Pixel-Verified Replica of an Invite-Only "AI" Indicator
A locked SmartAIML-branded indicator shown to be Lorentzian Classification plus Chandelier Exit at stock defaults, reproduced with zero differing pixels.
Nothing copied from the original: three open-source layers recombined, non-default styling measured off pixels, and the ML label colour used as proof that prediction values match, not just timing.
- 0 differing pixels on three timeframes
- Label colour match proves prediction values match
- Non-default styling measured, not guessed
- Nothing copied from the locked source
An indicator sold under an AI label with its source locked turned out to be three well-known open-source scripts stacked together. The replica recombines Lorentzian Classification (the machine-learning labels and kernel regression line), Chandelier Exit (the ATR trailing stops and state fills) and a multi-timeframe RSI table, all at their stock defaults, and matches the original with zero differing pixels on weekly, daily and hourly charts.
The verification is stronger than it sounds. The ML buy label's colour encodes the prediction total, so an exact colour match proves the k-nearest-neighbour output matches, not merely the signal timing. Several non-default style settings had to be measured from pixels rather than assumed: a wider kernel line, specific transparencies on the stop lines and fills, and the order of inputs so the status line reads the same.
The work was done with Win32-level automation of the TradingView desktop app (screenshots, pixel diffs, nearest-neighbour magnification and colour probes) before the DevTools route was discovered, and the pixel-compare helpers remain useful for any replica verification.
How it works
- Identify candidate open-source components from the visual signature
- Compose them in one script at stock defaults
- Load original and replica on the same chart, toggle visibility
- Pixel-diff screenshots and probe colours per region
- Adjust style constants until the diff is empty
Software and automation are infrastructure, not financial advice. No performance guarantees. Every live system needs staged testing, risk limits, and owner approval.