The thinking habits of macro researchers fall into three types: logical-deduction, grassroots-fact, and data-regularity. Each, used alone, has a structural fatal flaw; the optimal strategy is to use empirical regularity for the framework hypothesis, logic for deduction, and data for verification. When the three types of conclusions conflict, this framework holds that empirical regularity should be trusted first — because empirical regularity is, in essence, high probability.
The Framework As It Stands
This section is organized from the compiled research draft: it preserves the original framework’s structure, terminology, and key formulations, with editorial bridging and external factual annotations; diagrams are drawn by the compiler following the structure of the original text.
The three modes of thinking and their respective fatal flaws
Logical-deduction type: excited by A→B chains of derivation; must have logic to believe. Fatal flaw: too much logic can lead to different conclusions — one can list ten arguments each for why China’s economy is good or bad, with no way to judge the current dominant logic.
Grassroots-fact type: excited by field research; must experience firsthand to have intuition. Fatal flaw: small-sample risk and selection bias — one factory does not represent one city, and researchers tend to pick samples that fit their own priors.
Data-regularity type: excited by historically similar phases; must find a regularity to believe. Fatal flaw: “this time is different” — the Popperian counterexample: a single counterexample can overturn the entire logic.
The order of priority in combining the three
Optimal = combining all three, but in sequence, not in parallel:
① Form the framework hypothesis based on empirical regularity (starting point)
↓
② Deduce based on logical soundness
↓
③ Verify with macro and micro data
Conflict adjudication: when the three approaches yield conflicting conclusions, empirical regularity should be trusted first.
The core methodological proposition
Empirical regularity outweighs logic; so-called empirical regularity is, in essence, high probability.
When it is uncertain which logic is correct, seek relatively cold empirical regularities and find from them the high probability of future events — this is the fundamental basis for adjudicating uncertainty.
A supporting methodological case (for the full case library see Finding High Probability Through Empirical Regularities: Six Case Studies in Practice)
The late-2015 commodity bottom (one example to make the point): the year-on-year change of the CRB industrial raw materials index basically oscillated within ±20% over 60 years, breaking below negative 20% only three times in history (the 1950s / 1975 crude oil / 2008); at the end of 2015 it touched that level again; empirical regularity showed that from negative 20% to the final bottom takes only three to four months. Most people were extremely pessimistic at the time; empirical regularity pointed, against the tide, to a bottom, and was later validated.
Note: the case reflects the judgment made in 2015 at the time, presented as a methodological demonstration, not a current market conclusion.
Compiler’s Perspective
Coordinates: Cognitive Algorithms · Fa (methods) · Why It Is So
Connecting to the Dao layer
This entry’s distinctive increment: the failure mode of the three types of thinking is not “using the wrong tool,” but “making a single tool the adjudicator.” The most common failure of the logic type: within one macro judgment, being able to find ten arguments supporting both the bulls and the bears — this is not analytical rigor, but analytical breakdown. The failure of the grassroots type is more insidious: many field trips are made, yet the conclusions systematically lean toward one direction, because the visiting route itself has already been filtered by priors. The failure of the data type is using historical correlation coefficients to forecast, then not knowing what to do when the first counterexample appears.
The concrete wrong move of the old approach: upon receiving a macro signal, the first reaction is to “tell the logic” — first construct an A→B→C derivation chain, and once the logic is self-consistent, draw the conclusion. The framework’s correction: first find the historical extremes or similar positions of this indicator, see what the empirical regularity points to, then use logic to verify whether the driver still holds. With a different order, the stability of the conclusion is entirely different.
The practical meaning of the proposition “empirical regularity outweighs logic”: it is not “don’t use logic,” but that logic always comes after empirical regularity as verification, rather than before empirical regularity as adjudication. At the end of 2015, most of the logic on commodities pointed to further declines (capacity reduction still ongoing, demand weak) — it was precisely this logic that prevented most people from recognizing the bottom; whereas the CRB’s 60-year ±20% empirical regularity forcefully gave a time anchor of “only three to four months to the bottom,” against the consensus logic of the time.
See Also
-
The Nature of Macro Research and the Sense of Position — upstream master outline: the essence of research = narrowing the range of choices / sense of position
-
Finding High Probability Through Empirical Regularities: Six Case Studies in Practice — case implementation: five-element breakdown of six practical cases
-
The Cycle Superposition Framework: Adjudicating the Economy’s Overall Position — downstream: how to locate cycle coordinates under the guidance of empirical regularity
Sources
- Compiled research draft · collected 2026-07
Compiled draft z-0112 · collected 2026-07