Methodology

Why this page exists

For any investing site, the question that matters most is simple: where did this number come from? Bong Stocks recalculates metrics for roughly 2,650 Korean listed stocks and 1,140 ETFs on every trading day. This page documents the sources, formulas, assumptions and verification steps behind them. Our rule is that any figure we publish should be reproducible from what is written here.

We do not publish buy or sell opinions, price targets, or stock scores and rankings. We publish metrics computed from public data, and the reasoning behind how we read them.

1. Data sources

• Price, volume, market cap, shares outstanding — Financial Services Commission open “Stock Price Information” API (data.go.kr)
• Financial statements, filings, dividends, ownership — DART, the Financial Supervisory Service's electronic disclosure system (OpenDART API)
• Indices and market classification — Korea Exchange (KRX) public data
• FX rate (for USD display) — median of three public exchange-rate APIs

All of these are freely available public sources. We do not redistribute paid data, brokerage research, or analyst consensus figures. Rather than republishing raw feeds, we publish values we compute ourselves using the formulas below.

2. Core valuation metrics

• P/E = price ÷ earnings per share, based on the last confirmed fiscal year. This can be misleading when earnings change sharply.
• P/B = price ÷ book value per share
• P/S = market cap ÷ annual revenue
• ROE = net income ÷ shareholders' equity
• Dividend yield = dividend per share ÷ price

Every value is recomputed from the latest closing price and most recent confirmed financials. Market cap comes from the FSC feed as published, and we verify on every build that it matches close × shares outstanding (see section 8).

3. Forward P/E — the metric that matters most

Trailing P/E looks backward. For companies whose earnings move quickly — cyclical industries like memory semiconductors, turnarounds, high-growth names — that number can be far from reality. So we compute a separate forward P/E based on this year's expected net income and show both.

Forward P/E = market cap ÷ expected net income for the current year

How expected earnings are determined (in priority order)

• ① Company's own figures — if the company has published revenue, operating profit or net income targets in a DART business-plan filing, a corporate value-up plan, or its own IR materials, we use those.
• ② Our own estimate — when no official outlook exists:
  · We estimate operating profit first (recent confirmed quarterly operating profit, projected forward along its trend, adjusted for the margin cycle), then derive net income by applying taxes and net financial items.
  · We do not simply annualise one quarter by four. If quarterly profit is trending up, remaining quarters are projected along that trend; if it is falling, they are marked down.
  · Companies sharing the same industry cycle must come out mutually consistent. If two memory-semiconductor names show very different forward multiples with no difference in results, margins or share to explain it, we re-examine the estimate.
  · We sanity-check our estimate against the range of expectations commonly circulating in the market, internally only. Those external figures and their sources are never published.
• ③ Seasonality approximation — when neither of the above is possible, we approximate the year from the average quarterly earnings distribution of the previous three fiscal years.
• ④ Not available — if expected net income is zero or negative (sustained losses), we do not produce a forward P/E and mark it as unavailable.

Forward P/B follows the same logic: we add this year's retained earnings (expected net income less expected dividends) to book equity, then divide market cap by that projected equity.

Limitation: ② and ③ are our own unverified estimates. Actual results will differ, and the error is larger for companies with volatile earnings. Every page labels which basis was used — company figures, our estimate, or unavailable.

4. Sector-specific headline metrics

Ranking every company by a single P/E is a false comparison. For a bank, asset quality is what matters, so P/B comes first. A pre-revenue biotech has no meaningful P/E at all. A holding company should be read through the value of the stakes it owns (NAV).

So we classify listed companies into a three-level taxonomy of about 123 detailed sectors and pre-assign, for each sector, the one metric that should be read first. That is what the “headline valuation metric” card at the top of each stock page shows — together with why that metric takes priority for that business.

• Memory semiconductors → forward P/E (earnings swing too much for trailing figures)
• Banks and insurers → P/B
• Holding companies → NAV / sum-of-the-parts; consolidated P/E and P/B are distorted here
• Pre-profit growth companies → revenue-based metrics such as P/S

Classification is based on the business actually described in filings. Where the official industry code diverges from economic reality — a company coded as wholesale trade that is in substance a holding company, for example — we correct it to the real business.

5. Enterprise value metrics

Market cap alone hides debt and cash. Enterprise value accounts for both.

• Net debt = interest-bearing debt − cash and equivalents (negative means net cash)
• EV = market cap + net debt
• EV/EBIT = EV ÷ operating profit · EV/EBITDA = EV ÷ (operating profit + depreciation & amortisation) · EV/Sales = EV ÷ revenue
• FCF yield = (operating cash flow − capital expenditure) ÷ market cap

For large caps whose summary cash-flow statement omits depreciation, we reconstruct it from the XBRL source filing. For sectors where EV is not meaningful — banks, insurers — we do not compute or display it.

6. Intrinsic value (DCF estimate)

A discounted cash flow model estimates per-share value from future cash the business is expected to generate. We use the assumptions below and publish all of them on the page.

• Two-stage model — a 10-year explicit forecast plus a Gordon terminal-growth stage
• Discount rate — sector-standard weighted average cost of capital
• Base cash flow — free cash flow. For profitable companies whose FCF is temporarily negative because of heavy capex or a captive finance arm, we substitute projected earnings and state on the page which basis was used.
• Output is a bear / base / bull range, never a single “fair price”.

Limitation: DCF output swings widely with discount and growth assumptions. Read it as a range implied by assumptions, not an answer. We never present it as a price target.

7. Four-axis diagnostics and the valuation label

Each company is graded on financial health, growth, profitability and valuation, compared against the median of its own sector. The valuation label — Undervalued, Fairly valued, Overvalued or Inconclusive — follows these rules.

• Forward P/E takes priority. Where a forward figure exists, it is compared to the sector median. We never fall back to trailing P/E for a company that has one.
• Below-book guard. A company trading under book value (P/B below 1) is never escalated to “overvalued on an asset basis”. This rule was added after that adjustment wrongly flagged large-cap automakers as overvalued.
• Thin sector samples return Inconclusive. If there are too few comparables or the metric does not apply, we decline to judge rather than force a label.
• The displayed label reflects the sector headline metric together with the in-depth review, and the reasoning is shown in the “Valuation vs peers” section of each page.

Relative strength (RS) converts one-year return relative to the index into a universe-wide percentile. It describes price behaviour only — it is not a judgement about business value.

8. Quality gates that run before publication

If the numbers are wrong, nothing else matters. Every update runs the following checks across all companies, and a single violation stops the deployment.

• Market cap consistency — close × shares outstanding ≈ published market cap
• P/B and P/E consistency — price ÷ book value per share ≈ displayed P/B; price ÷ EPS ≈ displayed P/E
• Headline metric consistency — the value on the sector metric card matches a recomputation from source data
• Korean–English consistency — the same metric shows the same number in both languages
• Label consistency — the valuation label in the list matches the one on the detail page (added after the two disagreed)
• In-text figure consistency — P/E, market cap and similar figures written in the prose are checked against current computed values
• Price freshness — if the public feed falls behind actual trading days (excluding market holidays), we stop rather than publish stale prices
• Classification regression tests — automated tests prevent sector misclassification caused by partial name matches recurring