RoboQuant Review

How we rated RoboQuant

The claim RoboQuant is sold on is specific enough to be checked: describe a trading idea in plain English and the assistant writes Pine Script or Python that compiles. The page carrying the numbers behind that claim is the company’s own, a set of two hundred generation tasks it ran on itself and scored beside three named assistant models. Everything else a buyer needs before paying, the plan ladder, what a credit buys, the route an alert takes to a broker and the terms that govern a refund, sits on the company’s site; the list of what was read, with the date on each, is at the end of this review.

Each of the five scores is drawn from something the company publishes, and the sections below take them in the order a buyer meets them: the code the assistant writes, the path that code takes to a broker, the tests it runs before money is exposed to it, the material a new strategy can start from, and the plans with the credits they meter.

Code generation and accuracy

The assistant is the part RoboQuant is named for, and it is the part the company documents most closely. Its knowledge base is given as 5,000 verified code patterns, of which 3,200 are Pine Script v6 strategies, and the platform is described as checking generated code against the TradingView compiler before the result is handed over rather than leaving the first error to the reader. The languages named on the product pages are Pine Script for TradingView and Python for custom bots; a card on the same homepage adds MQL4, MQL5 and NinjaTrader, so the supported set is drawn at one width in the FAQ and another on the page beside it.

The number the company publishes for itself is a 94% first-try compilation rate across 200 Pine Script v6 generation tasks, set on the page beside three named models scored on the same run: Claude Sonnet 4.6 at 78%, GPT-5.2 Codex at 71% and Gemini 3.1 Pro at 58%. The bar for RoboQuant is labelled as working with the knowledge base, while the other three carry their makers’ names, and the note underneath says the figures come from internal testing and that results may vary. Read that way, the comparison sets the company’s assistant working with its own library against general models working without one, and the sixteen-point gap is the distance between the two. It is the sole figure of its kind published here, and nothing outside the company can rerun the 200 tasks.

4.5 for code generation and accuracy. The knowledge base is counted, the compiler check is described as part of the workflow rather than as a promise, and the company puts the test it ran on itself on the page with the caveat that goes with it. The score holds below the top because the figure comes from a run on the company’s own tasks, which makes it a claim about the product rather than something a reader can set beside it.

Automation and execution

Execution runs on TradingView alerts. An alert fires a webhook, the platform turns it into an order at the broker, and the company puts the round trip at under 50 milliseconds on Tradovate, which is the broker it names for futures. The same answer lists the funded-account programmes the routing covers, Apex, Topstep and My Funded Futures among them, and the higher plans are described as carrying unlimited accounts. What arrives around each order is a monitoring layer, a performance dashboard and an audit trail. Risk controls sit in the same feature list: a kill switch on each account, a daily loss limit and a cap on position size.

Two things are worth separating here. The first is that the reach is stated in more than one place and not identically: the homepage offers execution on any broker, a feature card names crypto exchanges beside Tradovate, and the FAQ answers the same question with Tradovate for futures. The second is the connector. Turning alerts into orders is the Webhook Connector, sold on its own at $39.90 a month with seven days free and included with Pro, Elite and Ultra, so a reader on the free account or the chart tool is buying analysis rather than execution.

4.3 for automation and execution. The path from alert to order is named end to end, the latency figure and the funded-account programmes are listed, and the controls a funded account needs are part of the feature set, while the broker the FAQ names for futures is a smaller set than the homepage wording implies.

Strategy testing and analytics

Strategy Lab is where a backtest is judged. Results are uploaded, for MetaTrader 5 either as a report file or through the import the changelog records, and the lab runs a Monte Carlo simulation with confidence intervals, tests for curve-fitting, and works walk-forward rather than reporting a single equity curve. A second tool takes the parameters a reader is optimising and runs them against the rules of the funded-account programmes the company names, Apex, Topstep, My Funded Futures, FTMO and Trading Pit, returning a probability of passing.

The lab reads work the reader brings rather than producing the backtest itself, which leaves the quality of the input outside the platform: a Monte Carlo run over a badly built test is a confident answer to the wrong question. The record around the tool is stronger than most products at this price offer. The changelog is dated, names the models in use, separates what is new from what is improved and what is fixed, and includes entries for corrections rather than only for additions; the MetaTrader 5 import and the fixes that followed it appear there too.

4.2 for strategy testing and analytics. The techniques are named one by one rather than gestured at, the funded-account rules are the firms’ own, and the changelog gives a dated account of what changed and when, while the analysis begins from results the reader uploads, so the quality of what it judges is set outside the platform.

Templates and platform coverage

The template library is the shortest route from the platform to a working strategy, and it is given as more than fifty strategies with the source included, grouped by the schools the company names, ICT, SMC and price action among them. Two converters sit beside it, one that rewrites MetaTrader code as Pine Script and one that builds a strategy from a video. The work is organised into workspaces, three on Pro, seven on Elite and fifteen on Ultra, with more available as an add-on.

The material is for reading and adapting rather than for taking away. Nothing in the published material describes the strategies as exportable, and the terms place the software and its content with the company, so a reader who wants the library as files outside the subscription will not find it here. The chart-reading tool is a separate purchase again: at $9 a month for three analyses a day and $19 for seven, it needs no workspace and no wallet, which makes it the cheapest way to see how the assistant reads a chart before committing to a plan.

4.1 for templates and platform coverage. The library is counted and its groupings are named, the two converters cover the most common way a reader arrives with code already written, and the plans include a free account and a low-priced way to test the analysis, while the strategies stay inside the subscription and none of it comes out as a file.

Price, credits and terms

Pro is $49 a month, Elite $99 and Ultra $199, each metering a monthly allowance of AI credits: 25,000 on Pro, 50,000 on Elite and 150,000 on Ultra, which the pricing page translates into roughly 165, 330 and 1,000 prompts. Paying for a year costs ten times the monthly rate, so the ladder is $490, $990 and $1,990, and the page calls the difference two months free. A free account carries no workspace and no credits, which makes it a way to read the documentation, and workspaces beyond the plan are $5 a month each.

Two things worth reading before paying sit on different pages. The refund is one: the pricing page offers the full amount back within the first seven days, while the terms handle refunds case by case. The two pages answer the same question differently. The credits are the other: they cannot be transferred and the terms say they expire according to the plan, so an allowance bought and not spent does not carry forward indefinitely. The rest is stated plainly. Stripe takes the payment, the contract is governed by Wyoming law and names FXC Algo Trade LLC, and liability is capped at what was paid in the preceding twelve months.

3.9 for price, credits and terms. The monthly rates sit in the middle of what this kind of platform charges, the annual price is genuinely ten months for twelve, and the free account gives the documentation away and says plainly that it carries no workspace and no credits, while the refund is described one way on the pricing page and another in the terms, and the allowance metered to each plan does not roll over.

Who RoboQuant is for

This suits someone who already has a view about how a market behaves and wants that view running without writing the Pine Script themselves, and who will test the result before money reaches it. The plan ladder is built for that reader at more than one size, and the execution path rewards the one who already trades futures through Tradovate or a firm the routing covers.

It fits two readers less well. Someone who expects the platform to choose the strategy will not get it here: the assistant writes what it is told, the lab judges what it is given, and the company’s own advice is to paper trade a strategy before it goes live, which the platform does not do for you. Someone who wants to trade an equity or spot crypto account is outside the routing as the FAQ describes it, since the broker named there is Tradovate for futures, and a service built around that venue will serve them better.