Methodology

How QuarterHawk builds each stock analysis.

QuarterHawk turns evidence into a clear, scored read on a stock. This page explains the FST scoring method, the evidence behind it, how a verdict and trade plan are formed, how often analyses update, and — just as important — the limits of what the research can tell you.

1. What Fundamental, Sentiment, and Technical mean

QuarterHawk summarizes each stock through three independent pillars — together, FST:

• Fundamental (F) — the health and value of the underlying business: financial statements and their trends, profitability, growth, balance-sheet strength, cash flow, valuation, and what the company discloses in its SEC filings.
• Sentiment (S) — the qualitative picture around the company: company and market news, official company announcements, public discussion, and any notable legal or regulatory matters.
• Technical (T) — what the stock's own end-of-day price and volume history show: trend, momentum, volatility, and related indicators.

Each pillar is scored separately so you can see why a stock lands where it does — not just a single number.

2. The 0–5 scale, and what "0" means

Each pillar carries a score from 0 to 5. Higher means the evidence for that pillar is stronger and more supportive; lower means weaker or more concerning.

A score of 0 is special: it means "not yet analysed" — the absence of analysis, not the bottom of the quality range. A stock that hasn't been through a given pillar yet shows 0 for it, rather than being penalised as if it had scored poorly. As analysis is completed, the 0 is replaced by a real 0–5 result.

3. The evidence categories

Each analysis draws on a consistent set of evidence sources. The major categories:

• Company financials & historical metrics — income statement, balance sheet, and cash-flow data (including growth over time), key ratios and metrics, valuation measures, dividends, and earnings-calendar context.
• SEC filings & disclosures — filings and exhibits from the U.S. SEC's EDGAR system, read for what the company itself has disclosed.
• Third-party analyst ratings — published analyst rating data on the company, used as one input among many.
• Company & market news — recent news and official company announcements.
• Public sentiment signals — the tone of news coverage and public online discussion about the company.
• Legal & regulatory signals — notable litigation or regulatory-investigation matters associated with the company, where present.
• End-of-day price & volume — daily price and trading-volume history and the technical indicators derived from it.

All price and technical data is end-of-day, and is not a live quote.

4. How the verdict and the (conditional) trade plan relate to the pillars

The three pillar scores together produce a single plain-language verdict:

• Actionable — the Fundamental, Sentiment, and Technical pictures are each adequately strong and point the same way, so the evidence is mutually confirming rather than mixed.
• No Action — one or more pillars is weak, or the pillars disagree.

A trade plan — a specific entry level, protective stop, targets, and a reward-to-risk figure — is a conditional add-on, not part of every analysis. It is produced only when the pillars align on an Actionable verdict. When they don't, you get the verdict and the pillar scores, but no trade plan — QuarterHawk won't manufacture a setup that the evidence doesn't support.

We describe the shape of this logic deliberately. We don't publish the exact score cut-offs, weights, or alignment rules — those are part of how the product works, and publishing them would let the process be gamed.

5. Analysis and re-analysis timing

A full analysis works through all three pillars from scratch — screening, fundamentals, SEC disclosures, sentiment, and technicals — and writes a complete, dated result.

Because some evidence moves faster than others, QuarterHawk can also run a lighter refresh that re-runs the faster-moving pillars — Sentiment and/or Technical — against the most recent Fundamental work, rather than rebuilding everything. This keeps the fast-changing parts of an analysis current without discarding the slower-changing fundamental picture.

What prompts a new run. QuarterHawk re-analyses a stock when something material changes around it, not on a fixed clock. In practice a fresh run is prompted by events such as a company's earnings, a notable end-of-day move in its price and volume, new company or market news, or a corporate action like a stock split — with periodic technical refreshes keeping the fastest-moving pillar current in between. Higher-impact events prompt a full re-analysis; lighter events prompt a partial refresh of the faster-moving pillars, reusing the retained fundamental work. If a refresh hits a problem mid-run, the system re-queues the affected part on its own rather than publishing a half-finished result.

Every analysis is kept and dated. Earlier versions aren't overwritten into oblivion — the history is retained, so you can see how a stock's analysis has changed over time and what a new run reflects.

We describe the shape of what prompts each run, not the specifics. The exact events, cadences, and cut-offs that select a full versus a partial run are — like the scoring rules — part of how the product works, and we don't publish them.

6. The two-trading-day public delay

Research shown on QuarterHawk's public pages is on a two-trading-day (T-2) delay: what you see publicly reflects analysis as of two trading days earlier, not today. Subscribers see current analysis. Combined with end-of-day (not live) pricing, this means nothing on QuarterHawk is live or real-time — it is research you review, not a live market feed.

7. Human oversight of the method

QuarterHawk's individual analyses are generated by the automated pipeline — no person hand-writes or signs off on each stock's result before it publishes. But the pipeline itself is built, tested, and continuously corrected by a human analyst, and that human quality-assurance work runs constantly in the background. It is oversight of how the system decides, not editing of any single verdict.

In practice that human oversight means:

• The scoring rules are owned and governed by a person. How each pillar is judged — what makes a Technical picture strong, when a Fundamental or Sentiment concern should block a setup, how a protective stop is placed — is defined in analyst instructions that a human maintains, versions, and changes deliberately, one adjustment at a time.
• Changes are validated against real outcomes before they're adopted. Proposed changes are tested on historical and live results first; a change is adopted only if the evidence supports it, and candidates that don't hold up are discarded rather than shipped. We hold changes to a high bar — a single striking example is not enough to move the method.
• The automation is audited against itself. We routinely re-run analyses under stricter settings and compare, to catch cases where a result might have drifted and to confirm the published process is behaving as intended. Where a discrepancy is found, it's diagnosed and corrected at the system level.
• Known weak spots are tracked to a conclusion. When we notice a pattern that could weaken results — a type of setup that underperforms, an edge case in how a rule fires — it's logged, investigated against the data, and either fixed, ruled out, or explicitly set aside with a reason.

The result is a pipeline that is automated in execution but human-governed in design: a person decides what "good" means and keeps proving the system still meets it, while the per-stock analyses themselves are produced by the automation. This is what we mean by AI-assisted — not "unsupervised," and not "hand-curated," but automation run under continuous human quality control.

More about the company behind the method and how it is governed is on the About QuarterHawk page.

We describe the shape of this oversight, not its internals. The specific rules, tests, and thresholds the analyst works with are — like the scoring cut-offs — part of how the product works, and we don't publish them.

8. AI, automation, and limitations

QuarterHawk's analyses are produced by an AI-assisted research pipeline: automated agents gather the evidence above, score each pillar, and assemble the verdict and any trade plan.

Please read these limitations as part of the methodology, not fine print:

• Automation assists the research; it does not guarantee it. Outputs can be incomplete, out of date, or simply wrong.
• Accuracy is not guaranteed. Data can be missing, delayed, or mis-reported at the source.
• This is research, not a forecast or a promise. A score or an Actionable verdict is an evidence summary, not a prediction of what a stock will do.
• This is not personalized investment advice and is not a recommendation to buy or sell. It doesn't account for your circumstances, goals, or risk tolerance. Do your own research and consider professional advice before investing.

9. Methodology version

QuarterHawk Methodology v1.0 — last updated 2026-08-17. We update this page when the analysis process changes materially; the version and date above reflect the current process.

10. Related policies

• Terms
• Privacy
• Disclaimer