"A rating you can't recompute yourself isn't a rating — it's a rumour with a number attached. Every figure NEP50 publishes traces back to a formula on this page."
NEP50 Research · Methodology Charter
01 / 09
What the score is for
NEP50 exists to answer one question about a NEPSE-listed company, using only public numbers and one published formula: is this a well-run, profitable, shareholder-respecting business, or isn't it? Nothing about price direction, nothing about timing — just quality.
The score is deliberately boring by design. It doesn't predict what a stock will do tomorrow. It measures what a company has actually done — its profitability, how it treats minority shareholders, whether it pays what it promises, and whether the market can trade it in reasonable size. A high score means "this business has a track record of doing the fundamentals well." It is not a buy signal.
In plain English
Think of it like a school report card, not a fortune teller. A report card doesn't predict what grade you'll get next year — it tells you, using clear rules everyone agrees on in advance, how you actually did on the things that were measured. NEP50 is a report card for companies: profit, fairness to shareholders, dividends, and consistency. It doesn't guess the future.
02 / 09
The eligible universe
Every NEPSE-listed company starts in the universe. Nothing is excluded by sector, size, or reputation before the gates are applied — the gates decide, not a person.
This matters more than it sounds. If a human analyst pre-selects "the good companies" before running the numbers, the numbers just confirm what the analyst already believed. NEP50 runs every listed company through the same four gates, mechanically, with no exceptions and no discretion. A company either clears all four gates or it doesn't get a score at all — it isn't scored low, it's simply not rated, the same way a student who skips an exam gets no grade rather than a zero.
In plain English
Imagine a doctor who checks every patient the same way, using the same four basic health tests, before deciding who needs a closer look. Nobody skips the tests because they "look healthy." NEP50 does the same thing to every company on the stock exchange — same four checks, no favourites.
03 / 09
The four mandatory gates
All four must pass simultaneously. Failing even one disqualifies the company from scoring entirely — there is no partial credit at this stage.
Gate 1 · Profitability
Weighted 3-yr ROE ≥ 15%
Return on Equity across the last three fiscal years, weighted 3×/2×/1× toward the most recent year, must average at least 15%.
Gate 2 · Dividend Discipline
Dividend paid ≥ 3 consecutive years
Cash or bonus dividend, paid without a gap, for at least three straight fiscal years — checked by walking backward from the most recent year until a miss is found.
Gate 3 · Liquidity
Median daily turnover ≥ NPR 1,000,000
A company can be excellent and still fail this gate if almost nobody trades its shares — a rating is only useful if you can actually act on it.
Gate 4 · Governance
Governance score ≥ 60 / 100
AGM timeliness, audit quality, board independence, and SEBON compliance rolled into a single 0–100 governance figure.
How the weighted ROE is actually computed
The most recent fiscal year counts three times as much as the oldest of the three, because a company's current profitability tells you more about where it's headed than what it did three years ago.
# weights = [3, 2, 1], applied to the 3 most recent fiscal years available
weighted_ROE = (ROE_year1 × 3 + ROE_year2 × 2 + ROE_year3 × 1)
÷ (3 + 2 + 1)
In plain English
It's like a final grade that counts your most recent test more than your first one — because it tells the teacher more about how you're doing right now, not three months ago.
04 / 09
The quality score formula
Every company that clears all four gates receives four sub-scores, each independently capped between 0 and 100, then combined into one composite Quality Score using fixed weights.
