Methodology
Stock scoring system: how Bullish Stock Alerts ranks research priority
Our stock scoring system gives each eligible US-listed company a score out of 100. It combines valuation, quality, growth, risk and technical context to show which companies deserve deeper research first. This page documents the public model (V1) behind the scanner: the score weights, the fair value estimate methodology, a worked example and the limits. It is educational research, not personalized investment advice.
What is a stock score?
A stock score is a single number that summarizes several measurable traits of a company, so that a long list can be sorted consistently. Ours answers one question: which companies are worth researching first? It is a research priority, not a forecast, a price target or a recommendation. Unlike an analyst stock rating system (buy, hold, sell), it does not tell you what to do with a share.
How to read the score
Each component is scored from 0 to 100, and the weighted total is rounded to a whole number. The scanner places the result in one of these categories:
- 75 to 100, High research priority: review first, if data coverage is sufficient.
- 60 to 74, Research candidate: worth deeper analysis.
- 45 to 59, Watchlist: not enough support yet for a higher priority.
- 0 to 44, Limited data: a weak research case or insufficient inputs.
- N/A, Not eligible: outside the universe or missing minimum data.
Read a score together with its confidence figure and its data date: an 80 built on stale or partial data says less than a 65 built on complete, fresh data.
Score weights: valuation 40%, quality 25%, growth, risk and technical
Valuation carries the most weight because the scanner looks for gaps between price and estimated value. Quality comes second, so that a low price alone is not enough to rank high.
| Component | Weight | What it measures in model V1 |
|---|---|---|
| Valuation | 40% | Starts at 50 and adds 1.1 points per percentage point of upside from the price to the base fair value estimate (a negative upside lowers it). It loses 1.2 points per P/E unit above 22 and 1.5 points per price-to-book unit above 6. Without a price or a calculable fair value, it is set to 35. |
| Quality | 25% | The quality score starts at 45 and moves with return on equity, return on assets, net margin and operating margin. |
| Growth | 15% | Starts at 45 and moves with the latest annual growth in earnings per share (EPS) and in revenue. |
| Risk | 15% | Starts at 80. A beta above 1 removes 28 points per unit above 1, a base fair value more than 10% below the price removes 12 points, and a missing beta removes 5. |
| Technical | 5% | Starts at 50 and moves with the gap between the price and its 50-day and 200-day moving averages, as supporting context only. |
Formula: Score = Valuation × 0.40 + Quality × 0.25 + Growth × 0.15 + Risk × 0.15 + Technical × 0.05, rounded to the nearest whole number. Weights are not renormalized when data is missing; the confidence figure flags the gap.
How fair value is estimated
In the scanner, fair value is our working estimate of intrinsic value. It follows fixed rules, applied identically to every eligible company. That keeps a large universe comparable; it does not replace a full valuation of any single company.
- Sector multiple. Start from a target price-to-earnings (P/E) multiple: 24 for technology and software, 21 for communication services, 19 for healthcare, 18 for consumer companies, 13 for financials and 16 for other sectors.
- Quality and risk adjustments. Add 4 points when return on equity is above 20% (2 points above 12%), add 2 points when net margin is above 20% and remove 3 points when beta is above 1.5. The final multiple stays between 8 and 32.
- Internal value. Multiply that multiple by earnings per share (EPS), when EPS is positive.
- Blend with the analyst consensus. When an analyst consensus target is available, the central estimate is 65% internal value and 35% analyst target. When only one of the two exists, it is used alone. When neither exists, the scanner shows N/A instead of inventing a value.
- Scenarios. The bear case is 80% of the central estimate, the base case 100% and the bull case 120%. Upside is the gap between the base case and the price.
This is an earnings-multiple model, not a discounted cash flow (DCF) model, and not a specialized model for banks, insurers or REITs. A fair value is an estimate under assumptions, never a fact or a price target.
Valuing one company in depth takes explicit assumptions. Our stock research reports go further, with a thesis, a dated intrinsic value range, scenarios and invalidation points. To test an estimate against your own assumptions, use the intrinsic value calculator (two-stage DCF) or the Benjamin Graham formula calculator, or read how to calculate the intrinsic value of a stock.
Worked example: how a score of 76 is built
The inputs below are fictional. They only show the arithmetic of the public model V1, in use in October 2026, and describe no real company.
Inputs: technology company, price $120, market cap $30 billion, EPS $5, analyst consensus target $150, P/E 24, price-to-book 5, return on equity 18%, return on assets 8%, net margin 22%, operating margin 25%, EPS growth 10%, revenue growth 8%, beta 1.20, 50-day average $115, 200-day average $110.
- Target multiple: 24 (technology) + 2 (return on equity above 12%) + 2 (net margin above 20%) = 28. Internal value: $5 × 28 = $140.
- Fair value range: $140 × 65% + $150 × 35% = $143.50 for the base case, with a bear case of $114.80 and a bull case of $172.20. Upside to the base case: 19.6%.
- Components: valuation 69, quality 100 (capped), growth 60, risk 74, technical 58.
- Score: 69 × 0.40 + 100 × 0.25 + 60 × 0.15 + 74 × 0.15 + 58 × 0.05 = 75.6, displayed as 76, in the High research priority category.
