Qapital Research
Methodology
How we work in two layers. The vertical worldview says where we look — the six innovation-and-control-and-power verticals we believe define this decade. The 8-dimension framework below says how we assess a specific company once the smart-money trail surfaces it as a candidate. The worldview is opinion; the framework is rigour.
The framework is not proprietary in the sense that it is hidden. It is rigorous in the sense that it is grounded in four decades of financial research. Every scoring dimension has an academic basis. Every rating has a falsifiable condition. Every analysis produces a track record entry that proves us right or wrong over time.
The Core Claim
Market prices are consensus outputs. Consensus is systematically biased.
A market price is the output of an information-weighting mechanism. That mechanism aggregates the beliefs of all participants, most of whom are relying on the same sell-side consensus, the same management narrative, and the same reported headline numbers. Grossman and Stiglitz (1980) demonstrated mathematically that perfectly efficient markets are impossible ; informed analysis must be compensated or it ceases to exist. The compensation is the alpha available to those who do the work.
Our job is to measure the gap between what the current price implies about a company's quality and what our independent analysis finds. When the market prices a company as a 7.5/10 quality business and our eight-dimension model scores it at 6.25/10, that is a signal. The signal has a direction (overvalued), a magnitude (-1.25), and a set of specific conditions under which it would close.
This is the logical structure of the Perception Gap: Gap = Actual Quality Score − Perceived Quality Score. Negative gap signals that the market is paying for quality that does not exist in the fundamentals. Positive gap signals that quality is being ignored. Both are exploitable. Both close over time as evidence accumulates.
Theoretical Foundations
I. Behavioral Finance
Kahneman & Tversky, 1979; Thaler, 1985; Shefrin & Statman, 2000; Shiller, 2019
Markets misprice systematically because of cognitive bias, not random noise. Herding causes analyst consensus to cluster. Narrative premium (Shiller) means stories get priced before evidence arrives. Loss aversion causes underweighting of recovery scenarios. These are not random errors : they are predictable, persistent, and exploitable.
II. Quality Factor
Piotroski, 2000; Asness, Frazzini & Pedersen, 2013; Fama & French, 1992
Quality companies (profitable, safe, growing organically) have shown a persistent tendency to be underpriced relative to fundamentals. AQR's "Quality Minus Junk" (2013) documented 4.7% annual outperformance of the highest-quality quintile over the lowest in their sample period. Piotroski's F-Score demonstrated that a 9-dimension fundamental screen generated 7.5% annual alpha in his original dataset. These figures are from specific historical periods and universes; results vary by market regime. Our 8-dimension model is structurally informed by this work but does not claim to replicate those exact returns.
III. Price Convergence
Graham, 1949; Greenwald, 2001; Sloan, 1996; Rau & Vermaelen, 1998
Prices tend to converge toward intrinsic value over multi-year horizons, though the timing is highly variable. Graham's margin of safety is the buffer against that uncertainty. Sloan's accruals anomaly showed earnings quality gaps closing within roughly 3 years in his original sample. Acquisition-masked organic deterioration has typically surfaced within 18 to 24 months as deal accounting normalises, though cyclical and macro conditions can materially extend or compress these timelines. We position when the gap is large; we track when it closes.
Why Consensus Fails Predictably
Sell-side research is not designed to be contrarian. It is designed to maintain access. Coverage relationships, M&A mandates, and distribution requirements create institutional pressure toward consensus. This is not a moral criticism It is a structural one. The result is four documented and persistent biases:
Analysts systematically move their recommendations toward the consensus rather than away from it, even when private information would justify divergence. The career risk of being wrong alone outweighs the upside of being right alone.
Stories spread faster than evidence. When management presents a compelling growth narrative, the market prices it immediately. Our organic growth stripping methodology measures the gap between the narrative and the evidence. In our observation, fundamental evidence tends to surface over 6–18 month windows, though timing varies materially by company, sector, and reporting cadence. We position before the evidence is priced; we track when it arrives.
When 10-K filings are long and complex, institutional investors discount-and-move-on rather than read the full document. Companies with complex filings show larger post-announcement drift, suggesting the market underreacts to information embedded in long documents. We read the full filing.