35%
Profitability
Weighted ROE, rescaled
25%
Governance
Direct governance score
25%
Dividend
Latest dividend yield, rescaled
15%
Trend
Price-trend classification
1 · Profitability sub-score
The weighted ROE from Gate 1 is rescaled onto a 0–100 line, anchored so that a 25% ROE lands near the top of the scale, with a floor of 10 points so that any company which passed the gate still starts from a non-zero base.
profitability_score = min(100, max(0,
(weighted_ROE ÷ 25) × 100 × 0.9 + 10
))
2 · Governance sub-score
Governance is already a 0–100 figure from the audit/disclosure review, so it passes through directly, only clamped to stay within range.
governance_score = min(100, max(0, governance_input))
3 · Dividend sub-score
The most recent dividend yield is rescaled the same way as profitability, anchored so an 8% yield sits near the top of the scale, with a 15-point floor.
dividend_score = min(100, max(0,
(dividend_yield ÷ 8) × 100 × 0.85 + 15
))
4 · Trend sub-score
Trend is the one qualitative input — a company's recent price trend is classified as up, down, or stable/neutral, then mapped to a fixed point value. This exists to slightly reward companies gaining market recognition and slightly penalise ones in a sustained slide, without letting short-term price action dominate the score the way it dominates a trading signal.
trend_score = 82.0 if classified "up"
trend_score = 54.0 if classified "down"
trend_score = 68.0 otherwise (stable / neutral)
Composite score
The four sub-scores are combined with their fixed weights, clamped to the 0–100 range, and rounded to one decimal place. This is the number displayed everywhere as the NEP50 Quality Score.
composite = 0.35 × profitability_score
+ 0.25 × governance_score
+ 0.25 × dividend_score
+ 0.15 × trend_score
NEP50_score = round( min(100, max(0, composite)), 1 )
In plain English
It's a weighted average, the same idea as a course grade made of "40% exams + 30% homework + 30% attendance." Nothing here is a guess or a black box — plug the same four numbers into this same formula and you will get the exact same score, every time.
05 / 09
From score to tier
The composite 0–100 score is translated into one of five plain-language tiers, so a reader doesn't need to interpret raw numbers to understand where a company stands.
Elite
≥ 85.0
Best-in-class across all four factors, with no material weakness.
Strong
72.0 – 84.9
Solid across the board; comfortably above the eligibility gates.
Attractive
58.0 – 71.9
Meets every gate but shows a clear soft spot in one factor.
Neutral
44.0 – 57.9
Passed the gates narrowly; several factors are middling.
Weak
< 44.0
Cleared the gates on a technicality; the composite tells a weaker story.
In plain English
Like a letter grade on top of a percentage — 92% and 87% might both round to an "A," but you don't need to remember the exact percentage to know both are good. Elite through Weak works the same way for a company's numbers.
06 / 09
Full worked example
To make every step checkable, here is the entire calculation done by hand for a fictional company. No real NEP50 constituent is used here on purpose — this example exists purely so you can re-run the arithmetic yourself.
Illustrative only · not a real company
"Himal Finance Bank Ltd." — inputs
| Input | Value |
|---|
| ROE, FY082/83 (most recent, weight 3) | 19.0% |
| ROE, FY081/82 (weight 2) | 17.0% |
| ROE, FY080/81 (weight 1) | 15.0% |
| Dividend paid, consecutive years | 4 years |
| 6-month median daily turnover | NPR 2,400,000 |
| Governance score | 78 / 100 |
| Most recent dividend yield | 4.0% |
| Price trend classification | Up |
Step 1 · Gate check
| Gate | Computed | Threshold | Result |
|---|
| Profitability | 17.67% weighted ROE | ≥ 15% | Pass |
| Dividend | 4 years | ≥ 3 years | Pass |
| Liquidity | NPR 2.4M | ≥ NPR 1M | Pass |
| Governance | 78 | ≥ 60 | Pass |
weighted_ROE = (19×3 + 17×2 + 15×1) ÷ 6 = 106 ÷ 6 = 17.67%. All four gates pass, so the company proceeds to scoring.
