How YieldIQ works
Three pillars behind every YieldIQ verdict
Every stock on YieldIQ is scored against the same three questions: what is it actually worth, how durable is the business, and where does it sit against its peers? Below, in plain English, is exactly how we answer each one.
Pillar 01
Fair Value (DCF)
Two-stage 10-year discounted cash flow with cyclical-trough handling and ADR currency awareness. Bear, base, and bull scenarios published together with a confidence interval.
Pillar 02
Moat
A five-signal moat formula — margin stability, ROIC vs WACC, market share, switching costs, network effects — labelled Wide, Moderate, Narrow, or None.
Pillar 03
Sector Percentile Valuation
Peer-cohort ranking against 41 to 345 same-sector NSE peers, mapped to six explicit bands from Notable discount to Notable premium.
01 — Fair Value
The DCF, in detail
YieldIQ’s fair-value engine is a two-stage discounted cash flow model: a five-year explicit projection, followed by a five-year fade to terminal growth. The terminal stage doesn’t snap to a single number on year 11 — it decays smoothly, so the path the market actually walks is modelled rather than caricatured.
WACCis industry-tiered. Regulated utilities discount at roughly 9%. NBFCs sit 50 basis points above their non-financial peers to reflect the cyclicality of credit costs. Banks follow a separate equity-only path because the “debt” on a bank’s balance sheet is its raw material, not its cost of capital.
Terminal growth is size-tiered. Large caps fade to 6%. Mid caps to 7%. Small caps to 8%. The reasoning is simple: a ₹20,000 Cr business has a lot more runway than a ₹5 lakh Cr one, and a single blanket terminal number is the failure mode that wrecks most public-domain DCF spreadsheets.
Every valuation is published in three scenarios — bear, base, and bull. The scenarios flex growth, margin, and reinvestment jointly, not one input at a time. The confidence interval reported on the verdict reflects the spread between those three end-states, which is what real uncertainty looks like.
Reverse DCF — what FCF growth and margin must the market be pricing in to justify today’s share price?
Alongside the forward model, YieldIQ runs a reverse DCF that solves for the free cash flow growth and steady-state margin implied by the current share price. When the implied number sits well above anything the business has ever delivered, the model says so plainly. When it sits below management’s own guidance, that’s a setup the forward DCF would never tell you about on its own.
Cyclical trough anchor.When intrinsic value divided by price drops below 0.2 on a stock the classifier flags as cyclical, the estimate is anchored to 0.95× the current price instead of being downgraded to data_limited. Real cycle bottoms — metals, autos, oil — produce depressed-trailing-FCF inputs that crush a naive DCF; the anchor is the discipline that stops us declaring “data limited” on the very stocks where the model has the most to say.
02 — Durability
The Moat formula
A high fair value with a thin moat is a value trap waiting to happen. The Moat score answers a separate question: how durable are the cash flows the DCF is discounting?
- Margin stabilityStandard deviation of operating margin across the cycle. A high score requires the margin to remain stable through input-cost shocks, not just average to a flattering mean.
- ROIC vs WACCThe persistent spread between return on invested capital and cost of capital. We score the multi-year average; one good year doesn't prove a moat.
- Market sharePosition within the sector cohort. Leadership matters, but so does direction; share losses to a faster competitor erode the score even if the lead survives.
- Switching costsA composite of customer concentration, contract tenor, and revenue retention where disclosed. Subscription IT and enterprise pharma score highest.
- Network effectsA signal limited to platforms and exchanges where active-user growth itself enhances unit economics. Most stocks correctly score zero on this axis.
The five signals composite to one of four labels: Wide, Moderate, Narrow, or None. A Wide moat lifts the YieldIQ composite by 10 points; a None actively penalises it.
The 18-stock bellwether allowlist — HDFCBANK, HUL, NESTLE, TITAN, ASIANPAINT, TCS, INFY, and others — floors at Wide. We refuse to publish a Narrow on a 30-year compounder because two quarterly metrics drifted.
The allowlist is small, named, version-controlled in source, and reviewed each year. It is not a fudge layer; it’s a guard against the single most common failure mode in quantitative moat scoring — the model penalising structurally great businesses for short-term noise.
03 — Quality
Piotroski F-Score, with a bank-aware mode
The classic Piotroski F-Score is nine binary fundamental signals: profitability, leverage, and operating efficiency. For non-financial businesses, YieldIQ runs it unchanged.
For banks and most NBFCs, the classic formula is wrong on its face. Half the signals — current ratio, gross margin, asset turnover — are not meaningful for a balance-sheet business. Running the classic 9-signal on a bank produces a stream of false WEAK ratings even on HDFC Bank or Kotak.
Bank mode scores the four signals that doapply — ROA positive (f1), operating cash flow positive (f2), ROA improving year-on-year (f3), and no equity dilution (f7) — and rescales to the standard 0–9 range so the score is directly comparable to non-financials.
Most public stock screeners run the classic 9-signal Piotroski on every ticker, including banks. This is the single biggest reason their “quality” rankings put PSU laggards ahead of HDFC Bank. We don’t.
