Alpha Vantage Asian Stock Exchanges: Symbol Suffixes

Which Asian stock exchanges does Alpha Vantage support?

Coverage of Asian markets is uneven. The documented examples for international equities include London (.LON), Toronto (.TRT and .TRV), XETRA (.DEX), India BSE (.BSE), Shanghai (.SHH) and Shenzhen (.SHZ). Those are the only Asian suffixes with documented examples for the daily time series endpoints.

Tokyo (.T or .TYO) and Hong Kong (.HK) are not in that documented list. A ticker from those markets may return an error, an empty object, or occasionally data, depending on what the vendor currently carries. Treat them as unverified until a probe says otherwise.

This is why a setup that works for London, India and Australia can fail for Japan and Hong Kong. The code path is identical, but the symbol universe behind it is not.

What is the correct symbol format for Alpha Vantage?

The format is the local ticker followed by a dot and the exchange suffix: SYMBOL.EXCHANGE. Prefix forms such as TYO.4446, HKG:0700 or SHA:600104 are not parsed as exchange qualifiers. The API treats the whole string as an unknown symbol and returns an error.

MIC codes (XTKS, XHKG, XSHG, XSHE) are also wrong. They are ISO market identifiers, not Alpha Vantage suffixes.

Each Asian market has its own ticker rules, and getting the digits wrong is as common as getting the suffix wrong.

How do I verify a symbol with SYMBOL_SEARCH?

SYMBOL_SEARCH is a fuzzy search over the vendor's symbol universe. If a ticker is not returned there, TIME_SERIES_DAILY will not work for it either, which makes this the fastest first check.

Search by company name or code, then read the matches. Each match includes the exact symbol string to use, plus the name, type, region, marketOpen, marketClose, timezone, currency and a matchScore.

1. Call SYMBOL_SEARCH with keywords set to the company name (for example toyota) and, separately, the code (for example 7203).
2. Read the 2. name and 4. region fields to confirm the match is the right company and market.
3. Copy the 1. symbol value exactly as returned, including the suffix, and use it in your data call.
4. Check 8. currency (JPY, HKD, CNY) so you do not mix currencies in a portfolio.
5. If bestMatches is empty for both the name and the code, treat the market as unsupported for that ticker.

https://www.alphavantage.co/query?function=SYMBOL_SEARCH&keywords=tencent&apikey=YOUR_KEY

# Response shape (fields abbreviated)
{
  "bestMatches": [
    {
      "1. symbol": "600104.SHH",
      "2. name": "...",
      "3. type": "Equity",
      "4. region": "Shanghai",
      "5. marketOpen": "09:30",
      "6. marketClose": "15:00",
      "7. timezone": "UTC+08",
      "8. currency": "CNY",
      "9. matchScore": "..."
    }
  ]
}

Why do I get an error, an empty response or an Information message?

Alpha Vantage returns HTTP 200 for almost everything, including failures. Checking the status code tells you nothing, so you have to inspect the JSON body keys.

Each body pattern points to a different cause, and the fix differs for each.

Python probe: which suffix returns data?

This script tries a code against several candidate suffixes and classifies each response by body, not by status code. It stops early if the quota is exhausted so it does not waste calls.

Run it once per market with a known liquid ticker, then record the results with the date, since coverage can change.

import time, requests

API = "https://www.alphavantage.co/query"
KEY = "YOUR_KEY"

def classify(body):
    if "Error Message" in body:
        return "invalid_symbol"
    if "Information" in body or "Note" in body:
        msg = (body.get("Information") or body.get("Note")).lower()
        return "premium" if "premium" in msg else "rate_limit"
    if "Time Series (Daily)" in body:
        return "ok"
    return "empty"

def probe(code, suffixes, function="TIME_SERIES_DAILY"):
    results = {}
    for sfx in suffixes:
        sym = f"{code}{sfx}"
        r = requests.get(API, params={
            "function": function, "symbol": sym,
            "outputsize": "compact", "apikey": KEY}, timeout=30)
        r.raise_for_status()
        body = r.json()
        status = classify(body)
        results[sym] = status
        if status == "ok":
            ts = body["Time Series (Daily)"]
            last = max(ts)
            print(sym, "last bar", last, ts[last]["4. close"])
        if status == "rate_limit":
            print("Quota hit, stopping")
            break
        time.sleep(1.2)  # stay under burst limits
    return results

print(probe("7203", [".T", ".TYO"]))
print(probe("0700", [".HK"]))
print(probe("600104", [".SHH", ".SHZ"]))
print(probe("000002", [".SHZ", ".SHH"]))

Daily vs adjusted data, currency and end-of-day caveats

TIME_SERIES_DAILY returns raw open, high, low, close and volume. TIME_SERIES_DAILY_ADJUSTED adds an adjusted close plus dividend and split columns, but it has been a premium-gated endpoint, so a free key may return an Information notice instead of data. Without adjusted prices, backtests will show false jumps at splits and distort returns on dividend-paying stocks.

The free outputsize=compact returns roughly the latest 100 daily bars. Full history (outputsize=full) is a premium parameter on many plans, so long backtests may require a paid key.

Data is end-of-day. Asian sessions close before Europe and the US open, so a bar may appear hours later than the local close, and it can be missing on local holidays. Timestamps are plain dates in the exchange's local calendar; do not join them against US-calendar data without aligning trading days.

Each series is in the local currency (JPY, HKD, CNY). Convert with a separate FX series if you aggregate across markets, and keep the conversion date aligned to the bar date to avoid look-ahead.

What are the alternatives when a market is unsupported?

If the probe shows no Tokyo or Hong Kong data, move that market to a provider that covers it explicitly. Keep a small adapter layer that maps your internal ticker to each vendor's format, so switching costs one function instead of a rewrite.

Licensing matters as much as coverage. Scraped or unofficial feeds are fine for research but are risky in anything redistributed or commercial.