# Machine Learning: Lorentzian Classification | NextCandle

[All indicators](https://nextcandle.trade/) / Machine Learning: Lorentzian Classification

## Machine Learning: Lorentzian Classification
review & score
By [NK2](https://nextcandle.trade/about/#nk2) · checked 2026-11-03
Free open-source · ML / "AI" · by jdehorty

[75](https://nextcandle.trade/methodology/) Provisional out of 100
Last reviewed 2026-11-03 · Updated 2026-11-03: pros, cons and verdict re-sourced

Risk warning: Trading involves substantial risk of loss. Past or backtested performance does not guarantee future results. Scores rate the product, not trade outcomes. [Risk disclaimer](https://nextcandle.trade/risk-disclaimer/)
Best forTraders curious about machine-learning signals who want to read and tweak every line of a nearest-neighbour classifier.
[View Machine Learning: Lorentzian Classification script](https://nextcandle.trade/go/lorentzian-classification)

- Price: Free ([source](https://www.tradingview.com/u/jdehorty/))
- Platforms: TradingView
- Source code: Open source
- Repaint status: Realtime-bar repaint only (source) (source-code review, not a live bar-replay test, 2026-11-03)

## Repaint check
Code review of the current Pine v6 source (663 lines): no request.security() calls; training labels compare src[4] with src[0] (line 335), so they only use past bars; Buy/Sell shapes plot with offset=0 (lines 566-567). The author's comment (line 565) and tooltip (line 216) say signals can change intrabar until the bar closes. The imported MLExtensions and KernelFunctions libraries were not reviewed. ([source](https://www.tradingview.com/script/WhBzgfDu-Machine-Learning-Lorentzian-Classification/))

## Score breakdown

Factor | Points | Max | 
Repaint risk (code review done, live test pending) | 17 | / 25 | 
Source transparency | 20 | / 20 | 
Price value | 15 | / 15 | 
Backtest / claim honesty | 10 | / 15 | 
Vendor transparency | 8 | / 15 | 
Platform reach + alerts | 5 | / 10 | 
See the [methodology](https://nextcandle.trade/methodology/). Code review done, live test pending.

## Pros and cons

Pros
- PineCoders named it one of the best Editors' Picks of 2023
- Open-source Pine v6 with volatility, regime and ADX filters and an optional kernel-regression line
- Training labels use only past bars and there are no request.security() calls (our code review)
Cons
- The on-chart win-rate table is computed on the same history the model ran on, so treat it as in-sample
- Signals can change on the open bar; the author's own tooltip mentions intrabar repainting
- Depends on two of the author's libraries (MLExtensions, KernelFunctions) we haven't reviewed
- 663 lines and many inputs, so it's slow to learn
From the script’s public page and our code review (source). Not from trading tests.
Runs on TradingView: [open TradingView](https://nextcandle.trade/go/tradingview).
TradingView link is an affiliate link: we may earn a commission at no cost to you. It never changes the score.

## Get repaint test results by email
We only email when Machine Learning: Lorentzian Classification’s repaint test results are published. Unsubscribe anytime by replying. See [Privacy](https://nextcandle.trade/privacy/).
WebsiteEmailNotify me

## Similar
- Squeeze Momentum Indicator (80)
- WaveTrend Oscillator (80)
- SuperTrend (75)