Prompt gallery

Prompts for Taiwan Market Data — Ready to Paste

Connected your AI and not sure what to ask first? Start with one of these and edit it. Each one says whether it needs an API key.

Not connected yet — three ways to connect.

Confirm the connection works

No key needed

No key, no account. If it returns prices, the connection works. The keyless tickers are 2330, 2317, 2454, 0050, 2603.

Using TW Market Data, get the last 5 sessions of closing prices for 2330. Show it as a table and state which dataset you used and the data date of each row.

Uses: twse-daily-price

Revenue momentum

Needs an API key

Monthly revenue from 2010, 333,036 rows. The shortest path into fundamentals.

Using TW Market Data, pull 24 months of monthly revenue for 2330, compute YoY for each month, point out any run of three consecutive months of acceleration or deceleration, and say which field you used.

Uses: monthly-revenue

What the institutions did

Needs an API key

Daily net buy/sell by foreign investors, investment trusts and dealers — the most-asked question on the flow side.

Using TW Market Data, pull 20 sessions of institutional net buy/sell for 2330. Break it out by foreign investors, investment trusts and dealers, and flag the days where the three disagreed.

Uses: institutional-flow

A backtest universe that includes delistings

Needs an API key

Price history is retained for 311 stopped-trading names. Most free sources drop them, which is what inflates a backtest.

Using TW Market Data, explain how to build a Taiwan equity backtest universe that INCLUDES delisted names: which datasets to use, how to tell whether a ticker stopped trading, and which names have an official delisting date on record versus which do not.

Uses: twse-daily-price · stock-delisting-lifecycle

Avoid look-ahead bias

Needs an API key

Every row carries a knowledge_date — when the figure became knowable. That is what makes a point-in-time backtest possible.

Using TW Market Data, pull monthly revenue for 2330 and explain: if I were making a decision on 2025-03-31, which months' revenue was already knowable at that point? Justify it from the fields in the response, not from assumption.

Uses: monthly-revenue

Ask about the holes first

No key needed

Gaps are reported in the response rather than filled in. Better to ask before the analysis than to discover it after.

Using TW Market Data, pull daily prices for 2330 and tell me whether the response reports any data_gaps and what source_role each row carries. If there are gaps, say so  do not interpolate.

Uses: twse-daily-price

Related articles