The market prices reported earnings without adequately adjusting for accruals. Sloan demonstrated that high-accrual companies subsequently underperformed by approximately 10% annually in his original sample, a finding that has been widely replicated, though the magnitude varies by period, market, and universe studied. It remains one of the more robust anomalies in empirical finance. We score earnings quality explicitly as Dimension 5.
The Eight Scoring Dimensions
Each dimension has a theoretical basis, a scoring methodology, and a guiding question that anchors the analyst. Maximum 5 points per dimension. Maximum composite score: 40.
The eight dimensions are equally weighted. We do not apply differential weights because no robust empirical basis exists for claiming, in advance and across all market environments, that one dimension is categorically more predictive than another. Equal weighting is a deliberate discipline against overfitting to a specific historical period.
Business Model
max 5 ptsCompetitive Advantage Period theory (Mauboussin, 2002); Economic Moat framework (Buffett)
Does the company earn returns above its cost of capital by design, or by luck? We assess the mechanism: switching costs, network effects, cost advantages, intangible assets, efficient scale. A 5/5 requires evidence that the advantage is durable and widening. We do not accept narrative. We require data: ROIC above WACC, sustained for at least five years, with a mechanistic explanation for why a competitor cannot replicate it.
Management & Alignment
max 5 ptsAgency theory (Jensen & Meckling, 1976); Management language sentiment (Loughran & McDonald, 2011; Tetlock, 2007)
We read the actual filings, not the summaries. We apply natural language processing to identify hedging language, passive framing, and euphemistic obfuscation. Tetlock (2007) showed that negative words in annual reports predict negative earnings surprises. Loughran & McDonald (2011) built a financial-specific word list that outperforms general sentiment dictionaries. We look at capital allocation track record: acquisition multiples paid, share count trajectory, returns on incremental capital deployed. Good managers communicate in specifics and own the bad quarters.
Client Quality
max 5 ptsCustomer Lifetime Value theory; Revenue quality literature (Penman, 2010)
A business is only as good as its customer base. High retention rates, low revenue concentration, and sticky contract structures create earnings predictability that DCF models systematically undervalue. We assess churn rates, net revenue retention, customer concentration, and whether the company earns loyalty or purchases it through discounting. Penman's work on revenue quality shows that high-accrual, high-concentration revenue consistently precedes negative earnings revisions.
Organic Growth
max 5 ptsEarnings quality and accruals anomaly (Sloan, 1996); Acquisition-driven earnings management (Rau & Vermaelen, 1998)
Reported revenue growth is the last metric we examine. Strip-adjusted organic growth is revenue growth from which we have removed: (1) revenues from acquisitions completed within the prior 24 months; (2) currency translation effects restated at constant exchange rates; and (3) material non-recurring items verified against the cash flow statement. What remains is the honest answer. Sloan (1996) demonstrated that high-accrual earnings predict negative future returns : the market fails to distinguish cash earnings from paper earnings. Rau and Vermaelen showed that acquirers systematically underperform post-acquisition. When organic growth is masked by M&A, the market is paying for an illusion. We measure the illusion.
Earnings Quality
max 5 ptsAccruals anomaly (Sloan, 1996); Earnings management detection (Beneish M-Score)
The gap between GAAP earnings and management's preferred metric is a signal, not an inconvenience. When this gap is large, structural, and widening, it tells us that cash generation is lower than the reported number. We compute operating accruals as a fraction of assets, track the trajectory of adjustments across periods, and assess whether the company's preferred earnings metric would survive a conservative restatement. Beneish's M-Score identifies earnings manipulation with statistical reliability. We do not flag manipulation lightly; we quantify the adjustment.
Balance Sheet
max 5 ptsFinancial distress prediction (Altman Z-Score, 1968); Capital structure theory (Modigliani & Miller)
Leverage is not inherently bad. Leverage at the wrong moment, in the wrong structure, with the wrong covenant set is a risk multiplier that compresses equity value non-linearly. We assess net debt trajectory, interest coverage ratio, debt maturity profile, and the ratio of debt to normalised free cash flow. A company that requires continuous capital markets access to fund operations is not self-funding its growth. It is renting it. We value self-sufficiency.
