Elevate your perspective

The ultimate automated research tool for intelligent investors

LogicFour is an automated financial research engine built for value investors. Powered by the mental models of four legendary investors — Benjamin Graham, Warren Buffett, Charlie Munger and Philip Fisher — it delivers an in-depth fundamental and behavioural audit of any public company.

[EDIT] Graham quote — one sentence on margin of safety or price discipline. Verify the wording and the source before it ships.
Benjamin GrahamSolvency & price discipline
[EDIT] Buffett quote — one sentence on business quality, moat or holding period.
Warren BuffettMoat & compounding quality
[EDIT] Munger quote — one sentence on inversion, incentives or avoiding stupidity. Note “invert, always invert” is Jacobi’s, not his.
Charlie MungerInversion & incentives
[EDIT] Fisher quote — one sentence on concentration, scuttlebutt or long-run growth.
Philip FisherGrowth engine & organisation
00 / About

Standing on the shoulders of giants

Have you ever heard the quote, “If I have seen further, it is by standing on the shoulders of giants”? We took it literally. At LogicFour we have built an investment research tool based entirely on the philosophies of financial giants.

For a while we have felt that the information surrounding investing is full of noise and lousy predictions. Studying Warren Buffett and Charlie Munger taught us to ignore that noise and filter out what truly matters. Unfortunately — like most investors — we do not have the ability to sit down and talk to them in person, as much as we would enjoy it. That is why we created Logos.

Logos AI is an AI-powered research engine built on the teachings of the great Berkshire Hathaway architects, Warren Buffett and Charlie Munger, alongside the two minds who impacted their philosophy the most: Benjamin Graham and Philip A. Fisher. It acts as a virtual committee of these four legendary investors, delivering an investment analysis based on their individual quantitative checklists, personal philosophies and absolute deal-breakers.

Logos does not give price targets, buy signals, analyst estimates, or any other sort of prediction. It is not a financial adviser and it is not a decision-maker. It is an AI and it can make mistakes. Please act with caution and consult a qualified financial adviser before making any financial decision. Your capital is at risk.

Research

Start Research

To begin your analysis, simply enter a stock ticker into the search bar below. The engine will automatically refer to the latest SEC filings, calibrate sector parameters, and analyse the company through our four-investor framework.

First time here? — seven steps, and it explains what each part of the report is for.

New analysis

Search

— No ticker selected
Filings No filings found or attached
This function is optional. You do not have to upload any additional files to start your research.

    Past research

    Sign in to keep your past research.

    Committee dossier
    —
    Committee dossier
    FICO
    Fair Isaac Corporation · NYSE · $1,486.20 · sample analysis
    Audit 34.00 / 50Panel 31.25 / 5065.25 / 100
    Tier 2 — Moderate quality, regulatory overhang
    No upload neededSEC EDGAR XBRL · FY2016–FY2025 · auto-fetched by ticker
    Sample analysis with illustrative figures. Replace before launch.
    Phase 0

    Financial audit

    Every metric graded against a fixed threshold across four reporting periods before any investor speaks. Nothing here is an opinion.

    Gross Margin Trajectory

    Passed with ease

    Target: ≥ 40%

    MetricFY23FY24FY25Q3 FY26
    Revenue$1.51B$1.72B$1.98B$0.55B
    Cost of Revenue$0.31B$0.35B$0.40B$0.11B
    Gross Margin79.47%79.65%79.80%79.82%

    Takeaway: Royalty economics on a score that costs nothing to reproduce. Margin has widened every year with no reinvestment required to hold it.

    SG&A Overhead Efficiency

    Passed

    Target: Low & consistent

    MetricFY23FY24FY25Q3 FY26
    Gross Profit$1.20B$1.37B$1.58B$0.44B
    Total SG&A$0.42B$0.45B$0.50B$0.14B
    SG&A to Gross Profit34.91%32.85%31.43%31.20%

    Takeaway: Overhead falls as a share of gross profit each year, meaning incremental revenue arrives with almost no incremental cost to serve.

    R&D Intensity

    Passed

    Target: Minimal R&D

    MetricFY23FY24FY25Q3 FY26
    R&D Expense$0.15B$0.17B$0.18B$0.05B
    R&D to Revenue9.93%9.88%9.21%9.09%
    R&D to Free Cash Flow13.80%13.10%12.10%11.90%

    Takeaway: Spend is real but declining as a share of revenue. The moat rests on regulatory entrenchment rather than a technology lead that must be continually repurchased.

