fintech-algorithms
Using a coding agent? Give it the skill: npx skills add IslamBaraka90/Fintech-Algorithms-Library What it does →

Regression Channel

Install and import#

bash
npm install fintech-algorithms
ts
import { regressionChannel } from "fintech-algorithms/technical-indicators/bands-envelopes-and-squeezes/regression-channel";

Signature#

regressionChannel(input)

Regression Channel: the end point of a rolling least-squares fit of close over period bars, with bands multiplier root-mean-square residuals above and below that fitted point.

Parameters#

NameTypeNotes
inputTopicInputbars is the required OHLCV array -- each bar carries timestamp, open, high, low, close, volume and an optional basis, strictly ordered by timestamp. From parameters this topic reads period (default 20, integer >= 2), the regression window, and multiplier (default 2, a finite number >= 0), the channel half-width in residual units.

Returns#

TopicResult

series holds center, the fitted close at the end of each window, plus upper and lower; latest carries the last of each. All three share a warm-up of period - 1 leading nulls, so ready_at is 19 at the default period.

Warm-up#

The first `period - 1` bars (19 at the default period) positions are null. The fit is recomputed from scratch on each full window, so nothing is emitted until period closes exist. The channel does not extend or anchor across windows.

Errors#

  • When parameters.period is not an integer >= 2 — throws Error
  • When parameters.multiplier is not a finite number, or is negative — throws Error
  • When a bar is missing open, high, low, close, or volume, or one of them is not a finite number — throws Error

Complexity: time O(n * period), space O(n).

Worked example#

executed Captured by running this function on the input its own test provides. Real output of real code — but not asserted against a published figure.

Input#

input
{
  "bars": [
    {
      "timestamp": "2024-01-02",
      "basis": "synthetic-unadjusted",
      "open": 100,
      "high": 101.45,
      "low": 98.695,
      "close": 100,
      "volume": 750000,
      "benchmark": 200
    },
    {
      "timestamp": "2024-01-03",
      "basis": "synthetic-unadjusted",
      "open": 101.49111452,
      "high": 103.38381693,
      "low": 100.05480022,
      "close": 101.78791214,
      "volume": 795117,
      "benchmark": 200.56326135
    },
    {
      "timestamp": "2024-01-04",
      "basis": "synthetic-unadjusted",
      "open": 102.45519048,
      "high": 104.6701838,
      "low": 100.91147007,
      "close": 102.9549389,
      "volume": 840234,
      "benchmark": 201.11020913
    }
  ],
  "parameters": {}
}

Call#

regressionChannel(input)

Returns#

object with 9 fields: topic_id, title, state, ready, ready_at, series, latest, parameters, …

{
  "topic_id": "D07-F07-A05",
  "title": "Regression Channel",
  "state": "calculated",
  "ready": true,
  "ready_at": 19,
  "series": {
    "center": [null, null, null, null, null, null],
    "upper": [null, null, null, null, null, null],
    "lower": [null, null, null, null, null, null]
  },
  "latest": {
    "center": 103.97453763199998,
    "upper": 108.29510693120406,
    "lower": 99.6539683327959
  },
  "parameters": {},
  "diagnostics": {
    "causal": true,
    "input_count": 96
  }
}

Diagrams#

Regression Channel — article hero
Regression Channel — concept map
Regression Channel — decision comparison
Regression Channel — worked example

Calculation flow#

Regression Channel calculation flow
flowchart LR
    A["ordered OHLC or close observations, a declared centerline,"] --> B["Validate order, basis, and finite values"]
    B --> C["Apply the selected Regression Channel convention"]
    C --> D["Emit value, readiness, and diagnostics"]
    D --> E["Interpret descriptively; test outcomes separately"]
    B -->|invalid or insufficient| X["Withhold output with a reason"]
Regression Channel readiness and evidence states
stateDiagram-v2
    [*] --> Waiting
    Waiting --> Ready: enough valid causal observations
    Waiting --> Rejected: malformed or unsupported input
    Ready --> Calculated: selected formula applied
    Calculated --> Interpreted: diagnostic and limitation retained
    Interpreted --> Ready: next observation arrives
    Rejected --> Waiting: corrected input and deterministic reset

How it works#

This page states the contract — how to call it correctly. The article explains the concept: why it works, and where it breaks.

Read the article →

References#

  • TA-Lib function groups — see linked primary or authoritative record
  • TA-Lib C/C++ API — see linked primary or authoritative record
  • TA-Lib maintained source — see linked primary or authoritative record
  • Evidence decision
  • Level 1 evidence map

The rest of the Bands, Envelopes, and Squeezes family#