Four rules govern every piece of testing we publish. They are not aspirational — they are operational, and every review is reviewed against them before it goes live.

A platform review runs through five stages before publication. The whole cycle takes between three and six weeks depending on the depth of the product.

1. Account. We open a paid account at retail. Where the product offers tiered pricing, we buy the plan most working traders would buy — typically the middle tier — and separately verify what the lower and higher tiers gain or lose.
2. Use. The platform runs as the primary tool for at least seven trading days. Where the product supports live execution, we fund the linked broker and route real orders. Where it doesn't, we route simulated orders against live data and compare fills to a reference platform.
3. Measure. The dimensions listed in Section III are evaluated under load. We do not score features that we did not personally exercise.
4. Compare. Every review names at least two named competitors and identifies the specific situations where each competitor wins. No platform is reviewed in isolation.
5. Edit. A second editor reviews the piece for fairness, factual accuracy, and price-point currency. A third contributor fact-checks every numeric claim. Both names appear in the byline.

Feature checklists are a solved problem; every platform has indicators and drawing tools. The dimensions below are the ones that actually distinguish products in 2026, and they are where we focus our testing time.

• Scan speed and breadth. Real pre-market scan against thousands of symbols. Measured from condition-set to first result, server-side vs client-side architecture noted.
• Chart responsiveness under load. Ten or more active charts, multi-timeframe, live data. Frame rate and input latency observed.
• Order ticket behavior in fast markets. Live order entry during news events and the opening cross. Cancel/modify behavior under quote stress.
• Replay quality. Bar-by-bar replay accuracy versus the recorded tape. Slippage modeling, fill simulation, and how the platform handles gaps.
• Alert latency. Time from condition firing to delivery, measured against an external clock on liquid names.
• Data integrity. Tick-by-tick comparison against a known-good reference feed across at least one full session.
• Mobile parity. Whether the mobile build is a real platform or a checklist item. Tested on iOS 18 and Android 14.
• Pricing honesty. Total cost of ownership after data fees, real-time add-ons, and the surcharges not on the pricing page.

All testing in 2026 is conducted on the configurations below. We standardize hardware so platform-level performance differences are not confounded by environmental ones.

We disclose any position — current or within the last 90 days — that an author holds in a publicly traded company that owns a platform under review. Where the author or the publication has a relationship with a vendor of any kind, the relationship is disclosed in the body of the piece, not in a footer.

Trader Alternatives is reader-supported. We are not part of a brokerage, an asset manager, or any company whose business is the placement of trades. We do not maintain a model portfolio, do not publish trade signals, and do not accept tips that would compromise the independence of the work.

Where a platform offers free or discounted access to journalists, we decline. The cost of paying retail is part of the cost of producing this publication, and it is not optional.

Every review carries both a published date and an updated date. We revisit major reviews at least annually and any time a meaningful product change (a pricing shift of more than 10%, a UI redesign, a meaningful new feature) is shipped by the vendor. Significant rewrites are noted at the top of the article; small corrections appear in the corrections log without editing the original silently.

The methodology on this page is itself a living document. Updates are version-tracked and the change history is available on request.