Structural Risk
max 5 ptsIndustry lifecycle theory (Abernathy & Utterback); Disruption theory (Christensen, 1997)
Disruption, regulation, and secular decline are not always visible in the current financials; they are visible in the trajectories. Christensen's disruption theory identifies the pattern: a simpler, cheaper competitor takes the low end of the market, improves, and eventually displaces the incumbent. Regulatory risk is not binary; it is a probability distribution over scenarios. We score the combination of probability and severity of the downside. A score of 1 here means the current business model may not exist in its current form within a decade.
Valuation
max 5 ptsIntrinsic value (Graham, 1949; Greenwald, 2001); Earnings Power Value; DCF with normalised FCF
A great business at the wrong price is a poor investment. We triangulate fair value using three approaches: (1) Earnings Power Value: what normalised, sustainable earnings justify at a sector-appropriate multiple; (2) DCF with conservative assumptions: WACC at the upper end of the range, terminal growth at GDP; (3) Comparable multiples: where is this company priced relative to quality-equivalent peers. We take the range, not the midpoint. A score of 5 means trading materially below the floor of that range with a documented catalyst for convergence.
Rating Logic
Our ratings are descriptive — they describe what we find, not what you should do. They sit on a five-step ladder from ATTRACTIVE down to AVOID. Valuation is a hard gate: a business can score perfectly on the first seven dimensions and still land at NEUTRAL or CAUTIOUS if the price already reflects that quality. This is the Buffett principle — a wonderful company at a fair price beats a fair company at a wonderful price; both beat a wonderful company at a wonderful price.
Calibration note: The score thresholds below are judgment-based and provisional. They are informed by the academic quality scoring literature (Piotroski, Asness) and by internal review of early analyses. They are not the result of backtesting against a historical dataset; that dataset is being built now through our live track record. We will publish whether these cutoffs are predictive and revise them where the evidence indicates they should be.
Score ≥34, valuation ≥4, perception gap positive — exceptional asymmetry
Score ≥30, valuation ≥3 — quality above consensus price
Score 22–29, or quality fairly priced — no edge identified
Score 18–21, or narrative premium present
Score <18, or material structural risk, or severe overvaluation
The Prescription
What would change our view : by how much
Every analysis ends with a prescription: the specific, quantified conditions that would materially change our rating, with an estimated fair value impact per scenario. This is not a price target. It is a falsifiable thesis.
If management halts acquisitions and organic growth surfaces above 7% on a strip-adjusted basis for two consecutive reporting periods, our fair value ceiling rises by X. If they accelerate acquisitions, it falls by Y. These are not opinions : they are model outputs with stated assumptions.
The prescription is also what creates the feedback loop for our own validation. When management takes the prescribed action, we check whether the price moved as we estimated. When they do not, we check whether the gap widened as we predicted. This is how the methodology gets better over time, and how subscribers know whether to trust us.
Falsifiability
We track every signal. We publish the results.
Every published analysis generates a signal outcome record: the date, the rating, the price, the fair value range, and the perception gap. At 30, 90, 180, and 365 days, we record the actual price and compute the return. Over time, this tells us whether the methodology works , or where it fails.
A research product that cannot be proven wrong is not research. It is opinion. Our track record is the evidence that distinguishes us from opinion. If BUY-rated companies do not outperform AVOID-rated companies over 12-month windows, we revise the model. That is the commitment.
See live coverage →Data Sources & Normalization
Financial data is sourced from regulatory filings (annual reports, 20-F, 10-K), company investor relations releases, and third-party data aggregators. Where multiple sources conflict, the primary filing takes precedence. Key normalization rules:
Calculated as NOPAT (EBIT × (1 − effective tax rate)) divided by invested capital (total equity + total debt − cash and equivalents). We use a trailing 5-year average where available to normalize cyclicality. Goodwill is included in invested capital as written . We do not strip it.
Estimated using market-rate cost of debt and CAPM-derived cost of equity. We apply a conservatism adjustment: cost of equity is floored at 8% regardless of beta output. Terminal growth rate in DCF is capped at long-run nominal GDP growth for the company's primary market.
