5D Library › Macro-Liquidity Regime Switching, Extreme Tail Dependence and Barbell Asset Allocation · 中文

Preface: Buffers, Tails and Regimes

Note

Risk notice: please read this first. This book presents mathematical models for teaching and research. It is not investment, trading, career, tax or legal advice. Backtests and simulations do not represent future returns; every return, price, reserve figure and economy in the book is fictional or simulated, not data from any real market or country. The book recommends no particular asset, exchange, currency or broker, and offers no method that “guarantees”, “always wins”, “cannot lose” or is “proven optimal”. Its models hold only under the assumptions stated; real markets can violate them at any time, and options and other derivatives can lose everything invested in them.

In calm years, stocks, bonds and commodities go their separate ways; in a few weeks of panic they seem suddenly to fall arm in arm. A country’s foreign-exchange reserves look like idle cost in normal times and far too small when capital suddenly leaves. An investor puts nine-tenths of her money in short-term government bills and one-tenth in options, believing this gives “a floor below and no ceiling above”. This book asks which of these intuitions are supported by mathematics and evidence, which need qualifying, and which are wrong.

Answering takes four layers of tools. The first is regimes: does the market switch between “calm” and “stressed”, and can a switch be recognised in time? The second is tails: how large and how frequent are extreme losses, and how can a quantity that is rarely observed be estimated from limited data? The third is dependence: do extreme losses arrive together, and why can’t the ordinary correlation coefficient answer that question? The fourth is buffers: under explicit assumptions, what protection can floor-protected portfolios and reserves provide, and at what cost?

What the book covers

Chapter 1 covers hidden Markov models and regime identification; Chapter 2 extreme value theory, VaR and ES and their backtesting; Chapter 3 copulas and tail dependence; Chapter 4 correlation in crises and vine copulas; Chapter 5 barbell allocation and stochastic dominance; Chapter 6 the adequacy and dynamics of sovereign reserves; Chapter 7 reads these beside the Book of Rites, the Mozi and the Book of Han; and Chapter 8 offers a four-step checklist and a pre-registered backtest.

The book neighbours several others in 5D Library; where they overlap, it only recaps briefly, cites the earlier book and does not repeat its programs. Book 19, Non-Ergodic Wealth Dynamics, the Kelly Criterion and Volatility Drag, treated time-average growth and position sizing. Book 20, Credit Cycles, Self-Organised Criticality and the Statistical Physics of Market Phase Transitions, treated GARCH, implied volatility and skew, and liquidity spirals (Chapter 2 only recaps GARCH). Book 28, Phase Transitions, Critical Slowing Down and Resilience in Complex Dynamical Systems, treated early-warning signals of critical transitions. Book 33, Scientific Methodology, Falsifiability and the Mathematical Statistics of the Replication Crisis, treated pre-registration, power and decision rules (Chapter 8 follows it). Book 38, Implied Volatility Surfaces, Skew Premia and Quantitative Risk Control of Vertical Spreads, treated the variance risk premium and the ever-normal granary (Chapters 5 and 7 only recap). Book 39, Liquidation Cascades, Order-Book Depth Depletion and Cointegrated Statistical Arbitrage, treated liquidation cascades and the sudden drying-up of market liquidity.

Claims corrected after checking

Several points in the author’s framework needed correcting or qualifying after checking. Each chapter says so where it arises, and the accompanying editorial notes list them together.

