P/E of 25 — Cheap or Expensive? Use Percentile Ranking
4-section structure: Concept / How We Compute / How to Read / Caveats.
1. Concept
"This stock's P/E is 25x" — that sentence alone is meaningless.
Whether 25x is expensive depends on:
- Its own history — has it traded at 10x or 40x before?
- Its peers today — are others at 40x or 15x?
So absolute multiples are useless; relative position matters.
Two dimensions of "relative":
- Time dimension: Self-history percentile — "I'm cheaper than X% of my own history"
- Cross-section: Industry percentile — "I'm cheaper than Y% of my peers today"
This platform shows both on a single card so you can see at a glance whether a stock is actually cheap or expensive.
2. How We Compute
2.1 Four Valuation Metrics
| Metric | Formula | Reading |
|---|---|---|
| P/E | Price / Annual EPS | Paying how much per $1 of earnings |
| P/B | Price / Book Value Per Share | Paying how much per $1 of book equity |
| P/S | Price / Revenue Per Share | For loss-makers & high-growth names |
| Dividend Yield | Annual Cash Dividend / Price | Direction reversed (higher = cheaper) |
EPS and BVPS use annual Q4 data; dividends from the latest record in FR_DIV_全_tw_h.
Companies with negative EPS are excluded from distributions.
2.2 Self-History Percentile
For each of the past 6 years, take yearly PE (high + low) / 2 as the representative value. Current PE is ranked in this 6-year distribution.
Example: past-6-year PE midpoints: 12, 14, 15, 18, 22, 25; current 20 → 4th position → ~P67 (higher than 67% of history).
2.3 Industry Percentile
Take all stocks in the same sector today; compute their PE/PB/PS/yield distributions; rank the target within them.
2.4 Visual Design
Each card has two percentile bars:
- Top = self-history (time)
- Bottom = industry (cross-section)
- Green → cheap, Red → expensive
- Yield direction reversed: P100 = highest yield = cheapest
2.5 Caching
API cached 15 min. During TW trading hours (~4.5h), this refreshes ~18× — sufficient.
2.6 Peer Leaderboard
Below the cards we list the top 10 peers by market cap in the same sector, with the target stock highlighted, side-by-side PE/PB/PS/yield/market-cap comparison.
3. How to Read
3.1 Four-Quadrant Interpretation (core)
| Self | Industry | Verdict |
|---|---|---|
| Low | Low | 🟢 Absolutely cheap — great entry zone |
| Low | High | 🟡 Company deteriorating — why are peers richer? |
| High | Low | 🟡 Sector out of favor — cyclical bottom? |
| High | High | 🔴 Overheated — risky across the board |
3.2 Which Metric to Prioritize
- Growth stocks: PE + P/S
- Value stocks: PE + P/B
- Financial / REIT / cyclical: P/B
- Income / high-dividend: Yield + payout stability
- Loss-making / SaaS: P/S
3.3 History vs Industry Signal Strength
Industry percentile is usually stronger because it's apple-to-apple and unaffected by company structural shifts. History percentile is still useful to detect sector-wide bubbles when everyone is at P50 industry but P95 history → sector bubble.
3.4 Yield Caveat
Yield percentile reverses direction:
- P100 = highest yield = cheapest
- P0 = lowest yield = expensive
But watch out for yield traps: extreme yields often follow price crashes (denominator collapse, not numerator stability).
4. Caveats
⚠️ High Growth Naturally Commands High P/E
A 30% growth stock deserves 30× P/E; a 5% growth stock at 30× is expensive. Use PEG or DCF alongside.
⚠️ Structural Change Invalidates History
Companies that have transformed (e.g., manufacturing → software) in the last 3 years have irrelevant history. Use industry percentile only, or wait 3 years to accumulate new history.
⚠️ Coarse Sector Granularity
Platform uses TWSE's sector classification. Within "semiconductors" you have IC design / foundry / OSAT / IDM — very different valuation logic. Do your own mental refinement.
⚠️ Small Industry Samples
Some sectors have only 5–10 stocks. Rankings become volatile (one company's change reshuffles everyone). System flags ⚠ insufficient sample in those cases.
⚠️ Financials' Distorted P/E
Banking/insurance/brokerage P/E is distorted by valuation gains/losses and regulatory basis changes. Prioritize P/B for financials.
⚠️ TW Only (for now)
US version coming in Phase 1 — requires SIC/GICS mapping and unit handling.
Further Reading
- What Is DCF? A Step-by-Step Valuation
- 6 Valuation Models Compared
- Altman Z / Piotroski F / Beneish M Three Scores
Try It
- Open Stock Analysis → Valuation — the top block is Relative Valuation
- Read the four-quadrant prompt; observe both self-history and industry bars
- Scroll down to the peer leaderboard for an apples-to-apples comparison
- Click 📐 for the full formula and parameters