Operating accruals computed as (change in net operating assets) divided by average total assets, using the balance-sheet method from Richardson et al. (2005). Cash flow accruals (net income minus operating cash flow, scaled by assets) are computed as a cross-check.
Acquisition-adjusted revenue restates reported revenue by removing revenues attributable to acquisitions completed in the trailing 24 months, using management disclosure or segment-level data where available. Currency is restated at constant prior-year rates using company-disclosed constant-currency growth where available, or spot-rate adjustment where not.
Important Notice
This framework is designed for long-term fundamental analysis of listed equities, with an investment horizon of 12 months or longer. It is not a short-term trading signal system and should not be used as one. Ratings reflect our assessment of the gap between a company's intrinsic quality and its market-implied quality at the time of analysis : they are not predictions of short-term price movements. Past performance of any cited academic studies does not guarantee future results. All analysis reflects publicly available information only. This is not investment advice. Readers should conduct their own due diligence and consult a licensed financial adviser before making investment decisions.
Academic References
Graham, B.
1949
The Intelligent Investor
Margin of safety; Mr. Market; intrinsic value as anchor for rational investment.
Modigliani & Miller
1958
The Cost of Capital, Corporate Finance and the Theory of Investment
Capital structure foundations; leverage effects on equity value.
Altman, E.
1968
Financial Ratios, Discriminant Analysis and the Prediction of Corporate Bankruptcy
Z-Score: multi-dimensional financial distress signal. Conceptual ancestor of our balance sheet scoring.
Jensen & Meckling
1976
Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure
Agency problem as the quantifiable cost of misaligned management incentives.
Grossman & Stiglitz
1980
On the Impossibility of Informationally Efficient Markets
Markets cannot be perfectly efficient; informed analysis must be compensated or it ceases to exist.
Kahneman & Tversky
1979
Prospect Theory: An Analysis of Decision under Risk
Cognitive biases as systematic mispricers. Loss aversion, availability bias, anchoring.
Abernathy & Utterback
1978 / Christensen 1997
Industry Lifecycle & Innovator's Dilemma
Disruption follows predictable patterns; structural risk can be scored before it appears in financials.
Fama & French
1992
The Cross-Section of Expected Stock Returns
Three-factor model; value and size premia. Quality as the missing fourth factor.
Sloan, R.
1996
Do Stock Prices Fully Reflect Information in Accruals and Cash Flows?
Accruals anomaly: high accruals predict negative returns. Foundation for our earnings quality scoring.
Rau & Vermaelen
1998
Glamour, Value and the Post-Acquisition Performance of Acquiring Firms
Acquirers systematically underperform. Basis for our organic growth stripping methodology.
Shefrin & Statman
2000
Behavioral Portfolio Theory
Investors construct layered portfolios based on mental accounts, not mean-variance optimisation. Creates systematic mispricings.
Greenwald, B.
2001
Value Investing: From Graham to Buffett and Beyond
Earnings Power Value as an alternative to DCF. Reproduction cost as a valuation floor.
Mauboussin, M.
2002
Measuring the Moat
Competitive Advantage Period: the time horizon over which a company earns excess returns. Framework for business model scoring.
Piotroski, J.
2000
Value Investing: The Use of Historical Financial Statement Information to Separate Winners from Losers
F-Score: 9-dimension fundamental screening. Documented 7.5% annual outperformance. Structural precedent for our 8-dimension model.
Tetlock, P.
2007
Giving Content to Investor Sentiment: The Role of Media in the Stock Market
Negative words in media predict negative returns. Foundation for our management language read.
Asness, Frazzini & Pedersen
2013
Quality Minus Junk
Quality companies are systematically underpriced. Highest-quality quintile outperforms lowest by 4.7% annually. Direct theoretical basis for our framework.
Loughran & McDonald
2011
When is a Liability Not a Liability? Textual Analysis, Dictionaries, and 10-Ks
Financial-domain sentiment word lists. Basis for our management language semantic scoring.
Shiller, R.
2019
Narrative Economics
Stories drive price before evidence arrives. The narrative premium is the primary source of our gap signal.