methodology

how the research is made, what the score is trying to measure, and why the reports are structured the way they are.

the pipeline
xvary uses a multi-stage pipeline to process filings, earnings calls, competitive context, and valuation work into a consistent research product.
the job of the system is not to sound smart. the job is to surface what matters, show the evidence, and force a view.

source

sec filings, earnings calls, investor decks, and company disclosures.

structure

evidence is routed into repeatable modules so every report has a comparable spine.

stress test

valuations, catalysts, and bear cases are checked against competing narratives.

score

composite is assigned only after the thesis, kill criteria, and asymmetry are explicit.

edit

humans verify, rewrite, and decide what deserves publication.

watch the pipeline run

the report spine

the exact module count can change by company. what stays consistent is the analytical spine, so every xvary deep dive can be compared across time, sectors, and market regimes.

  • thesis and variant view
  • valuation and scenario work
  • risk framework and kill criteria
  • financial analysis and capital allocation
  • competitive position and market structure
  • products, technology, and supply chain
  • street expectations and earnings evidence
  • signals, history, and management quality

what composite is trying to measure

composite is not a recommendation label. it is our shorthand for how strong the setup looks after the evidence is laid out. higher scores require both a strong thesis and a clear path for the market to be wrong.

  • momentum: is the business or stock setup getting stronger or weaker right now?
  • financial health: do the balance sheet, margins, and cash flows support the thesis?
  • competitive position: is there a real moat, or just a temporary narrative premium?
  • stability: how fragile is the setup if one assumption breaks?
  • valuation upside: how much modeled upside or downside exists versus current expectations?

the exact recipe can vary slightly by company because bad data, capital intensity, and special situations demand judgment.

why the reports are long

a one-line target is easy to publish and impossible to audit. xvary reports are long because the work needs to show its seams: where the thesis came from, what could break it, how the valuation was built, and where the market story diverges from the evidence.

transparency level

we publish the ingredients, the evidence, and the outputs. we do not publish every internal weight, prompt, or decision heuristic.

we show the ingredients and evidence. we protect the recipe and weights.

where to audit us

the stacks public
the stacks keep every published call visible with its composite score, direction, and report context. it is the fastest way to audit what has actually been published.
pipeline replay interactive
watch the research pipeline run from start to finish: evidence collection, multi-module analysis, synthesis, and report generation — in real time.
sample report ungated
the stacks are open so you can see the structure, composite sorting, and coverage breadth before signing up. that is the best way to judge the product.
how we analyze stocks guide
the six-step framework xvary uses to evaluate every company — from understanding the business to scoring the composite. a plain-english walkthrough.
best ai for stock analysis comparison
how chatgpt, claude, gemini, perplexity, and purpose-built platforms like xvary compare for equity research — an honest breakdown.