  1. Hamilton’s 1989 model. It is a Markov switch in the mean of quarterly US post-war real GNP growth, used to describe expansions and recessions, not a model of volatility regimes; applying the same framework to volatility regimes in returns is a later extension (Hamilton 1989; Ang and Bekaert 2002). His estimated persistence probabilities for expansion and recession were about 0.90 and 0.75, giving expected durations of about 10 and 4 quarters (Chapter 1).
  2. “Expected duration = 1/(1 − p_ii)”. The formula is right, but it holds only under the Markov (memoryless) assumption: in the model, a regime that has already lasted 100 days has the same expected remaining duration as one that has just begun. Even with the true states known, ten years of daily data give a 90 % interval for the duration of the stress regime running from 23 to 48 days; with Hamilton’s parameters and 131 quarters, the 90 % interval for recession duration is 2.5 to 6.6 quarters (Chapter 1).
  3. Regime identification. HMMs can divide history into regimes fairly accurately after the fact, but real-time (filtered) probabilities lag the after-the-fact (smoothed) ones: in the best case, with the true parameters known, the median delay is 2 days and the 90th percentile 7 days. When returns are themselves fat-tailed, both an information criterion and out-of-sample likelihood prefer a third “regime” that does not exist. So “an HMM can identify regimes” should read “an HMM gives a probabilistic division into regimes, given a number of states and distributional assumptions, and it is far clearer in hindsight than in real time” (Chapter 1).
  4. The Pickands–Balkema–de Haan theorem and the Hill estimator. The theorem concerns the limit of the excess distribution as the threshold approaches the right end of the distribution, and requires the distribution to lie in the domain of attraction of an extreme value law; in finite samples the shape parameter is biased (in the program, samples with a true ξ of 0.25 give a median estimate of about 0.15). The Hill estimator applies only to power-law tails with ξ > 0, and it produces plausible-looking numbers for normal data too (Chapter 2).
  5. Zero tail dependence of the Gaussian copula. The accurate statement is that for ρ < 1 the tail-dependence coefficient λ, a limit, equals 0. It does not mean that extreme losses never occur together: at the 1 % quantile, a Gaussian copula with correlation 0.5 still gives a conditional probability of 0.125 of a joint crash; it merely tends to zero further out. “The formula that killed Wall Street” was a magazine headline (Salmon 2009); MacKenzie and Spears found that practitioners had deep misgivings about such models before the crisis and that the crisis’s main mechanism was associated with simpler uses of the model (MacKenzie and Spears 2014). The book therefore does not say that “the Gaussian copula caused the financial crisis” (Chapter 3).
  6. The Clayton and Gumbel copulas. λ_L = 2^(−1/θ) and λ_U = 2 − 2^(1/θ) are correct; but the Clayton copula has zero upper-tail dependence and the Gumbel copula zero lower-tail dependence, so describing joint crashes with a Gumbel copula requires its rotated (survival) version (Chapter 3).
  7. “Correlations go to one in a crisis”. This is not a law. Forbes and Rigobon showed that correlations conditional on high-volatility periods are biased upwards by heteroskedasticity, and after adjustment found “virtually no contagion” in the crises of 1987, 1994 and 1997 (Forbes and Rigobon 2002); Longin and Solnik, using extreme value methods, found that the correlation of large negative returns is far higher than multivariate normality would imply, while that of large positive returns is not (Longin and Solnik 2001); and the Forbes–Rigobon adjustment itself depends on model assumptions (Corsetti et al. 2005). The book sets the two bodies of evidence side by side (Chapter 4).
  8. “Stochastic dominance” of the barbell. It cannot be proved in general. The book proves that when the barbell has a floor and the stock can lose more than that floor, the stock can neither first- nor second-order dominate the barbell; if, in addition, the options expire worthless more often than the stock falls below the floor (56.4 % against 20.5 % in the book’s model), the distribution functions must cross and the barbell cannot first-order dominate the stock either. Second-order dominance requires an expected return that is no lower, which depends on option prices. In the author’s jump-diffusion model, under four option prices (implied volatility 0.13 to 0.28), the barbell shows neither first- nor second-order dominance over the stock or over “90 % bills + 10 % stock”; when options are rich (0.22 and 0.28), it is itself second-order dominated by “90 % bills + 10 % stock” (Chapter 5).
  9. “A protected floor”. “You cannot lose more than 10 %” holds only in nominal terms and only if the bills do not default; bill interest makes the nominal maximum loss about 7.3 %, but inflation erodes real value. Taleb’s barbell is cited only as the source of the idea (Taleb 2012); the academic comparisons are capital-protected portfolios and CPPI (Black and Perold 1992), and CPPI with a multiplier of 6 ends below its floor in 6.4 % of simulated years (Chapter 5).
  10. The Greenspan–Guidotti rule and the ARA metric. The former was proposed by Argentina’s deputy finance minister Pablo Guidotti and popularised by Greenspan in a 1999 speech: usable reserves should cover external debt falling due within a year (Greenspan 1999). The IMF’s ARA metric, proposed in 2011 (International Monetary Fund 2011), gives emerging markets regime-dependent weights; the current weights are short-term external debt 30 %, other portfolio liabilities 20 % (15 % under a floating rate), broad money 10 % (5 %) and exports 10 % (5 %), with reserves of 100–150 % of the metric suggested (International Monetary Fund 2016). The weight on other liabilities was raised by 5 percentage points after 2011, when it was 15 % (10 %), so the current weights should not be attributed to the 2011 paper. The same fictional economy scores 108 % under a fixed rate and 163 % under a floating one (Chapter 6).
  11. The classical texts. The passage from the “Royal Regulations” checks out as “三年耕,必有一年之食;九年耕,必有三年之食。以三十年之通,虽有凶旱水溢,民无菜色”, preceded by “国无九年之蓄曰不足,无六年之蓄曰急,无三年之蓄曰国非其国也”; some online transcriptions miswrite 凶旱 as 凶早. The ever-normal granary comes from the Book of Han’s “Treatise on Food and Money” (Geng Shouchang, under Emperor Xuan), not from the Records of the Grand Historian (as checked in Book 38). The phrase “reckoning over thirty years” occurs twice in the “Royal Regulations”; the first time it concerns setting expenditure: “the state’s expenditure is set by reckoning over thirty years, measuring income to determine outgoings”, that is, budgeting on a long-run average (Chapter 7).

