Home / Indicator library
Indicator library
Valuation, ownership, fundamentals and research tools, grouped so you can find the series you need and try it on a real ticker. Every number is either the official public source shown as-is, or a derivation you can reproduce from it.
Prefer to start from a sector? The Taiwan stock sector index lists all 34 sector classifications covering 1,971 tickers, and each sector page shows every ticker in it with P/E, P/B and dividend yield from that day's snapshot.
Want the historical base rate of how prices behaved after a given filing? See Event study statistics — 27 categories of Taiwan stock filings with their historical abnormal returns, CAR and BHAR benchmarks side by side, flagged where the two disagree.
Want to chart one first? The valuation river chart plots eight percentile bands from a ticker's own PER, PBR or dividend-yield history, each point computed from the snapshot available on that date — free on the demo ticker, no signup.
Or start with one ticker's whole picture: the stock checkup puts seven dimensions of public disclosure — financial quality, valuation percentile, revenue momentum, ownership concentration, insider pledges, disposal filings and buybacks — on a single page, each as a traffic light with one plain sentence and no composite score.
Tracking who holds the float? The cross-source ownership timeline lays weekly mega-holder distribution, insider transfer filings, director pledge changes and buyback windows on one axis, each placed on the date it became available rather than the date it describes — free on the demo ticker.
Weighing several names against each other? The multi-stock comparison dashboard puts up to four tickers on one axis across ROE, gross margin, revenue growth, valuation percentile, dividend yield and mega-holder ratio, with a latest-value table underneath — and the comparison set is carried in the URL, so a shortlist is a link you can send.
New to this market? Taiwan requires every listed company to file its previous month's revenue by the 10th, which most markets have no equivalent for. The monthly revenue momentum chart turns that filing into a level read — 3-month against 12-month moving averages with year-over-year growth — free on the demo ticker.
Running a backtest on Taiwan names? An ex-dividend date drops the price by the payout, and a return computed on raw prices books that as a loss. The backward-adjusted price chart puts the adjusted candles against the raw close with the per-day adjustment factor, and puts a number on how much a raw-price return understates the holding period — free on the demo ticker.
Watching leverage rather than valuation? The market-wide margin maintenance ratio stopped being published as an official statistic and largely vanished from the public record. The market margin maintenance ratio is rebuilt here from the public daily files, unbroken from 2004, with the 140 percent warning band drawn and every approximation written out — free, no signup. Its companion on the sentiment and participation side is the market heat dashboard: day trading share of turnover for both boards since 2014, and the advance-decline line drawn against the index so you can see whether a rally is broad or has narrowed to a handful of names.
Both of those read the market from the inside. For the backdrop they sit in, the macro regime chart puts the official five-colour business cycle signal behind narrow and broad money supply growth, with every crossing between the two marked. The cycle series runs monthly from 1986, which makes it the longest continuous history on this site — free, no signup, no per-ticker limits.
Working bottom-up instead? The financial ratio time series puts fifty-three ratios per ticker — profitability, balance sheet safety, growth and a DuPont decomposition with its ROE cross-check — recomputed from the official quarterly filings, with every observation gated on the legal disclosure date so a screen built on it cannot read a quarter before it was public.
And when the question is finally what the thing is worth, three valuation models run side by side on the same ticker — discounted cash flow, Gordon dividend discount and the Peter Lynch fair value — each returning a pessimistic to optimistic range rather than a single figure, with growth, discount rate and terminal growth on sliders that map straight onto the endpoint's query parameters. The assumption set is printed next to the answer, including which inputs are derived rather than filed.
Starting from the other end — no ticker in mind yet — the cross-source screener intersects consecutive sessions of foreign institutional net buying with a cheap valuation percentile and returns the numbers that caused each hit. It also reports how many tickers were actually scanned on every call, because the two underlying panels do not cover the same universe and an empty table should read as coverage rather than as a verdict on the market. We publish our own negative backtest of the combination alongside it.
Back on the ownership side, the timeline above reads filings as events. For the level they add up to, the insider and mega-holder ownership levels chart plots the combined board shareholding filed every month — a running level, which most markets do not publish — against the weekly share of the float held by accounts over a thousand lots, with the insider pledge ratio on its own axis. Where the percentage denominator is derived rather than filed, the chart says so on every call.
Valuation
Where a stock sits against its own history
Taiwan Stock PER Band & Valuation Percentile
Free Taiwan-stock PER band lookup: where a stock's PER / PBR / dividend yield sits inside its own historical distribution (percentile + z-score). Official-source direct, look-ahead-bias-free.
Dividend Yield River
Taiwan dividend-yield river API: cash yield versus its own historical percentile bands, built from daily official snapshots with as-of versions for backtests. Deterministic, open formulas. A data tool, not dividend-stock picks.
Taiwan Point-in-Time Backtest Data Bundle
Taiwan backtest data bundle: market-wide panel with a snapshot_date on every row, monthly revenue aligned to actual announcement dates, statements gated by statutory deadlines — the data-layer fix for look-ahead bias. Parquet.
Ownership and flows
Who is buying, who is stepping back
Taiwan Shareholder Distribution (TDCC 15 Brackets)
Taiwan shareholder-distribution lookup: official TDCC 15-bracket data structured into mega-holder ratio, retail ratio and concentration change — weekly, snapshot-stamped. A data tool, not a signal.
