Skip to content

ADX — Average Directional Movement Index

Measures the strength of a trend regardless of direction. Values above 25 indicate a strong trend; below 20 suggest a weak or ranging market.

Inputs: [high, low, close] | Options: [period] | Outputs: [adx]

Basic

use tulip_rs::indicators::adx::indicator;

let high  = vec![82.15, 81.89, 83.03, 83.30, 83.85,
                 83.90, 83.33, 84.30, 84.84, 85.00_f64];
let low   = vec![81.29, 80.64, 81.31, 82.65, 83.07,
                 83.11, 82.49, 82.30, 84.15, 84.11_f64];
let close = vec![81.59, 81.06, 82.87, 83.00, 83.61,
                 83.15, 82.84, 83.99, 84.55, 84.36_f64];

let inputs = [high.as_slice(), low.as_slice(), close.as_slice()];
let (outputs, mut state) = indicator(&inputs, &[14.0], None).unwrap();
println!("{:?}", outputs[0]); // ADX values

// State continuation — feed new bars without reprocessing history
let new_high  = vec![85.20_f64];
let new_low   = vec![84.50_f64];
let new_close = vec![85.00_f64];
let continued = state.batch_indicator(
    &[new_high.as_slice(), new_low.as_slice(), new_close.as_slice()],
    None,
).unwrap();
println!("{:?}", continued[0]);

import numpy as np
import tulip_rs

high  = np.array([82.15, 81.89, 83.03, 83.30, 83.85,
                  83.90, 83.33, 84.30, 84.84, 85.00], dtype=np.float64)
low   = np.array([81.29, 80.64, 81.31, 82.65, 83.07,
                  83.11, 82.49, 82.30, 84.15, 84.11], dtype=np.float64)
close = np.array([81.59, 81.06, 82.87, 83.00, 83.61,
                  83.15, 82.84, 83.99, 84.55, 84.36], dtype=np.float64)

outputs, state = tulip_rs.indicators.adx.indicator([high, low, close], [14.0])
print(outputs[0])  # ADX values

# State continuation
new_high  = np.array([85.20], dtype=np.float64)
new_low   = np.array([84.50], dtype=np.float64)
new_close = np.array([85.00], dtype=np.float64)
continued = state.batch_indicator([new_high, new_low, new_close])
print(continued[0])

import * as ti from 'tulip-rs-node';

const high  = [82.15, 81.89, 83.03, 83.30, 83.85, 83.90, 83.33, 84.30, 84.84, 85.00, 85.90, 86.58, 86.98, 88.00, 87.87];
const low   = [81.29, 80.64, 81.31, 82.65, 83.07, 83.11, 82.49, 82.30, 84.15, 84.11, 84.03, 85.39, 85.76, 87.17, 87.01];
const close = [81.59, 81.06, 82.87, 83.00, 83.61, 83.15, 82.84, 83.99, 84.55, 84.36, 85.53, 86.54, 86.89, 87.77, 87.29];

const [outputs, state] = ti.adx.indicator([high, low, close], [14]);
console.log('ADX(14):', outputs[0]);

// State continuation
const n = high.length - 5;
const [, state2] = ti.adx.indicator([high.slice(0, n), low.slice(0, n), close.slice(0, n)], [14]);
const continued = state2.batchIndicator([high.slice(n), low.slice(n), close.slice(n)]);
console.log('Continued ADX:', continued[0]);

import { init } from 'tulip-rs-wasm';
import * as ti from 'tulip-rs-wasm';

await init(); // bundler resolves the WASM asset automatically

const high  = [82.15, 81.89, 83.03, 83.30, 83.85, 83.90, 83.33, 84.30, 84.84, 85.00, 85.90, 86.58, 86.98, 88.00, 87.87];
const low   = [81.29, 80.64, 81.31, 82.65, 83.07, 83.11, 82.49, 82.30, 84.15, 84.11, 84.03, 85.39, 85.76, 87.17, 87.01];
const close = [81.59, 81.06, 82.87, 83.00, 83.61, 83.15, 82.84, 83.99, 84.55, 84.36, 85.53, 86.54, 86.89, 87.77, 87.29];

const [outputs, state] = ti.adx.indicator([high, low, close], [14]);
console.log('ADX(14):', outputs[0]);

// State continuation
const n = high.length - 5;
const [, state2] = ti.adx.indicator([high.slice(0, n), low.slice(0, n), close.slice(0, n)], [14]);
const continued = state2.batchIndicator([high.slice(n), low.slice(n), close.slice(n)]);
console.log('Continued ADX:', continued[0]);

Optional Outputs

adx exposes 3 optional outputs: dx, atr, tr. Pass a boolean mask as the third argument — one bool per optional output, in order.

