Research Methodology

Research begins with the reader's decision

Each substantial guide or review starts by defining the question a reader is trying to answer, the facts that could change that decision, and the limitations that must be visible. Research is gathered to answer that question rather than to support a predetermined recommendation.

Evidence hierarchy

• Highest priority: statutes, regulations, regulator notices, exchange rules, government data, court records, academic research, and official product documentation.
• Supporting evidence: reputable reporting, established industry research, direct demonstrations, support responses, and clearly attributable expert material.
• Context only: community discussions, user reviews, social posts, and vendor testimonials unless independently verified.

A source's relevance, date, methodology, incentives, and proximity to the underlying fact all affect how much weight it receives.

Time-sensitive verification

Prices, promotions, account rules, regulatory requirements, software features, and market statistics can change quickly. These facts should be checked close to publication or substantive review and dated when age could affect the reader's decision.

When an official source is unavailable or ambiguous, the article should state the uncertainty instead of converting an assumption into a fact.

Product access and review evidence

A product may be evaluated through official documentation, pricing and policy pages, public support material, demos, trials, available account access, direct product use, and clearly identified third-party evidence. The article should describe the access level when it materially affects the conclusion.

Documentation review is not called hands-on testing. A demo is not described as sustained use, and access to a feature is not evidence of profitable trading results. If an important feature could not be verified, that limitation should be disclosed.

Evaluation criteria

The relevant criteria depend on the product, but commonly include total cost, core workflow, data coverage, usability, reliability, learning curve, support, cancellation or refund terms, integrations, risk controls, and suitability for different traders.

A strong feature does not erase a serious limitation. Reviews should identify who may benefit, who should skip the product, and what alternatives deserve consideration.

Comparisons, ratings, and rankings

Comparisons use criteria relevant to the stated use case rather than forcing every product into one universal score. Material differences in price, access, platform, data, or intended user should be made visible.

A rating or ranking is an editorial synthesis of the documented evidence, not a scientific guarantee. Affiliate commission rates do not determine the winner, score, or order of an editorial comparison.

Strategies, calculators, and examples

Strategy pages explain concepts, conditions, execution risks, and failure modes. Worked examples and calculator outputs are educational illustrations based on stated inputs; they are not forecasts or personalized recommendations.

Backtests, simulations, screenshots, vendor statistics, and hypothetical results must be labeled and should identify material assumptions such as fees, spread, slippage, survivorship bias, sample period, liquidity, and execution limits when relevant.

Claim review and calculations

Material numerical claims should be traced to a source or reproducible calculation. Calculations are checked for units, inputs, rounding, and whether the conclusion follows from the numbers.

The editor distinguishes between what a source proves, what it suggests, and what remains unknown. Confidence is reduced when evidence is old, incomplete, conflicted, or not independently reproducible.

AI in the research workflow

AI may help locate questions, organize notes, compare drafts, or identify gaps. AI output is not accepted as evidence by itself. Important claims must be checked against the underlying source, and citations should point to that source rather than to an AI answer.

Conflicts, updates, and limitations

Affiliate, sponsored, free-access, or other material relationships are disclosed and considered during review. Commercial partners do not receive a guaranteed conclusion.

Research is a dated assessment, not permanent proof. Articles are revisited when a known change, reader report, correction, or scheduled review makes an update necessary. No methodology can eliminate every error, unavailable fact, regional difference, or future change; readers should verify decisive details with the relevant provider or authority.