StrategyQuant X review: strategies from the machine

What StrategyQuant X is

StrategyQuant X (SQX for short) is the main product of StrategyQuant s.r.o. in Prague. Per its company page, the firm has worked on algo-trading software for 15 years and cites more than 23,000 customers. SQX is a desktop platform that generates trading strategies: you set market, timeframe, building blocks and requirements, the program combines entry and exit rules using machine learning and genetic programming, tests every combination and keeps what passes your filters.

The family also includes QuantAnalyzer for analysing finished results, QuantDataManager for historical data, the visual strategy editor AlgoWizard (built into SQX) and AlgoCloud for automated stock trading in the cloud. This report covers SQX itself.

1. 11. Sub-menu: Overview, Features, Pricing, Download, Stories and More. Prices, features and download details in this report come from these pages.
2. 22. Free Trial and Buy Now: the trial runs 14 days with full functionality and no credit card.
3. 33. Systems: “Runs on Windows 11/10, Mac, Linux, Windows Server”. Further down the same page still says “Windows 10/8.1/7”; the download page is authoritative.
4. 44. Modules: Getting started, Builder, Retester, Optimizer, Data Manager, Custom projects and AlgoWizard.
5. 55. Result card: profit factor, symbol, timeframe and profit of a sample strategy, next to “Passed” for walk forward, Monte Carlo and advanced robustness.

The home page advertises “180 days” the old way against “14 days” with StrategyQuant X and says artificial intelligence makes you “1000x faster”. These are advertising claims we cannot verify. The FAQ is more sober: it says SQX provides no ready-made profitable robots and guarantees no results, and that how much you can earn cannot be predicted reliably.

How strategies are made: Builder, Retester, Improver, Optimizer

The core is the Builder. It generates strategies randomly or by genetic evolution from building blocks: per the vendor, all common indicators such as RSI, MACD or CCI, candle patterns, four entry types and six exit types. Ranking uses criteria you choose, such as net profit, return to drawdown or Sharpe ratio. Alongside are the Retester (check existing strategies on other markets, timeframes or settings), the Improver (search only part of a strategy anew, such as the exit) and the Optimizer (optimise parameters simply or in walk-forward mode).

1. 11. Step 1, build process: machine learning and genetic programming generate new strategies for forex, futures and equities and, per the vendor, check millions of combinations of conditions, order types and price levels.
2. 22. Step 2, robustness tests: Monte Carlo, walk-forward optimisation and matrix, System Parameter Permutation, Optimization Profile and what-if, optionally applied automatically as part of the build.

Important for your own ideas: strategies can be generated from your own templates. A template is a strategy with placeholders that SQX fills randomly, so the core idea stays yours and only parts are searched. Add strategies across several timeframes and symbols (e.g. trading H1 with confirmation on H4 and D1), fuzzy logic (a signal counts if a set share of conditions is true) and trade management with several partial exits (ATM). On ATM, StrategyQuant writes that even a simple setting improves 10 to 30% of strategies; that is a preliminary vendor claim.

If you already know what you want to trade, you rebuild the rules without code in the visual editor AlgoWizard. Custom indicators and building blocks work via Java snippets or as “custom blocks” in AlgoWizard. Per the FAQ, MetaTrader indicators cannot be imported.

Robustness tests and automated workflows

This is where SQX is strongest, and where the editions differ most. Per the features page there are:

• –Two separate Monte Carlo tests with more than nine simulation types that play through random variations. Strategies that fail can be discarded automatically; you can add your own simulation models.
• –Walk-forward optimisation and walk-forward matrix: the matrix tests many combinations of windows and shows in a 3D view whether stable regions exist or only single peaks.
• –System Parameter Permutation (SPP) and Optimization Profile: two methods after Dave Walton and Robert Pardo that check whether a strategy works only with one exact parameter set.
• –Any number of out-of-sample periods with separate filters per period, plus MAE/MFE and daily equity.
• –Custom Projects: your own workflows of any number of steps (build, retest, optimise, filter) that run for hours or days without intervention.
• –Custom analysis: your own evaluations and filters, optionally calling external programs or Python scripts.

1. 11. Title bar: “StrategyQuant X Ultimate Build 133 (Full license)”. So the image is several versions older than the current build 145.
2. 22. Module bar: Getting started, Builder, Retester, Optimizer, Data Manager, Custom Projects, AlgoWizard and SQ 4 Business.
3. 33. Optimisation tabs: Walk-Forward Results, Overview, Optimization profile, Sys. Param Permutation, List of trades, Equity chart, Trade analysis and Strategy config.
4. 44. Walk-forward chart: 20 runs with 30% out-of-sample; the red lines mark window boundaries, with volume, drawdown and periods below.
5. 55. Result list: original strategy and walk-forward variant, with filter result “PASSED” and the chosen window.
6. 66. System note: Windows 11/10, Mac, Linux and Windows Server.

