LuxAlgo Quant: what the AI coding agent can and can't do, per its docs and the reviews

LuxAlgo built its name on TradingView indicators. In 2026 it moved the spotlight to Quant, an AI agent inside its own charting platform, Vela. The pitch on the Quant page is direct: 'Generic AI meets markets the way a tourist meets a city: confidently, and wrong.' Here is what can be checked. Plans and prices are tracked on our LuxAlgo Quant tool page.

What Quant is documented to do

According to LuxAlgo's product page and FAQ, read October 14, 2026:

• Write and repair Pine Script from a plain-English description, or read and debug code you paste in.
• Work from an image: give it a chart screenshot and it rebuilds the logic as an editable script.
• Draw on the chart: fair value gaps, liquidity, market structure and key levels as editable drawings, with prices taken from the chart's own bars.
• Backtest in the same window, with an equity curve, drawdown and performance analytics.
• Validate its own code: the page says it 'validates every line of Pine Script it writes' and 'won't ship anything it can't prove runs'.

The benchmark: real method, missing numbers

LuxAlgo published a Pine Script accuracy benchmark dated February 21, 2026, across 100 indicator evaluations of varying complexity. Its failure definition is stricter than most: syntax or runtime errors, NaN outputs, constant unexpected values, or failing to do what was asked. The blog says Quant posted a pass-rate advantage over ChatGPT and Gemini. The pass rates themselves are not stated in the page text we read; they appear only in a chart graphic, and the test set is not published. Pass rate: not disclosed in text. Interval: not disclosed. Run by the vendor.

That definition still leaves the hardest failure out. A script can run, return sensible values and implement the wrong idea. LuxAlgo's own strategy guide says it plainly: a successful run 'does not establish that the logic is correct or profitable.'

What reviewers found

PickMyTrade's review (June 25, 2026) spent a week on builds and concludes the code rarely arrives broken thanks to the validation loop, but that Quant 'builds indicators, not edges.' It reports LuxAlgo's advice to start a fresh conversation after roughly 50 messages to avoid burning credits on long context, and that heavy users on lower tiers run out of credits faster than expected. One caution about that review: the plan prices it lists ($39.99, $67.99, $119.99) do not match LuxAlgo's pricing page on October 14, which shows how quickly review prices go stale.

TraderJury's review of the Signals & Overlays toolkit, LuxAlgo's best-known paid script, is not about Quant but matters for anyone using Quant to build signals: LuxAlgo states that a signal on the most recent candle can change and is only confirmed at the next candle's open, the code is closed source, and 'no independent forward test is published.'

We found no independent, scored test of Quant's chart markup or backtest engine, and no published forward test of any strategy it built.

What it costs (read October 14, 2026)

Credits are shared between Quant and the Backtesting Assistant and reset on the first of each month. LuxAlgo does not publish how many credits a typical request uses. The pricing page says promotional pricing applies to the initial payment and renewals bill at the then-current standard rate; the 30-day money-back guarantee covers the first payment only, which TraderJury also notes excludes renewals.

What the marketing shows, and what we leave out

LuxAlgo's product pages animate a Quant backtest report with a net profit figure, a win rate and a drawdown. We do not reproduce those numbers: they are a hypothetical demo, and LuxAlgo's own disclaimer on the same pages says simulated results 'do not represent actual trading.' Its legal text also says outputs from Quant are 'produced from criteria and inputs you control', which puts the responsibility for any strategy on the user.

A practical way to use it

• Ask Quant for an indicator whose correct output you already know, then compare signals bar by bar.
• Separate live-candle signals from confirmed ones before you judge any backtest.
• Re-run any backtest on data the idea was not built on, with fees and slippage set to realistic values.
• Watch the credit gauge in the first week before choosing a tier, and note the renewal price, not just the first-year price.