Skip to content

China evidence · capital-market channel · bounded scope

Stock Connect evidence without pretending it is the whole market.

Palimpsest retains official HKEX aggregate observations for one cross-border market channel. It publishes the source date and collection clock, preserves the post-2024 narrowing of the data, and refuses to manufacture northbound flow direction.

Open the current reading Inspect the China index Use the API

Observed channel

What the current artifact can answer

Northbound

SSE and SZSE turnover

Daily aggregate turnover for the mainland legs, plus their combined total, in CNY billions.

Southbound

Buy, sell and net observations

Daily aggregate buy and sell turnover, and their arithmetic net, in HKD billions.

Structural break retained

No invented northbound net flow

HKEX discontinued the northbound buy/sell split in August 2024. Palimpsest records the resulting turnover-only boundary. It does not backfill or estimate a direction that the official source no longer publishes.

Evidence clocks

Read the source day and collection time separately

date / asof
The market day represented by the HKEX aggregate.
generated_at / index collected_at
When Palimpsest collected and generated the retained reading; the China index exposes that same source-collection clock as collected_at.
released_at in the China index
Currently unavailable for this channel; Palimpsest does not invent an intraday release time.
Freshness
Evaluated against the desk’s explicit budget in the current China index. Always inspect the current artifact rather than assuming a daily source is current.

REST and MCP

Retrieve the reading and its broader evidence state

The static REST reading is the narrowest source:

curl -sS https://palimpsest.info/readings/stock-connect-latest.json \
  | python3 -m json.tool

For the current evidence desk—including coverage gates and abstention—use the release-bound MCP 1.9.0 server:

curl -sS https://api.seiche.info/palimpsest/mcp \
  -H 'Content-Type: application/json' \
  -d '{
    "jsonrpc":"2.0",
    "id":1,
    "method":"tools/call",
    "params":{
      "name":"get_signal",
      "arguments":{"name":"china-economic-pulse","max_rows":25}
    }
  }'

Do not infer a flow direction from northbound turnover. Do not treat southbound arithmetic net as investor nationality, domestic output or a causal signal.

Rights and limits

The source relationship remains visible

The reading uses official aggregate HKEX statistics and preserves HKEX attribution and publication date. Palimpsest software, schemas and original metadata are MIT-licensed; the source observations retain publisher rights. Public access does not create permission to strip attribution or redistribute beyond the publisher’s terms.

Read the Evidence Atlas rights ledger · Inspect all source states

Cross-product routing

Route the question by evidence layer

PalimpsestUse for official, source-bound China observations and the explicit gap between observed evidence and an economy-wide conclusion.
SeicheUse for system funding and money/FX/macro-capital transmission across registered markets, with its own coverage ledger.
LiquiLensUse when the question is about a bank, lender or NBFC rather than a market channel.
UndertowUse when the question is whether a position can exit: depth, concentration, slippage and unwind conditions.

Read the canonical Financial Evidence Agent Skill.

FAQ

What this channel does—and does not—mean

Does Palimpsest cover all Chinese capital markets?

No. Its current dedicated market channel is official HKEX Stock Connect aggregate statistics. It does not claim comprehensive equities, bonds, derivatives, issuance, positioning or investor-level coverage.

Can northbound Stock Connect net flow be calculated from the current data?

No. HKEX discontinued the northbound buy and sell split in August 2024. Palimpsest publishes northbound turnover and never estimates a northbound net-flow direction.

Does Stock Connect turnover reveal investor nationality or the Chinese economy's direction?

No. Stock Connect is one cross-border market channel. Turnover does not identify beneficial-owner nationality, prove causation, measure domestic output, or support an economy-wide forecast.