The last article answered “how big is this stock” (market cap). This one answers “is it expensive?” All valuation metrics share one idea: divide market cap by some operating number of the company, and see how much you’re paying per unit of profit / assets / revenue. Different divisors, different metrics — and completely different situations where each applies.
P/E (price-to-earnings): what you pay for each yuan of profit
P/E = total market cap ÷ annual net profit. A P/E of 20× means: at current unchanging profits, the price you paid would take 20 years of company profits to earn back (that’s the intuition, not an actual payback promise).
Market apps show P/E in three flavors, and the numbers can differ a lot — you must tell them apart:
- Static P/E: market cap ÷ last complete fiscal year’s net profit. The most certain, but the stalest — halfway through this year it’s still using last year’s profits.
- Forward P/E （动态市盈率）: market cap ÷ estimated current-year net profit (usually by annualizing published quarterly reports — e.g., Q1 profit × 4). The freshest, but badly distorted for strongly seasonal companies — a retailer that earns its money in Q4 will show an annualized loss off its Q1 report.
- TTM P/E (trailing twelve months): market cap ÷ net profit of the most recent four quarters. Both fresh and complete — this is the default for analysis.
P/E’s fatal blind spot: when the company loses money, P/E is negative and meaningless. That’s why a loss-making stock’s P/E field in market apps either says “loss” or stays blank.
P/B (price-to-book): what you pay for each yuan of net assets
P/B = total market cap ÷ net assets (shareholders’ equity). Net assets are, on the books, “the part that would belong to shareholders if the company were liquidated.” P/B of 1× means the market price equals book value; P/B of 0.6× means buying the book assets at a 40% discount.
P/B has a narrower applicable range than P/E, but in specific domains it beats P/E:
- Banks, insurers, brokers: assets are mostly financial, so book values are relatively real, while earnings are strongly cyclical (P/E looks cheap at profit peaks and expensive at troughs — exactly backwards). That’s why the common language of bank valuation is P/B — A-share banks trade at 0.5–0.8× P/B for years on end, and trading below book (P/B < 1) is the norm, not evidence of absurd cheapness.
- Heavy-asset, strongly cyclical industries (steel, shipping): profits at the top and bottom of the cycle differ dozens of times over; P/E lies, P/B stays relatively stable.
- Not applicable to asset-light companies: the value of internet and consumer-brand companies lives in brand, users, and network effects — almost none of it on the books. Moutai trades above 10× P/B year-round; that doesn’t mean it’s 10× overpriced — it means P/B is the wrong ruler for it.
P/S (price-to-sales): for companies without profits yet
P/S = total market cap ÷ annual revenue. Its use is highly specific: when a company has no profit (it’s loss-making), P/E fails, and at least there’s revenue to compare. Common for growth-stage tech and innovative drug companies.
The caveat: revenue is not profit. Two companies with identical revenue — one with 60% gross margin, one with 10% — at the same P/S mean completely different things. P/S can only compare companies in the same industry with the same business model; cross-industry P/S comparisons are meaningless.
Dividend yield: annual dividends as a percentage of your purchase price
Dividend yield = annual dividend per share ÷ share price. A ¥20 stock paying ¥1 a year has a 5% dividend yield — think of it as the stock’s “interest rate.”
Three details:
- It moves inversely with the price: for the same company, the further it falls, the higher the yield. So “high yield” is sometimes a value opportunity and sometimes a value trap — a yield propped up by a collapsing denominator, with deteriorating results and unsustainable dividends, is no gift.
- A-share high-dividend stalwarts are banks, coal, highways, and utilities, and a whole dividend strategy family has grown from them (dividend ETFs track exactly this kind of index).
- Dividends go ex (Part 4), so receiving one isn’t free money — the price reference drops by the dividend amount on the ex-date. The real meaning of dividends is that the company hands profits to you with certainty; reinvesting dividends over the long term is where the compounding comes from.
ROE (return on equity): is the business itself profitable
The first four metrics are all “price relative to the company” (the market’s view). ROE is a metric of “the company itself”:
ROE = net profit ÷ net assets. An ROE of 15% means that for every ¥100 of net assets shareholders have put in, the company earns back ¥15 a year. It answers the most essential question: how profitable is this business?
Buffett’s famous line, loosely: the long-run return from holding a stock converges to the company’s ROE, regardless of the price fluctuations at which you bought. Rule-of-thumb scale: ROE persistently above 15% is an excellent business, 10%–15% is passable, and below 8% you’d better ask why. Note that high leverage can prop up a fake-high ROE (more debt means less net assets), so glance at the debt-to-asset ratio whenever you look at ROE.
The shared limitation: every denominator is history
The denominators of all valuation metrics — net profit, net assets, revenue, dividends — are financial data that has already happened. But the market buys the future: a company about to explode upward always looks expensive on P/E, and a company about to blow up always looks cheap (the “low P/E trap” has harvested countless beginners at cyclical tops — P/E is lowest when profits are best, then profits collapse and the price and P/E go to heaven and hell together).
The correct usage: valuation metrics are for comparing and filtering, not for predicting. Who’s cheaper within the same industry? Is the stock expensive or cheap relative to its own historical range (the “P/E percentile” in market apps is exactly this use)? Valuation metrics answer those two questions very well. As for “will this stock go up” — they stay silent.
Next article completes the dividend story: cash dividends, bonus shares, reserve conversions, the record date, how the ex-rights/ex-dividend reference price is computed, and the holding-period tax tiers on dividends.
This article explains terminology only and does not constitute investment advice.