use tulip_rs::indicators::adx::indicator;

let close = vec![81.59, 81.06, 82.87, 83.00, 83.61, 83.15, 82.84, 83.99, 84.55, 84.36_f64];
let high  = close.iter().map(|x| x + 1.0).collect::<Vec<_>>();
let low   = close.iter().map(|x| x - 1.0).collect::<Vec<_>>();

let mask = [true, true, false];
let (outputs, _state) = indicator(
    &[high.as_slice(), low.as_slice(), close.as_slice()],
    &[14.0],
    Some(&mask),
).unwrap();

let adx = &outputs[0]; // adx (primary)
let dx  = &outputs[1]; // dx (optional — requested)
let atr = &outputs[2]; // atr (optional — requested)
// tr not requested — omitted from outputs

import numpy as np
import tulip_rs

close = np.array([81.59, 81.06, 82.87, 83.00, 83.61, 83.15, 82.84, 83.99, 84.55, 84.36], dtype=np.float64)
high  = close + 1.0
low   = close - 1.0

outputs, state = tulip_rs.indicators.adx.indicator(
    [high, low, close], [14.0],
    optional_outputs=[True, True, False],
)

adx = outputs[0]  # adx (primary)
dx  = outputs[1]  # dx (optional — requested)
atr = outputs[2]  # atr (optional — requested)
# tr not requested — omitted from outputs

adx exposes 3 optional outputs: dx, atr, tr.

const [allOut] = ti.adx.indicator([high, low, close], [14], [true, true, true]);
const adx = allOut[0]; // primary
const dx  = allOut[1]; // optional 0: dx
const atr = allOut[2]; // optional 1: atr
const tr  = allOut[3]; // optional 2: tr

// Request only dx
const [partial] = ti.adx.indicator([high, low, close], [14], [true, false, false]);

SIMD

By assets — same options, N assets in parallel:

use tulip_rs::indicators::adx::indicator_by_assets;

let inputs: [&[&[f64]; 3]; 4] = [
    &[h1.as_slice(), l1.as_slice(), c1.as_slice()],
    &[h2.as_slice(), l2.as_slice(), c2.as_slice()],
    &[h3.as_slice(), l3.as_slice(), c3.as_slice()],
    &[h4.as_slice(), l4.as_slice(), c4.as_slice()],
];
let results = indicator_by_assets::<4>(&inputs, &[14.0], None).unwrap();
for (i, asset_outputs) in results.iter().enumerate() {
    println!("Asset {}: {:?}", i + 1, asset_outputs[0]);
}

By options — same asset, N option sets in parallel:

use tulip_rs::indicators::adx::indicator_by_options;

let opts: [&[f64; 1]; 4] = [&[7.0], &[14.0], &[21.0], &[28.0]];
let results = indicator_by_options::<4>(&inputs, &opts, None).unwrap();
for (i, out) in results.iter().enumerate() {
    println!("Period {}: {:?}", opts[i][0], out[0]);
}

By assets — same options, N assets in parallel (must be 2, 4, 8, or 16):

simd_inputs = [
    [h1, l1, c1],
    [h2, l2, c2],
    [h3, l3, c3],
    [h4, l4, c4],
]
outputs_list, states = tulip_rs.indicators.adx.simd_by_assets(simd_inputs, [14.0])
for i, asset_outputs in enumerate(outputs_list):
    print(f"Asset {i+1}: {asset_outputs[0]}")

By options — same asset, N option sets in parallel:

simd_options = [[7.0], [14.0], [21.0], [28.0]]
outputs_list, states = tulip_rs.indicators.adx.simd_by_options(
    [high, low, close], simd_options
)
for i, out in enumerate(outputs_list):
    print(f"Period {simd_options[i][0]}: {out[0]}")

By assets — same period applied to 4 assets in parallel:

const simdInputs = [
    [[...high], [...low], [...close]],
    [high.map(v => v * 1.1), low.map(v => v * 1.1), close.map(v => v * 1.1)],
    [high.map(v => v * 0.9), low.map(v => v * 0.9), close.map(v => v * 0.9)],
    [high.map(v => v * 1.02), low.map(v => v * 1.02), close.map(v => v * 1.02)],
];
const [results] = ti.adx.simdByAssets(simdInputs, [14]);
results.forEach((out, i) => console.log(`Asset ${i + 1}:`, out[0]));

By options — same asset, 4 different periods in parallel:

const simdOptions = [[7], [14], [21], [28]];
const [results] = ti.adx.simdByOptions([high, low, close], simdOptions);
results.forEach((out, i) => console.log(`Period ${simdOptions[i][0]}:`, out[0]));