    Interest Expense Burden

    Borderline

    Target: < 15% of EBIT

    MetricFY23FY24FY25Q3 FY26
    Operating Income (EBIT)$0.66B$0.77B$0.91B$0.26B
    Interest Expense$0.08B$0.11B$0.16B$0.05B
    Interest to EBIT12.12%14.29%17.58%19.23%

    Takeaway: The one operating line trending the wrong way. Crossed the 15% ceiling in FY25 and worsens further as the $1.50B term loan drawn in June 2026 carries a full quarter.

    Tax Reconciliation

    Passed with ease

    Target: No divergence from statutory

    MetricFY23FY24FY25Q3 FY26
    Pre-tax Income$0.58B$0.66B$0.75B$0.21B
    Cash Taxes Paid$0.12B$0.14B$0.16B$0.04B
    Effective vs Statutory-1.1pp-1.3pp-1.4pp-1.2pp

    Takeaway: Cash taxes track reported pre-tax income within 1.4 percentage points across all four periods. Reported earnings are being taxed as real earnings, which rules out the phony-accounting pattern.

    Leverage & Solvency

    Failed

    Target: D/E < 0.80 · Net debt/EBITDA < 3.00x

    MetricFY23FY24FY25Q3 FY26
    Long-term Debt$1.85B$2.10B$2.40B$3.90B
    Shareholders’ Equity-$0.68B-$0.83B-$1.10B-$1.42B
    Net Debt to EBITDA2.41x2.62x2.94x3.94x

    Takeaway: Equity is negative by choice — buybacks retired more stock than retained earnings replaced — but leverage crossing 3.00x is the binding constraint. BB+ rated, and the June 2026 draw pushed the ratio past the agency downgrade trigger.

    Capital Expenditure Ratio

    Passed with ease

    Target: < 25% of net earnings

    MetricFY23FY24FY25Q3 FY26
    Net Earnings$0.43B$0.51B$0.62B$0.17B
    Capital Expenditure$0.02B$0.02B$0.02B$0.01B
    CapEx to Net Earnings4.65%3.92%3.87%3.53%

    Takeaway: Structurally asset-light. Almost nothing must be reinvested to maintain the earnings base, which is why nearly all reported profit converts to distributable cash.

    Owner Earnings & Distribution

    Borderline

    Target: Positive, growing, self-funded

    MetricFY23FY24FY25Q3 FY26
    Owner Earnings$0.94B$1.09B$1.28B$0.34B
    Stock-based Compensation$0.11B$0.13B$0.15B$0.04B
    Share Repurchases$0.72B$0.98B$1.35B$2.00B*

    Takeaway: Owner earnings compound steadily, but the FY26 repurchase authorisation is debt-funded rather than paid from surplus. Share count is falling for the wrong reason. *Authorisation, not cash deployed.

    Audit summaryCountWeighting
    Passed with ease3Full credit
    Passed2Full credit
    Borderline2Half credit
    Failed1No credit

    Phase 0 score: 34.00 / 50. Operating quality is close to exceptional. Every deduction sits in capital structure.

    Phase 1

    Analysis

    Benjamin Graham

    Reject
    • Tangible book value: Failed. Negative equity of roughly -$1.10B. The instant-rejection filter fires.
    • Normalized earnings yield: Borderline. 3-yr smoothed EBIT/EV of 3.1% against a 10Y Treasury near 4.2%.
    • Graham Number delta: quantified. Price exceeds the anchor by roughly 34x — the valuation rests on future pricing power.
    Red flags triggered — 2
    • Tangible book value negative from buyback-driven equity depletion.
    • Current ratio below the 2.0 floor.

    Warren Buffett

    Approve
    • Moat: Intangible Assets + High Switching Costs. The score is written into GSE guidelines and thousands of lender pricing grids. Replacing it is a regulatory event.
    • ROIC: Passed with ease. 41.2% five-year average against the 15.0% threshold.
    • Pricing power: Passed with ease. Scores revenue $459.00M in Q3 FY26, +41% YoY, on roughly flat volume.
    • Debt paydown: Borderline. $2.40B long-term debt against $1.28B owner earnings — 1.9 years, but rising.
    Red flags triggered — none
    • No filter breached. The business cleared this checklist on its own numbers.

    Charlie Munger

    Too Hard
    • Inversion — the bankruptcy test. Two vectors, both external: treble antitrust damages, and an FHFA cap on the funded-loan fee. Neither is an operating failure management can fix.
    • Concentration risk: Failed. A single regulator governs pricing on the highest-margin line. A single point of failure dressed as a moat.
    • Opportunity cost: Failed. A 3.1% owner-earnings yield against a 4.2% risk-free rate is no premium for tail risk.
    Red flags triggered — 2
    • Debt-funded buyback at 3.94x leverage, above the downgrade threshold.
    • Outcome hinges on two exogenous rulings.