These are exactly the tools Starter lacks: per the price table, advanced robustness tests, the Optimizer (simple and walk-forward) and Custom Projects start with Professional. Starter can build, retest and improve, but cannot check what makes a generator usable in the first place. If you are serious about generating, treat Starter only as an entry point with a later upgrade (per the pricing page $550, or $350 on request by email).

The real danger: lucky hits in a mountain of data

A generator that tests millions of rule combinations on the same historical data will always find some that look good, even if none of them has a real edge. That is not a weakness of StrategyQuant in particular but statistics: with enough attempts, lucky hits are certain. A user quote on the home page illustrates this unintentionally well: with tight filters, it says, SQX finds about one strategy that “works” per million iterations, roughly six a day.

StrategyQuant knows this and builds in the robustness tests for exactly that reason. They reduce the risk but do not remove it: Monte Carlo and walk-forward also work on the past, and if you tweak filters long enough, you end up fitting the tests. What helps in practice:

• –Keep out-of-sample truly untouched: one period neither Builder nor Optimizer ever sees, checked only once at the end.
• –Retest on your broker’s data: the free built-in data comes from Dukascopy. Exported EAs belong in your platform’s strategy tester with your broker’s data before going live; StrategyQuant’s FAQ describes this step too.
• –Forward-test on demo or a small account: only when a strategy holds up for months outside all tests is it more than a candidate.
• –Prefer simplicity: few rules and few parameters survive new data more often than complicated combinations.

More in our guides on backtesting and forward testing and why EAs fail live.

Engine, data and export

SQX has its own backtest engine, per the vendor with tick precision and thousands of backtests per second depending on data and precision. It is said to be built to match your trading platform’s backtests “exactly”, only several times faster. We could not check that; different data, spreads, commissions and execution rules almost always cause deviations in practice. Per the changelog, build 145 added a full NinjaTrader backtest engine.

Data: Dukascopy tick data for forex, CFDs and crypto with more than ten years of history is included free, equity data can be downloaded from Yahoo Finance, and your own data can be imported. US equities and futures in minute and daily resolution are available as a data subscription; per the data page from $59 for six months depending on the package (equities only, daily data); the complete package with minute data costs $269 a year. All data prices were shown as reduced on the check date. Subscription data works only in the full version and cannot be exported. New in build 145 are weekly COT data for 95 futures symbols from SpreadCharts and Volume and Market Profile building blocks.

• –Export as complete, open source code for MetaTrader 4, MetaTrader 5 (hedging and netting), TradeStation, MultiCharts, JForex and, since build 145, NinjaTrader 8 (NinjaScript). Per the FAQ the code is not locked and may be modified.
• –Commercial use: with a full licence the strategies are yours; you may trade, sell or rent them. Ultimate adds “SQ for Business” with support for the MetaQuotes MQL Market.

For MT5 users, retesting on broker data is where SQX and the ATS Backtest Manager complement each other: SQX generates and pre-selects, the real MT5 Strategy Tester confirms or refutes on the broker’s data. Results from both can then be combined in QuantAnalyzer.

Q: the AI assistant in build 145

With build 145 of 2026-11-03, StrategyQuant built in the assistant Q. Per the changelog you describe a trading idea in plain language, and Q builds and backtests strategies, compares databanks, configures Builder, Retester, Optimizer and Custom Projects, analyses markets with Python and writes custom indicators and snippets for MT4, MT5, NinjaTrader 8 and EasyLanguage. The Q page explicitly compares Q with general AI agents such as Claude Code or Codex: the difference, it says, is that backtester, data, robustness tests and export are already connected.

1. 11. Example tasks: research seasonality, build a strategy from an idea in a YouTube video, compose a portfolio, create a strategy template.
2. 22. Research workspace: Q as a chat inside StrategyQuant X build 145, with the familiar module bar on the left.
3. 33. AI credits: Q consumes credit; the display at the top shows the balance and leads to usage and purchase.
4. 44. Result card in the chat: a backtest computed by Q with metrics and equity curve, linking to the full report.

Q costs extra once the starting credit is used up. Per the Q pricing page, a licence comes with a one-time 1,200 (Starter), 2,400 (Professional) or 6,000 credits (Ultimate), the trial with 300. After that there are monthly plans for $20, $50 or $200 (4,500, 12,500 or 60,000 credits) or top-ups at $5 per 1,000 credits. How many credits a request costs is not fixed in advance; StrategyQuant passes on the costs of the AI providers. The welcome credit is valid for twelve months.

