Finding High Probability Through Empirical Regularities: Six Case Studies in Practice

The “finding high probability through empirical regularities” method applied in practice: when key coordinate indicators reach historical extremes or comparable low positions, the method identifies future high probabilities via mean reversion or strong-correlation regularities, then cross-validates with logic and multiple shadow indicators. Six case studies cover commodity bottoms, economic turning points, interest rate floors, Juglar cycle initiation, foreign-reserve/Treasury correlation, and two rounds of export slowdown.

The Framework As It Stands

This section is compiled from the research draft: the original framework’s structure, terminology, and key formulations are retained, including editorial bridging and supplementary external fact notes; charts are drawn by the compiler following the original framework structure.

Unified Five-Element Template

Each case can be broken into five elements: key indicator → historical extreme / comparable position → empirical regularity (often with mean reversion) → logical validation → result. The common methodological principle: find coordinate indicators with mean-reversion or strong-correlation properties, check whether the current reading falls at the position the regularity points to, and use this to identify the future high probability; empirical regularity is, in essence, high probability.

① Key indicator (mean reversion / strong correlation)
       ↓
② Historical extreme / comparable position
       ↓
③ Empirical regularity → high probability
       ↓
④ Logical validation (whether the driving factor still holds)
       ↓
⑤ Result + multiple shadow indicators pointing in the same direction

Case 1 · 2015 Commodity Bottom

Key indicator: CRB Industrial Materials Index, year-on-year.
Historical position: Over 60 years, the reading has stayed broadly within ±20%; only three times in history has it broken below −20% (the 1950s / 1975 oil / 2008). In late 2015 it reached that level again.
Empirical regularity: From −20% to the ultimate bottom has taken only three to four months.
Logical validation: The driving factor was slowing BRIC growth rates. By late 2015 China had stabilized at 6.5% and India had recovered to 7%, meaning the earlier downward expectations had already been priced in.
Result: Subsequently confirmed.

Case 2 · Q3 2016 Economic Turning Point Upward

Key indicator: Industrial-enterprise finished-goods inventory, year-on-year.
Historical position: Three historical lows — 2002 / 2009 / Q3 2016 — all at the same level.
Empirical regularity: By the principle of mean reversion, when inventory reaches a historical low, economic rebound is the high probability (ROE / ROIC converge toward equilibrium).
Logical validation + result: Confirmed after multiple indicators converged.

Case 3 · Q3 2016 Interest Rate Floor

Key indicator: Finished-goods inventory and interest rates have been highly correlated for 15 years, with rates lagging inventory by about half a cycle; inventory had already rebounded for more than two months.
Empirical regularity: Interest rate = expected investment return + inflation compensation = nominal growth rate; inventory bottom → rates rise.
Result: Subsequently confirmed.

Case 4 · Late 2016 Juglar Cycle Launch

Key indicator: Equipment utilization among 5,000 industrial enterprises, one cycle every 8–10 years (The Juglar Cycle: The Equipment Capex Mid-Cycle and Its ROE Essence).
Comparable position: Rise in 1998 / peak in 2006 / trough in 2009 / peak in 2011; the late-2016 bottom was squarely at the empirical trough.
Empirical regularity: The core of the normal distribution for Four Trillion Stimulus depreciation is 7–8 years; the cycle completed at end-2016 → a new round of investment is the high probability.
Logical validation: Cross-validated with capex / profit margin / fixed-asset turnover / micro-level leverage ratio all pointing the same direction.

Case 5 · 2018 Ten-Year Treasury (Foreign-Reserve/Treasury Correlation)

Key indicator: Foreign exchange reserves growth and ten-year Treasury yields are positively correlated (not spurious: foreign reserves are a shadow indicator of nominal growth; when reserves are high, policy tightens liquidity).
Empirical regularity + result: Using goods/services trade to capture the reserve trend, the ten-year Treasury yield fell from a high of 4.0% in 2018, consistent with the regularity.

Case 6 · Two Rounds of Slowdown in 2018 (Exports–Nominal GDP)

Key indicator: China exports and nominal GDP are historically correlated — when exports are strong, the economy is generally not weak; when exports are poor, the economy is generally not strong.
Historical position: Exports were at a high level in early 2018.
Empirical regularity + framework: Two stages distinguished — Q4 2018: orders / expectations deteriorated; H1 2019: export volumes materialized on the downside; the two stages correspond respectively to two rounds of nominal GDP slowdown.

Closing

Common thread across all six cases: find coordinate indicators with mean-reversion / strong-correlation properties → benchmark against historical extremes or comparable positions → derive future high probability → validate with logic + multiple corroborating indicators. Empirical regularity is, in essence, high probability.

Compiler’s Perspective

Coordinates: Cognitive Algorithm · Fa · Its Place in the Whole

Bridge layer

This entry’s position within the three-piece macro methodology group (z-0107 / z-0112 / z-0098): if The Nature of Macro Research and the Sense of Position answers “what is research for,” and Three Modes of Macro Research Thinking and the Primacy of Empirical Regularity answers “which tools to use and in what order,” then this entry answers “what the tool actually looks like” — the six cases are reusable prototypes of the five-element template.

This entry’s exclusive increment: the six cases contain a hidden asymmetric structure — Cases 1–4 derive their high probability from “mean reversion after a coordinate indicator reaches a historical extreme,” while Cases 5–6 derive theirs from “a stable long-run correlation between two macro variables.” The former is vertical comparison (the indicator’s own historical extreme), the latter is lateral transmission (one variable leading or coinciding with another). Conflating these two mechanisms leads to misapplication: the mean-reversion type requires assessing “whether the extreme is extreme enough,” while the correlation-transmission type requires first ruling out spurious correlation (the foreign-reserve/Treasury case specifically emphasizes it is not spurious).

The concrete error under the old approach: seeing CRB fall to a historical low, the first instinct is “keep bearish — the de-capacity logic is still in play.” The five-element template’s corrective is: first benchmark against the 60-year historical frequency (only three breakdowns below −20%, each reaching bottom within three to four months); then ask about logic — “has the downward expectation already been priced in?” If logic also confirms, then high probability is established, and a counter-consensus stance then has a basis. The more corroborating indicators, the more credible — but the direction still comes from empirical regularity first, with logic providing subsequent validation.

Counter-consensus reasoning appears in at least four of the six cases (extreme pessimism on commodities in 2015 / most forecasting rates to 1.6% in 2016 / Juglar not yet priced in during Q3 2016 / ten-year Treasury declining from a high in 2018) — empirical regularity provided the basis for counter-consensus at those junctures, not logical deduction. That is precisely the practical value of “primacy of empirical regularity.”

Source

Compiled research draft · incorporated July 2026

Compiled draft z-0098 · incorporated July 2026

Z-Finance

Explorer