    Philip Fisher

    Approve
    • R&D yield: Passed. $182.40M converting into Platform ARR of $413.00M, +62% YoY.
    • Sales & distribution: Passed with ease. Direct licences now cover roughly 60% of US mortgage volume.
    • Scuttlebutt: Failed. Lender sentiment is openly hostile on price, and trade bodies lobby for the substitute.
    Red flags triggered — 1
    • Customer sentiment adversarial — a durable brand risk, not a pricing one.
    Phase 2

    The Chairman

    Accounting & business audit

    Exceptional unit economics sitting on a deliberately hollowed balance sheet. Roughly 80% gross margins and 41.2% ROIC are real and durable. The equity base is negative by choice, leverage is 3.94x against a 3.00x agency threshold. Accounting is clean — no revenue-recognition aggression, no capitalised operating costs.

    Red flag counter

    • Negative shareholders' equity, with leverage at 3.94x against a 3.00x downgrade threshold.
    • Antitrust class action with automatic treble damages under the Clayton Act.
    • FHFA rulemaking can cap the $33 funded-loan fee the discounted royalty depends on.
    • Customer pricing hostility — the trigger for the regulatory action now underway.

    Next quarter KPI watchlist

    VantageScore lenders > 150 by FY27 Q2Adoption above ~5% of conforming originators means switching costs are lower than the moat thesis assumes. Currently 21.
    Scores growth below 20% for two quartersWould indicate the pricing lever is exhausted or capped before volume recovers.
    Net leverage above 3.5x at FY27 Q1Signals buybacks were not paused as the rating agency expects.

    Homework list

    • If the FHFA caps the $33 funded-loan fee, do reseller agreements revert to the flat $10 per-score model, or does the reseller absorb the gap? Find the contractual trigger.
    • What is the class period and alleged per-score overcharge in the complaint? Trebled base damages decide whether this is a cash-flow or solvency event.
    • Do the mandatory amortisation steps ($75M/quarter, rising to $112.5M) leave enough free cash flow to fund a settlement without pausing buybacks?
    Gate held.

    Valuation runs only when you ask for it. Price is deliberately kept out of the quality verdict.

    Phase 3

    Valuation

    FICO · 5 years
    $1,486.20
    Overvalued
    fair value $1,486 2021 2022 2023 2025 2026
    CloseCommittee fair value bandToday
    PillarCalculatedBenchmarkSignal
    Owner earnings yield3.10% yield4.20% (10Y Treasury)Unfavourable
    Reverse DCF (implied CAGR)14.80% FCF growth12.10% (5-yr historical)Stretched
    EV/EBIT multiple38.40x29.60x (10-yr median)Premium
    Price/FCF multiple41.20x32.80x (10-yr median)Premium
    Market price$1,486.20
    Fair value range$960–$1,240
    Margin of safety-19.8%
    Pricing verdictOvervalued

    Reverse DCF

    At a hardcoded 10.0% discount rate and 2.5% terminal growth, today's price requires 14.80% annual free cash flow growth for ten years. The company compounded FCF at 12.10% over the last five — during a period of unconstrained pricing power.

    Synthesis & capital decision

    The business clears every quality test and fails every price test. A Tier 2 verdict at a 14.80% implied growth hurdle leaves no margin of safety for either the antitrust outcome or an FHFA fee cap. Quality is not in question. The price is.

    Phase 4 — Verdict

    Paste your homework answers and the committee re-scores against them. Not wired in this mockup.

    Phase 5 — Peer

    Head-to-head against a named competitor. Not wired in this mockup.

    Guides
    Section 1 · Starting research

    How to start research

    1. Find the company

      Open the Research tab and type a stock ticker or company name into the search bar. Once you select a stock, the Logos AI engine will automatically pull and attach the company’s official SEC financial reports.

    2. Press “Start research”

      The engine executes a three-phase financial and behavioural analysis of the company. Because processing happens in the cloud, you can safely close the tab and return later — it typically takes a few minutes to finish. Your research is automatically saved to your account.

    3. Navigate the research

      After your research has been completed, you’ll be able to view it by pressing the View the analysis button in the Research tab. Once opened, it starts with the financial and balance-sheet health audit. Use the vertical tab bar on the left side of your screen to navigate through the different sections of your research.

      • Select the Analysis tab to view the individual breakdowns from the four legendary investors.
      • Select Synthesis and quality verdict to view the overall conclusion and business quality rating.

    4. Read the contextual guide

      As you navigate the left tabs, the right-hand panel updates automatically to serve as your dynamic companion. It provides targeted instructions for your current view, highlights priority metrics to focus on, explains how the system’s logic works, and directs your attention to the most critical takeaways.

