How We Validate Macro Factors: Technical Deep Dive #7
michaeljiangmingfeng-debug
started this conversation in
Show and tell
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
Every strategy starts with a macro factor question. Here's how we validate each one:
CNH-CHAU: Does offshore RMB reflect real cross-border capital flows?
Validation: Track CNH vs. CNY spreads against actual cross-border transaction data, FDI flows, capital account data. The offshore RMB market is where global capital votes on China's economy in real-time. This isn't a pattern — it's capital voting.
DIP-A: Is limit-down an accurate panic gauge?
Validation: Compare limit-down frequencies against retail positioning data, margin liquidation events. China's limit-down rule (unique globally) triggers forced selling cascades. When 1000+ stocks hit limit-down, it's not noise — it's systematic panic. 5 trades in 6 years, about once per year.
DIP-US: Does VIX spiking to extremes create mean-reversion edges?
Validation: Measure realized volatility post-spike vs. implied. Result: 21 trades since 2008, 100% win rate historically. Buy TQQQ at extreme VIX, hold until reversal signal.
IF-IC: Can capital controls create two independent fund channels?
Validation: Measure correlation between IF (large-cap futures) and IC (small-cap futures) during regime shifts. Under capital controls, domestic and foreign capital flow through separate channels, creating rotation opportunities across four regimes.
The Key Difference
We don't ask: "What pattern maximizes returns?" We ask: "What is the real macro factor, and is it persistent?"
Patterns break when regimes shift. Macro factors are structural — they change slowly because they're rooted in economic forces, not statistical coincidences.
This is why we publish full mechanism descriptions on AI Hall. We want you to validate these factors independently.
Philosophy: No pure statistical development. Every strategy preserves its character — we won't distort it to chase dead benchmarks.
Beta Was this translation helpful? Give feedback.
All reactions