Taiwan Institutional Net-Buy Streak
Free Taiwan institutional net-buy streak lookup: foreign / trust / dealer consecutive same-sign days, refreshed daily from official sources, snapshot-stamped. A data tool, not a trading signal.
Chip Concentration
Taiwan chip-concentration API: HHI concentration and mega-holder weekly deltas computed from official weekly holder-distribution filings. Deterministic, open formulas, availability-date stamped. A data tool, not stock picks.
Institutional Cost Line
Taiwan institutional cost-line API: volume-weighted cost estimate from official daily institutional flows and closing prices. Open methodology, deterministic, explicitly labelled an estimate. A data tool — not a support guarantee.
Retail–Mega Holder Divergence
Taiwan retail–mega divergence API: weekly deltas of mega-holder (1,000+ lot) vs retail share computed from official holder-distribution filings. Deterministic, availability-date stamped. A data tool, not stock picks.
Institutional Net-Buy Intensity
Taiwan net-buy intensity API: official daily institutional flows normalized by trading volume, returning rolling intensity and z-scores comparable across large and small caps. Deterministic, open formulas. A data tool, not stock picks.
Fundamentals
Whether the business actually earns its keep
Taiwan Piotroski F-Score
Taiwan Piotroski F-score: nine financial-strength checks computed in-house from official quarterly filings — open formula, per-line contributions, snapshot-aware.
Taiwan Altman Z-Score
Taiwan Altman Z-score bankruptcy risk: manufacturing Z' / non-manufacturing Z'' dual formula auto-applied, financials excluded (formula N/A). Computed from official statements, open formula.
Beneish M-score
Beneish M-score API for Taiwan stocks: the 1999 academic 8-variable model for earnings-manipulation risk, every variable disclosed individually, unavailable inputs honestly imputed neutral. Recomputed deterministically from filings. A data tool, not an accusation.
Taiwan Monthly Revenue
Taiwan monthly revenue API: official filings structured into YoY, MoM and momentum series, each row stamped with its disclosure date. Historical queries return as-of versions — no lookahead on unpublished revenue.
Taiwan Structured Earnings-Call Summaries
Taiwan structured earnings-call summaries: an LLM extracts guidance, capex plans and margin outlook into structured JSON — Chinese-language calls made machine-readable.
Research tools
Screening, indicators and event base rates
Taiwan Multi-Factor Stock Screener
Taiwan multi-factor screener API: combine PER percentile, institutional streaks, holder concentration and revenue momentum into one query. Deterministic filtering with transparent hit details. A data tool, not stock picks.
Taiwan 16-Indicator Library
Taiwan 16-indicator API: RSI, MACD, KD, Bollinger, OBV plus institutional cost line, chip concentration, PER band and revenue-YoY z-score — deterministically computed, open formulas, methodology attached. A data tool, not investment advice.
Taiwan Event-Study Stats
Taiwan event-study stats: historical abnormal-return base rates for 20+ event types (institutional streaks, insider filings, block trades, pledges, dividends), CAR / BHAR dual benchmark. Statistics, not predictions.
Taiwan Smart-Money Timeline
Taiwan smart-money timeline: aligns foreign/trust flows, insider filings, block trades, pledges and holder concentration from multiple official sources into one dated timeline.
How this library is organised
Grouped by the question you are asking
The five groups map to the questions a researcher actually asks — who is buying, is it expensive, does the business earn, is my backtest clean, what is the macro backdrop — not to how the data is stored.
Every entry links to something you can run now
Cards lead to working tools, not documentation. Type a ticker and the chart draws. The demo ticker runs with no signup, so you can see the shape of the data before deciding whether to go deeper.
Every historical point is the version available that day
All series here sit on the snapshot layer: a figure appears on a date only if it was computable on that date, and later corrections are not folded back. That is the foundation of the whole library, not a premium add-on on a few pages.
Limits are stated on the page, not buried in terms
Each page says what it can and cannot answer. Where our own backtests found no significant edge in a popular combination, the page says so.
Terms in plain English
Common questions
Where does the data come from?
Taiwan's official public disclosures: the exchange, the OTC market, the public filing observatory and the central depository. Every number is either reproduced as published or derived by a formula we document on the feature page, so you can recompute it from the raw filing.
How much of this is free?
Every page in the library opens and runs its full chart on the demo ticker with no signup. Single-indicator queries are open on the whole universe for free accounts. As-of history, arbitrary cross-source screening conditions and bulk market-wide pulls are paid tiers.
Why do your numbers differ from my broker's terminal?
Most terminals show the latest restated figures — corrections, dividend adjustments and share-count changes get folded back into history. Here each historical point keeps the version that was actually publishable on that date. That difference is the line between an honest backtest and one that peeked at the future.
Can I use these as trading signals?
No, and we do not sell signals. Everything here is a historical statistic or a current disclosure rendered as-is. Even the event-study base rates only say what happened on average after that kind of filing. Where our own backtests found no statistically significant edge, the feature page says so in plain words.
What happens when a value is missing?
It stays missing. Nothing is filled, interpolated or carried forward. Sources publish at genuinely different rhythms — quarterly for statements, monthly for revenue, daily for market data — and forcing them onto one grid invents numbers that never existed.