Step 2 · Sub-scores
| Factor | Formula applied | Sub-score |
|---|
| Profitability | (17.67 ÷ 25) × 100 × 0.9 + 10 | 73.6 |
| Governance | direct pass-through | 78.0 |
| Dividend | (4.0 ÷ 8) × 100 × 0.85 + 15 | 57.5 |
| Trend | classified "up" | 82.0 |
Step 3 · Weighted composite
| Factor | Sub-score × weight | Contribution |
|---|
| Profitability | 73.6 × 0.35 | 25.76 |
| Governance | 78.0 × 0.25 | 19.50 |
| Dividend | 57.5 × 0.25 | 14.375 |
| Trend | 82.0 × 0.15 | 12.30 |
Falls in range
58.0 – 71.9 → Attractive
Notice how close this result sits to the Strong threshold (72.0) — a real reminder that tiers are a summary, and the underlying composite score is where the real precision lives. Two companies both labelled "Attractive" can still be meaningfully different; always check the composite number, not just the tier badge, when comparing two close constituents.
07 / 09
Data sources & refresh cadence
A methodology is only as trustworthy as the numbers fed into it. Here's exactly where every input comes from and how often it changes.
Inputs
Financial statements (ROE, dividend history, governance inputs) are drawn from companies' own audited annual reports and unaudited quarterly disclosures filed with NEPSE/SEBON. Price and turnover data (for the liquidity gate and trend classification) comes from daily NEPSE trading data. Governance inputs are compiled from AGM filing dates, audit opinions, and board-composition disclosures.
Refresh cadence
Scores recompute whenever a company files new quarterly or annual financials — there's no fixed calendar delay. Price-derived inputs (liquidity, trend) refresh on a rolling daily basis once a live price feed is connected. Every recomputation uses the exact formulas on this page; nothing is manually overridden.
Methodology governance
Review cycleReviewed semi-annually by NEP50 Research.
Change noticeMaterial formula or threshold changes announced with a minimum 30-day public notice.
Version historyMaintained publicly; this document is v1.0, effective May 2026.
08 / 09
Limitations, honestly
A credible methodology names its own weak points before someone else does.
Interim data can be unaudited. Quarterly figures used between annual reports haven't been through a full audit, and NEP50 flags any score built substantially on unaudited interim data as provisional.
The formula is deliberately simple, not statistically adaptive. These are fixed linear formulas with hard-coded anchors (25% ROE, 8% dividend yield), not a statistically normalised model that reweights itself against the whole market each quarter. That's a design choice for transparency and reproducibility — it trades some statistical sophistication for the ability of anyone to recompute a score by hand, as shown in Section 6.
The trend factor is the least "fundamental" input. Because it's derived from recent price action rather than company financials, it's the one place a short-term market mood can nudge a long-term quality score. It's capped at 15% weight specifically to limit that influence.
This is not investment advice. The NEP50 score measures historical business quality. It is not a price target, a forecast, or a recommendation to buy or sell any security. Past dividend consistency or ROE performance does not guarantee future results.
09 / 09
Glossary
Plain-language definitions for every term used on this page.
ROE
Return on Equity. How much profit a company makes for every rupee shareholders have invested in it. Higher generally means the business is good at turning owners' money into profit.
Weighted average
An average where some numbers count more than others — here, the most recent year's ROE counts three times as much as the oldest of the three years used.
Dividend yield
The dividend a company pays per share, shown as a percentage of the share price — like the interest rate you'd earn, but from stock ownership instead of a bank deposit.
Governance score
A 0–100 score reflecting how well a company follows good corporate practice: holding AGMs on time, publishing clean audits, and keeping an independent board.
MDTV
Median Daily Traded Value — the "typical day's" rupee value of shares actually traded, used to check that a stock is liquid enough to buy or sell without moving the price a lot.
Gate
A pass/fail eligibility rule. A company must clear all four gates before it's given a score at all.
Composite score
The final 0–100 number, made by combining the four sub-scores using their fixed weights (35/25/25/15%).
Tier
The plain-language label (Elite, Strong, Attractive, Neutral, Weak) attached to a composite score range, so readers don't need to interpret raw numbers.