04 — Relative value
Sector Percentile Valuation
A DCF tells you what a stock is worth in absolute terms. Sector Percentile tells you where it sits inside its cohort — the question every investor asks second. YieldIQ ranks each ticker against 41 to 345 same-sector NSE peers, depending on sector density.
- Notable discount to peersBottom decile on the cohort's blended valuation axis. The market is paying a meaningful discount versus same-sector competitors.
- Below peer rangeBelow the cohort median but inside the normal range. Discounted versus peers, but not in a notable discount band.
- In peer rangeInside the inter-quartile band. No relative-value edge either way on the cohort.
- Above peer rangeAbove the cohort median but inside the normal range. Pricier than peers without being extreme.
- Notable premiumTop decile on the cohort's blended valuation axis. The market is paying a meaningful premium versus same-sector competitors.
- Insufficient peer dataThe cohort is too thin for a meaningful percentile (typically newly listed sectors or sub-40-stock cohorts). We say so explicitly rather than guess.
Each axis on the YieldIQ Hex carries a per-axis percentile score: 0–100 percentile mapped to a 0–10 score via the linear transform score = 10 − pct/10. That keeps each axis directly comparable across stocks even when the underlying distributions differ wildly.
Cohort taxonomy is sector-specific. IT services run against IT services; private banks against private banks; PSU banks against PSU banks. The General catch-all cohort is used only when no specialised taxonomy applies, and we flag it when it does.
05 — The composite
The YieldIQ Score
The composite score is a single 0–100 number, decomposed across five axes — Quality, Safety, Value, Growth, and Moat — and rendered as a letter grade for quick scanning.
- A+ / ATop-tier composite. High scores on all axes; no single axis below the cohort median.
- B+ / BSolid. High on most axes with one or two soft spots that are usually known and disclosed.
- C+ / CMixed. Real high points offset by real low points; the verdict is a thinking aid, not a green light.
- DBelow median. Multiple axes below the cohort median or red-flagged on Safety. Read the report before reading the grade.
Defence-in-depth: margin-of-safety is clamped to [−50%, +50%] before bucketing. A single classifier gap should never be enough to push a stock to A+ on broken data.
The clamp exists because we’ve watched competitor screeners issue glowing grades on stocks where one upstream calculator returned a runaway MoS. The clamp ensures that the composite degrades gracefully when a single input is wrong, rather than flipping to a top grade on a single bad number.
06 — Inputs
Data sources & discipline
- NSE XBRL filingsPrimary source for fundamentals. Approximately eight years of structured filings, FY18 through FY25, parsed line-item by line-item with validators in front of every field.
- yfinance fallbackUsed only for tickers and fields where NSE XBRL has gaps. Cached aggressively, gated by a process-wide circuit breaker, and validator-checked for unit-jump corruption.
- NSE corporate actions APICanonical source for dividends, splits, bonuses, and bulk-deal feeds. Used for total-return and dilution checks.
- 12 NSE sectoral indicesBank, IT, Pharma, Auto, FMCG, Energy, Metal, Realty, Media, PSU Bank, Private Bank, and Financial Services. These are the canonical sector taxonomy used for cohort assignment.
- BSE filingsUsed for announcements, news, and disclosures that surface on BSE first. Read-only — fundamentals come from NSE.
Cache discipline: every CACHE_VERSION bump is documented in source with a before/after canary-diff snapshot on 50 reference stocks. No exceptions.
07 — Regulatory
What we won’t do
YieldIQ is a research surface, not a recommendation engine. We are not registered with SEBI as an investment adviser or research analyst. The product is built so that it cannot issue advice even if we wanted it to.
- We don’t issue directional recommendations — none of the four rating words conventional sell-side analysts use to flag relative or absolute performance appear in our copy.
- We don’t issue price targets.
- All output is a model estimate, not investment advice.
- A vocabulary lint runs on every PR. Any forbidden term in user-visible copy fails the build.
Verdicts are descriptive, not imperative. Where data is insufficient we apply an explicit Insufficient peer data or data_limited label rather than guess. Consult a SEBI-registered adviser before making any investment decision.
08 — Honesty
What YieldIQ doesn’t do
- 8 years of historyNot 15+. NSE’s structured XBRL archive starts in FY18; older filings exist as PDFs but not as machine-readable line items. We use the cleanest data available rather than scraping unreliable sources for an extra five years.
- 3,005 active NSE tickersThe full active equity universe today. We’re expanding to a 5,000-ticker target as we add the BSE-only and SME segments, but we won’t add a stock until the data quality clears the same bar as the rest.
- ADR / cross-listed namesCross-listed ADR data quality is harder. Currency conversion, ADR-to-ordinary-share ratios, and FY mismatch all introduce noise. We flag these tickers with cross-listed (ADR / USD reporting) and data_limited where appropriate.
- Regulatory and M&A shocksA DCF cannot model a sudden tariff, a competition-commission ruling, an aggressive acquirer, or a single management decision that erodes a moat overnight. The model is silent on those events; the human reading the report is not.
For the long-form analyst appendix — sector-by-sector assumptions, the verdict-band table, and the full SEBI posture — see the methodology page. For pricing tiers and access, see pricing.