- Confidence: 45 to start, plus 10 for the analyst target, 8 for positive EPS, 5 for a positive P/E, 6 for the price and 4 for the market cap = 78 out of 100.
A 76 puts this fictional company near the top of the research list. It does not mean the price will reach $143.50: the next step is to check the EPS in the filings and ask whether a P/E of 28 suits this business.
How our fair value differs from Morningstar fair value, GF Value and Simply Wall St
Several services publish a “fair value” for US stocks, and the same company can show four different numbers on the same day. That is expected: each number comes from a different method.
| Estimate | Main method, as publicly described |
|---|---|
| Bullish Stock Alerts scanner (model V1) | Sector earnings multiple adjusted for quality and risk, blended 65/35 with the analyst consensus target; bear and bull cases 20% below and above. The same rules apply to every company. |
| Morningstar fair value estimate | Set by Morningstar equity analysts, mainly with a discounted cash flow model, and shown alongside economic moat and uncertainty ratings. |
| GuruFocus GF Value | Based on the multiples a stock has historically traded at (such as P/E, price-to-sales, price-to-book and price-to-free-cash-flow), adjusted for past returns and growth and for estimates of future business performance. |
| Simply Wall St fair value | Mainly a two-stage discounted cash flow model built on analyst forecasts, with other models for some companies, such as banks, insurers, REITs and steady dividend payers. |
None of these numbers is the true value of a company. Ours is deliberately simple and transparent: useful for sorting a large universe, less suited than a company-specific DCF to valuing one business. When estimates disagree widely, the gap shows which assumptions (growth, margins, discount rate) to check first.
Third-party methods are summarized from each provider’s public explanations as of October 2026 and may change. Names are trademarks of their owners; Bullish Stock Alerts is not affiliated with them.
Data sources
The universe comes mainly from Nasdaq Trader listings and SEC EDGAR, which also provides official filings and XBRL company facts. Prices, fundamentals and macroeconomic data can come from configured providers such as FMP, Twelve Data, Alpha Vantage, EODHD, Tiingo, FRED or BEA.
Data freshness
Scores are recalculated from stored data by scheduled server jobs, not when you open a page. Each row carries a data date: fresh (under 24 hours), partial (1 to 7 days), stale (7 days or more) or unknown. The public preview is delayed by at least 24 hours.
Data confidence
Confidence, from 35 to 92, reflects which inputs are available, such as an analyst target, positive EPS, a valid price and the market cap. Rows below the publication threshold (40 by default) are hidden. Confidence is not a prediction of return and not a measure of certainty.
Exclusions: why some stocks are hidden
Coverage is limited to US-listed stocks. ETFs, ADRs, SPACs, shell companies, warrants, units, low-volume securities and securities outside the configured market-cap filters may be excluded from public views. A company can also be hidden when its price is missing, its fundamentals are insufficient, its data is stale or its confidence is below the publication threshold.
A hidden stock is not a negative opinion. It only means the model cannot score it reliably with the data available.
Limits: what the score does not tell you
- It does not predict returns. It was not designed or validated as a return forecast, and no backtested performance is claimed for it.
- It is a general model. The sector multiple only partly captures sector differences. Compare scores within a sector, and take extra care with banks, insurers, REITs and cyclicals.
- It leans on reported earnings and analyst targets. A one-off gain, a cyclical peak in profits or an outdated consensus can inflate the fair value; a loss-making year removes the internal estimate.
- Some important factors are not scored. Free cash flow, debt and interest coverage, dilution, capital allocation and management quality are not active in model V1. They belong in your own research.
- Cheap is not the same as undervalued. A low price can reflect a real problem the data has not caught yet: the classic value trap.
- Data can be wrong or late. Check key figures against the company’s own reports.
How to use the scanner
Filter by sector, risk and market cap, then sort by score or upside. Treat the top of the list as a research list, not a shopping list: check the data against official filings, compare peers, build your own thesis and decide in advance what would prove it wrong. Our guide on how to find undervalued stocks walks through that process in five steps, and the stock catalyst calendar lists upcoming earnings, ex-dividend dates, splits and IPOs.
The public preview of the undervalued stock screener is delayed by at least 24 hours and limited to 8 rows. Members see the full list with scores, fair value ranges and company details (see plans and pricing).
Stock score FAQ
What is a stock score?
A number from 0 to 100 that combines valuation (40%), quality (25%), growth (15%), risk (15%) and technical context (5%) to rank companies by research priority. It is not a forecast or a recommendation.
Does a high score predict returns?
No. It means the available data puts the company near the top of the research list. No backtested performance is claimed for the score.
How often are scores updated?
Scheduled server jobs recalculate them from stored data, within provider limits. Each row shows its data date; the public preview is delayed by at least 24 hours.
Why are some stocks hidden?
They fall outside the US-listed universe and its filters, or their data is missing, stale or below the confidence threshold. It is not a negative opinion.
How is fair value calculated?
Positive EPS times a sector target P/E adjusted for quality and risk, blended 65/35 with the analyst consensus target when one exists. Bear and bull cases sit 20% below and above. It is not a DCF.
Important disclaimer
This website provides educational research only. Nothing on this page is personalized financial advice or a recommendation to buy, sell or hold securities. Read the full disclaimer.