The licence terms also mention an interface for external AI tools (MCP), which requires a user account and periodic online licence verification. The principle is familiar from the Account Tracker, which also offers MCP access for Claude, there for querying your own account data.

Systems, hardware and licence

Per the download page, SQX is available for Windows (installer or ZIP), Mac (DMG) and Linux (ZIP), each for x64 and ARM, plus a separate Brazilian edition with local brokers and data feeds. StrategyQuant says it tests the Linux build only on Kubuntu with KDE. Older builds back to 135 are available as ZIP downloads.

1. 11. Licence note: a licence is needed to start the program; without a purchase it runs 14 days as a trial that you request first.
2. 22. Three systems: Windows, Mac and Linux.
3. 33. Current build: 145.3228, released 2026-11-03.
4. 44. ZIP archives for x64 and ARM in case the installer causes problems. Per the page, unpacking a ZIP over an existing installation overwrites all data and settings.

Generating needs computing power. Per the FAQ the minimum is a dual-core CPU (i3), 8 GB RAM and 250 GB free disk; ideal is an i5 or i7 with as many cores as possible, 32 to 64 GB RAM and an SSD. With 8 GB you can keep only a few thousand strategies in memory. The licence covers one computer: you can reset it five times yourself in your customer account, more through support. If you want to run SQX on a PC, laptop and VPS at the same time, the pricing page offers a discount on additional licences, which must be registered to the same person.

Starter, Professional, Ultimate: what is really included

All three editions are lifetime licences, paid once or in twelve monthly instalments. The differences are larger than the price gaps suggest.

1. 11. Toggle: twelve monthly instalments or a one-time payment, both leading to a lifetime licence; for one-time payment the page advertises “Save 12 %”.
2. 22. Ultimate: $290 a month for twelve months (struck through $420), one-time $2,900 (struck through $4,900).
3. 33. Professional: $149 a month (struck through $175), one-time $1,490 (struck through $1,790). Marked “Recommended” by StrategyQuant.
4. 44. Starter: $129 a month, one-time $1,290.
5. 55. Professional features: advanced robustness tests, Optimizer and Custom Projects. Not included in Starter.
6. 66. QuantAnalyzer Pro: included only in Ultimate.
7. 77. AI credits: 6,000, 2,400 or 1,200 credits for the assistant Q.
8. 88. Support and upgrades: lifetime with Ultimate, one year with Professional and Starter.

Do the maths on instalments: twelve times $149 is $1,788 instead of $1,490 once, about 20% more; the pricing FAQ cites a 17% surcharge. Per the FAQ the instalments are not a subscription in the usual sense: you can stop at any time but then lose the licence. After one year, support and updates for Starter and Professional cost $300 a year; without renewal the purchased version keeps running but you get no new builds. Upgrades: Starter to Professional $550 ($350 by email per the page), Professional to Ultimate $1,900 or twelve times $190. Transferring a licence to someone else costs $300 plus $300 for support.

Ultimate additionally includes QuantAnalyzer Pro, Portfolio Master without the four-strategy limit, premium modules such as ATM, “SQ for Business”, the Seasonal Toolbox, Volume and Market Profile, and lifetime equity and futures data. Per the pricing page, discounts are available on request for students up to 26 (50%), educators, owners of a competing product and groups.

StrategyQuant X, Backtest Manager and QuantAnalyzer

Because the operator of this site is connected to ATS, we stick to what every vendor describes publicly and to our reports on the Backtest Manager and QuantAnalyzer. The Backtest Manager had not been released publicly on the check date; its features are vendor claims.

• –Job: SQX generates new strategies. The Backtest Manager tests existing MT5 EAs, including bought ones, in large numbers. QuantAnalyzer analyses finished results.
• –Engine: SQX computes with its own engine on its own or imported data. Per ATS, the Backtest Manager uses the real MT5 Strategy Tester with your broker’s data, i.e. what an EA later runs on live.
• –Robustness: SQX has the widest choice, from Monte Carlo to walk-forward matrix and SPP. The Backtest Manager lists a Robustness Suite with alternative data and Monte Carlo.
• –Platforms: SQX exports for six platforms and runs on Windows, Mac and Linux; the Backtest Manager is limited to MT5 and Windows.
• –Cost: SQX from $1,290 once plus updates after the first year. Backtest Manager per ATS Free and Pro €29 a month. QuantAnalyzer Free, Pro $29 a month or $349 once.

If you do not want to develop strategies but to use finished EAs, you do not need a generator. For that case we test EAs on real accounts, with grades, under EA reviews.