跨市场传导

全球市场 Lead-Lag 检查表

分层检查美股、日韩、台湾、欧洲与 A/H 市场之间的隔夜信号、机制映射和时效边界。

返回方法论索引在 GitHub 查看源文档 ↗
REVIEWED METHODOLOGY查看源文档
使用边界这是用于组织调查、证据和复核的方法框架;重要事实需以最新主源验证,财经内容不构成投资建议。

Global Lead-Lag Checklist

Use this reference when the user asks how overnight markets may affect today's or tomorrow's sector move.

Time Windows

For A/H or China-related next-session reads, pin exact dates and use this order:

  1. local_close: What happened in the local sector before the overnight window?
  2. after_local_close: company filings, exchange notices, policy/news after close.
  3. us_session: US indices, sector ETFs, ADRs, key large-cap peers, earnings/news.
  4. us_after_hours: earnings, guidance, mega-cap conference calls, futures.
  5. asia_preopen: Japan, Korea, Taiwan, Singapore, commodity futures, FX/rates if relevant.
  6. local_preopen: domestic media, exchange filings, broker notes, regulatory comments.

Do not blur "today", "yesterday", and "tomorrow"; always name the calendar dates and markets.

Signal Types

Signal Meaning Reliability
same_chain_confirmation A global leader confirms the same demand/order/pricing chain High when from earnings, orders, guidance, official capex
same_chain_shock A global leader weakens a required link in the same chain High when primary source, medium when reported
adjacent_beta Similar sector moves but no direct supply-chain link Medium
macro_risk Rates, FX, liquidity, index risk, geopolitical risk dominate Medium to high depending on source
local_only Domestic policy, positioning, quota, index inclusion, or local news dominates Medium
insufficient Movement lacks a sourced mechanism Low

Sector-Specific Lead Markets

Use these as starting maps, then verify current facts.

Sector/theme Lead signals to check
AI hardware / AI capex Nasdaq, SOX, NVIDIA, AMD, Broadcom, Micron, Marvell, Meta, Microsoft, Alphabet, Amazon, Oracle, CoreWeave/Nebius where relevant; Samsung Electronics, SK hynix, TSMC, Tokyo Electron, Advantest; HBM, memory, optics, PCB/CCL, power/cooling news
Robotics / physical AI Tesla, NVIDIA robotics/edge AI signals, industrial automation peers such as FANUC, Yaskawa, Keyence, SMC, THK, Harmonic Drive, Nabtesco; robotaxi/autonomous driving policy; actuator, reducer, sensor, servo and battery suppliers
EV / battery chain Tesla, BYD, CATL, lithium carbonate, nickel/cobalt, charger/power semiconductor peers, subsidy and tariff news
Memory / storage Micron, Samsung Electronics, SK hynix, DRAM/NAND/HBM contract and spot price indicators, capex plans
CPO / optics / networking NVIDIA/ASIC networking demand, Broadcom, Marvell, Arista, Coherent/Lumentum where relevant, hyperscaler capex
PCB / CCL / materials AI server buildout, copper foil, glass fiber, resin, laminate pricing, capacity announcements, customer qualification timelines
Commodities / resources Futures curve, spot price, inventory, USD, rates, China demand, policy, supply disruption
Internet / platform Nasdaq, sector ADRs, large platform earnings, ad/ecommerce/cloud metrics, regulation
Pharma / biotech FDA/EMA/NMPA events, trial readouts, peer drug data, reimbursement and pricing policy

Overnight Read Template

隔夜结论:
<same_chain_signal / adjacent_beta / macro_risk / local_only / insufficient>

日期窗口:
- 本地收盘:
- 美股时段:
- 亚洲盘前:

关键大票/指数:
| 市场 | 标的/指数 | 变化 | 原因 | 对本地板块传导 |
| --- | --- | --- | --- | --- |

链条映射:
| 全球信号 | 打到本地产业链哪一环 | 传导强度 | 需要验证 |
| --- | --- | --- | --- |

今日/明日情景:
| 情景 | 触发条件 | 板块含义 | 验证点 |
| --- | --- | --- | --- |

Interpretation Rules

  • Give more weight to a large-cap move when it comes with guidance, capex, orders, pricing, or product-cycle evidence.
  • Give less weight to a move when it only reflects index beta, positioning, or one stock's idiosyncratic issue.
  • Treat a customer's "unused capacity" or "rental" story as a possible supply/pricing narrative shock, not as proof of order cancellation unless primary evidence confirms it.
  • Treat Asian semiconductor moves as useful confirmation only when they occur before the local market opens and are tied to the same chain link.
  • Check whether the local sector already priced in the overnight move before assuming fresh downside/upside.
  • When evidence is mixed, write multiple scenarios with triggers instead of a single prediction.