How to read the boxes

As in the other books of 5D Library, different kinds of statement are marked separately.

Evidence

Evidence boxes report published research findings, with sources.

Calculation

Calculation boxes give short Python programs. Each runs unchanged in a standard Python 3 environment, uses only the standard library and fixes its random seed; its actual output is printed below it. All returns, prices, reserves and economies in the book are fictional or synthetic, and parameters are illustrative, not estimates for any real market or country.

Hypothesis · not established

Hypothesis boxes contain the author’s ideas, marked “not established”, together with the result that would show them wrong. The author’s own models and definitions are labelled “the author’s model” in the text.

Source text

Source boxes quote ancient texts with a plain-language rendering.

Note

Note boxes give reminders and explain choices.

About this edition

This is version 1.0.0, the first complete edition: a preface and eight chapters, in Chinese and English. It contains 16 Python programs and 5 vector figures. The text, boxes and code were drafted with an AI writing tool from the author’s title, framework and notes; the author is responsible for reviewing and revising them and for checking sources. AI-drafted text can get bibliographic details wrong; corrections are welcome through the 5D Library website. The barbell comparison model of Chapter 5, the reserve-buffer model of Chapter 6 and the granary model of Chapter 7 are the author’s models; their mathematical properties are not evidence of predictions about any real market or country.

Disclosure

The author uses the works of 5D Library, including this book, as the methodological basis of his community courses, the 5D Ascension Cards (Dimensional Shift Card Decks) and the Yafu AI knowledge base, and therefore has a commercial interest in the subject of this book. The book promotes no course, card deck or service; the Chapter 8 trial is designed to be run and analysed by researchers independent of the author’s business and does not recruit from the author’s community, students or customers; all the code and the study protocol are published openly for anyone to check. The author develops and maintains quantitative research scripts and provides related open-source research and independent consulting services; while writing this book he may have held routine observation positions in US equity index funds (such as SPY) or related derivatives. All models in this book are offered solely for mathematical and historical-statistical discussion; they constitute no investment advice or trading signal, and the book recommends no specific trade or asset.

References

Ang, Andrew, and Geert Bekaert. 2002. “International Asset Allocation with Regime Shifts.” The Review of Financial Studies 15 (4): 1137–87. https://doi.org/10.1093/rfs/15.4.1137.
Black, Fischer, and André F. Perold. 1992. “Theory of Constant Proportion Portfolio Insurance.” Journal of Economic Dynamics and Control 16 (3–4): 403–26. https://doi.org/10.1016/0165-1889(92)90043-E.
Corsetti, Giancarlo, Marcello Pericoli, and Massimo Sbracia. 2005. “‘Some Contagion, Some Interdependence’: More Pitfalls in Tests of Financial Contagion.” Journal of International Money and Finance 24 (8): 1177–99.
Forbes, Kristin J., and Roberto Rigobon. 2002. “No Contagion, Only Interdependence: Measuring Stock Market Comovements.” The Journal of Finance 57 (5): 2223–61. https://doi.org/10.1111/0022-1082.00494.
Greenspan, Alan. 1999. Currency Reserves and Debt.
Hamilton, James D. 1989. “A New Approach to the Economic Analysis of Nonstationary Time Series and the Business Cycle.” Econometrica 57 (2): 357–84. https://doi.org/10.2307/1912559.
International Monetary Fund. 2011. Assessing Reserve Adequacy. IMF Policy Paper. International Monetary Fund.
International Monetary Fund. 2016. Guidance Note on the Assessment of Reserve Adequacy and Related Considerations. IMF Policy Paper. International Monetary Fund.
Longin, François, and Bruno Solnik. 2001. “Extreme Correlation of International Equity Markets.” The Journal of Finance 56 (2): 649–76. https://doi.org/10.1111/0022-1082.00340.
MacKenzie, Donald, and Taylor Spears. 2014. “‘The Formula That Killed Wall Street’: The Gaussian Copula and Modelling Practices in Investment Banking.” Social Studies of Science 44 (3): 393–417. https://doi.org/10.1177/0306312713517157.
Salmon, Felix. 2009. “Recipe for Disaster: The Formula That Killed Wall Street.” Wired 17 (3).
Taleb, Nassim Nicholas. 2012. Antifragile: Things That Gain from Disorder. Random House.

Fan, Y. (2026). Macro-Liquidity Regime Switching, Extreme Tail Dependence and Barbell Asset Allocation: Vine Copulas, Hidden Markov Models and Sovereign Buffer Dynamics [宏观流动性体制转换、极值尾部相依与杠铃资产配置] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.23828049 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).