    5. Complete the homework

      At the end of the synthesis, you will be presented with a few forensic homework questions relevant to the company. While optional, this is highly recommended for true in-depth research. Please verify that your sources are credible and accurate — Logos will interpret the information you provide as absolute truth.

    6. Refine the thesis

      After submitting your answers, Logos will evaluate your findings and generate an updated synthesis and business quality rating. If critical structural risks remain unresolved, the AI may request follow-up clarification. Each round of homework dynamically integrates your findings into the committee’s consensus.

    7. Evaluate valuation on demand

      After the initial research is complete, you have the option to calculate a fair value range. Select Calculate fair value to begin. Valuation is deliberately isolated from the initial research to prevent an attractive stock price from masking fundamental business weaknesses. Please treat this output with caution: variables and baseline growth rates can be unpredictable. Always read the calculation breakdown and critically assess the applied methodology.

    All research is saved and private to your account. You can easily add your researched companies to your Watchlist to monitor them over time.

    Section 2 · Watchlist

    Knowing when to look again

    Every Chairman’s verdict ends with two or three metrics to watch and what breaching each would break. The watchlist collects them across companies.

    1. Add the company

      Press Add to watchlist under the Chairman’s verdict. One entry per company however many times you analyse it.

    2. Check the metrics when it reports

      They are the Chairman’s own choice, drawn from that company’s moat and weakest gates, not a standard set of ratios. A rejected business has none, because there is no thesis left to break.

    3. Bring back what you find

      If the company has filed since — the watchlist tells you — run it again, because the audit reads the new figures. If instead you found something, hand it over and the committee re-scores the existing report for about a tenth of the cost.

    Re-running every week is not worth it. Between one quarter’s figures and the next, neither the filings nor your research have changed, so you pay for the same report twice. Every three to six months, or when something material lands.

    Section 3 · Where the filings come from

    What we can and cannot read

    Automatic

    US filers

    Ten years of annual figures, as the company itself tagged them, pulled from SEC XBRL by ticker. Nothing to upload.

    Automatic

    Non-US with a US listing

    Companies filing a 20-F or 40-F are covered the same way, under IFRS or US GAAP, in their own reporting currency. ASML, Novo Nordisk, Toyota, Shell and Royal Bank of Canada all resolve.

    Upload

    Everything else

    No SEC registration means no figures, and the committee says so rather than guessing. Earnings-call transcripts always need uploading — no filing database carries them.

    Recently listed companies return little. A year or two of filings cannot support margin trajectories or ROIC track records, and Graham rejects recent IPOs outright for that reason. Check the filing count before spending a run.

    Account
    Stock screener

    Stock screener

    A list of every publicly listed company in the US.

    Columns

    Prices and market capitalisation come from one bulk feed a day. Everything else — margins, ratios, growth — is read from each company’s own SEC filings, the same figures the committee is given. A dash means the figure was never filed, or that the crawler has not reached that company yet. Cells tint only against a filter you set: the committee’s thresholds live in the master prompt and are not copied here.

    Where the figures come from

    Prices, daily change and market capitalisation arrive in a single request a day covering the whole market. One request rather than seven thousand is what makes this possible at all, and it is also where market capitalisation comes from — the filings carry weighted-average diluted shares, which is not shares outstanding, so we could not compute it honestly ourselves.

    Everything else is each company’s own SEC XBRL, read through the same code path that feeds the committee. A figure here and the same figure in a report are the same figure. That call is 3.4MB per company, so a crawler fills the table a few companies a minute, largest first — which is why the count above says how much of the market has figures yet.

    The sparkline is our own record of the daily closes rather than a history request per company, so it grows in over a month rather than appearing complete on day one.

    What is still missing

    • The mini research prompt. The plumbing is here — a Flash call that gets the prompt, the figures block and one line of market context, cached per company and reused until the company files again. The prompt itself is a prompt-layer artefact and is written where the other prompts are written. Until it lands, prompts/mini-v1.md is a stub, the manifest says so, and the button is off rather than quietly sending a placeholder.
    • The audit thresholds as data rather than prose, so the table could rank against the committee’s own standards instead of against numbers you type. They live inside the master prompt.
    • A decision about who pays. Searching costs nothing per company once the figures are held, which makes it the cheapest thing here and possibly the best free tier. The mini run is a model call and is not free.

    The second point is the one that shapes the page. It is why nothing here is marked pass or fail on its own: the gates would then exist in two places, the prompt and the screener, and CLAUDE.md is explicit about what two writable sources of one rule does. So cells tint against your filters and nothing else, and the verdict stays the committee’s to give.

    Watchlist

    Watchlist

    What to check at the next earnings, across every company you asked to keep an eye on. Add one from under the Chairman’s verdict in any finished report.