG, BBANDS

Basic Usage¶

An indicator instance is a callable. Applied to a DataFrame, series-based indicators use the close column by default — the item parameter selects another column. A pandas Series or numpy array can be passed directly as well (results always come back as pandas objects):

In [2]:

Copied!

prices = sample_prices()

SMA(50)(prices)
prices = sample_prices()

SMA(50)(prices)

Out[2]:

date
1980-12-12           NaN
1980-12-15           NaN
1980-12-16           NaN
1980-12-17           NaN
1980-12-18           NaN
                 ...    
2026-04-20    260.501999
2026-04-21    260.268199
2026-04-22    260.239200
2026-04-23    260.234200
2026-04-24    260.145400
Length: 11433, dtype: float64

In [3]:

Copied!

SMA(50, item="open")(prices)
SMA(50, item="open")(prices)

Out[3]:

date
1980-12-12           NaN
1980-12-15           NaN
1980-12-16           NaN
1980-12-17           NaN
1980-12-18           NaN
                 ...    
2026-04-20    260.634220
2026-04-21    260.527001
2026-04-22    260.325201
2026-04-23    260.328401
2026-04-24    260.289601
Length: 11433, dtype: float64

In [4]:

Copied!

RSI(14)(prices["close"])
RSI(14)(prices["close"])

Out[4]:

date
1980-12-12          NaN
1980-12-15          NaN
1980-12-16          NaN
1980-12-17          NaN
1980-12-18          NaN
                ...    
2026-04-20    66.533046
2026-04-21    56.723876
2026-04-22    62.742587
2026-04-23    62.948689
2026-04-24    59.706160
Length: 11433, dtype: float64

Chaining¶

The | operator chains indicators left to right: LOG() | EMA(20) | ROC(1) applies LOG first, then EMA, then ROC. .then() is the fluent equivalent, and .alias() names the result:

In [5]:

Copied!

(LOG() | EMA(20) | ROC(1)).alias("trend")(prices)
(LOG() | EMA(20) | ROC(1)).alias("trend")(prices)

Out[5]:

date
1980-12-12         NaN
1980-12-15         NaN
1980-12-16         NaN
1980-12-17         NaN
1980-12-18         NaN
                ...   
2026-04-20    0.000893
2026-04-21    0.000370
2026-04-22    0.000779
2026-04-23    0.000721
2026-04-24    0.000503
Name: trend, Length: 11433, dtype: float64

The Assign Idiom¶

Because indicators are callables, they can be passed directly to prices.assign, which invokes each with the DataFrame. EVAL evaluates a pandas expression string against the columns — and since assign processes keyword arguments sequentially, it can reference columns created earlier in the same call:

In [6]:

Copied!





result = prices.assign(
    sma50 = SMA(50),
    sma200 = SMA(200),
    rsi = RSI(14),
    slope = LOG() | EMA(20) | ROC(1),
    uptrend = EVAL("sma50 > sma200")
).iloc[:, -5:]

result
result = prices.assign(
    sma50 = SMA(50),
    sma200 = SMA(200),
    rsi = RSI(14),
    slope = LOG() | EMA(20) | ROC(1),
    uptrend = EVAL("sma50 > sma200")
).iloc[:, -5:]

result

Out[6]:

	sma50	sma200	rsi	slope	uptrend
date
1980-12-12	NaN	NaN	NaN	NaN	0.0
1980-12-15	NaN	NaN	NaN	NaN	0.0
1980-12-16	NaN	NaN	NaN	NaN	0.0
1980-12-17	NaN	NaN	NaN	NaN	0.0
1980-12-18	NaN	NaN	NaN	NaN	0.0
...	...	...	...	...	...
2026-04-20	260.501999	252.134106	66.533046	0.000893	1.0
2026-04-21	260.268199	252.400445	56.723876	0.000370	1.0
2026-04-22	260.239200	252.719730	62.742587	0.000779	1.0
2026-04-23	260.234200	253.040016	62.948689	0.000721	1.0
2026-04-24	260.145400	253.342819	59.706160	0.000503	1.0

11433 rows × 5 columns

Multi-Output Indicators¶

Multi-output indicators return a DataFrame, so they cannot be assigned to a single column — join the result instead:

In [7]:

Copied!

prices.join(BBANDS(20)(prices))
prices.join(BBANDS(20)(prices))

Out[7]:

	open	high	low	close	volume	upperband	middleband	lowerband
date
1980-12-12	0.098298	0.098725	0.098298	0.098298	469033600	NaN	NaN	NaN
1980-12-15	0.093597	0.093597	0.093169	0.093169	175884800	NaN	NaN	NaN
1980-12-16	0.086758	0.086758	0.086331	0.086331	105728000	NaN	NaN	NaN
1980-12-17	0.088468	0.088895	0.088468	0.088468	86441600	NaN	NaN	NaN
1980-12-18	0.091032	0.091460	0.091032	0.091032	73449600	NaN	NaN	NaN
...	...	...	...	...	...	...	...	...
2026-04-20	270.329987	274.279999	270.290009	273.049988	36590200	270.976486	257.638998	244.301511
2026-04-21	271.500000	272.799988	265.399994	266.170013	50209800	271.890660	258.372999	244.855337
2026-04-22	267.820007	273.739990	266.869995	273.170013	43249200	274.037695	259.449500	244.861304
2026-04-23	275.049988	275.769989	271.649994	273.429993	33399600	275.925261	260.489999	245.054738
2026-04-24	272.760010	273.059998	269.649994	271.059998	38124500	277.074557	261.398499	245.722442

11433 rows × 8 columns

Pandas Expressions¶

With pandas >= 3.0, as_expr() converts an indicator into a pandas Expression. For multi-output indicators, as_expr(item) picks a single output — which makes them usable inside assign after all:

In [8]:

Copied!





prices.assign(
    upper = BBANDS(20).as_expr("upperband"),
    lower = BBANDS(20).as_expr("lowerband"),
).iloc[:, -2:]
prices.assign(
    upper = BBANDS(20).as_expr("upperband"),
    lower = BBANDS(20).as_expr("lowerband"),
).iloc[:, -2:]

Out[8]:

	upper	lower
date
1980-12-12	NaN	NaN
1980-12-15	NaN	NaN
1980-12-16	NaN	NaN
1980-12-17	NaN	NaN
1980-12-18	NaN	NaN
...	...	...
2026-04-20	270.976486	244.301511
2026-04-21	271.890660	244.855337
2026-04-22	274.037695	244.861304
2026-04-23	275.925261	245.054738
2026-04-24	277.074557	245.722442

11433 rows × 2 columns