5D Library › Regime-Switching Fractional Kelly, Maximum-Drawdown Constraints and Lifecycle Leverage Allocation · 中文

Chapter 7. “When Things Flourish They Decline”, Feng Xuan’s Three Burrows, and Vince Reconsidered

Why read these sources here

The earlier chapters had three themes: regimes that turn, the danger of betting everything on one estimate, and the size of a bet relative to what can be lost. Two passages of early Chinese history speak to the first two. This chapter checks the texts and asks which correspondences can be computed.

The Shiji on flourishing and decline

Source text

Sima Qian, Shiji, “Treatise on the Balanced Standard” (Ping zhun shu)

是以物盛則衰,時極而轉,一質一文,終始之變也。

(司马迁 1959)

“Thus when things flourish they decline, when a time reaches its extreme it turns; now plainness, now refinement: such are the changes of end and beginning.” Sima Qian closes his treatise on the economy of the Han with this sentence. In the language of Chapter 3 it says more than that regimes switch: it says that the longer a good time lasts, the nearer its turn, which is a rising hazard. The Markov model of this book has no such memory. Part (b) of p13 compares it with a run length that does.

Calculation

# Chapter 7. p13: Two classical ideas as arithmetic.
# (a) Feng Xuan's "three burrows": each refuge fails in a given year
# with chance 0.10. If failures are independent, all three fail with
# chance 0.10^3. If one common shock (the king's disfavour, chance s a
# year) takes all three at once, and otherwise they fail independently
# with the rest of the 0.10, the chance that all fail is computed below.
# (b) "When things flourish they decline; when a time reaches its
# extreme it turns" (Shiji, Ping zhun shu) as duration dependence. In
# the Markov model of p02 a bull run ends in the next 12 months with
# the same chance whatever its age. A Weibull run length with shape 2
# and the same mean (50 months) gives a rising hazard instead.
from math import exp, gamma

if __name__ == "__main__":
    p = 0.10
    print("   (a) common shock   one burrow  three burrows  ratio")
    for s in (0.0, 0.01, 0.03, 0.05):
        q = (p - s) / (1 - s)
        three = s + (1 - s) * q ** 3
        print("   {:16.2f}   {:10.3f}  {:13.4f}  {:5.1f}".format(
            s, p, three, p / three))
    lam = 50 / gamma(1.5)
    surv = lambda t: exp(-(t / lam) ** 2)
    print("   (b) age of bull run  Markov: ends in 12 mo  Weibull k=2")
    for age in (6, 24, 50, 80, 120):
        mk = 1 - 0.98 ** 12
        wb = 1 - surv(age + 12) / surv(age)
        print("   {:19d}  {:20.3f}  {:11.3f}".format(age, mk, wb))
#    (a) common shock   one burrow  three burrows  ratio
#                0.00        0.100         0.0010  100.0
#                0.01        0.100         0.0107    9.3
#                0.03        0.100         0.0304    3.3
#                0.05        0.100         0.0501    2.0
#    (b) age of bull run  Markov: ends in 12 mo  Weibull k=2
#                      6                 0.215        0.087
#                     24                 0.215        0.202
#                     50                 0.215        0.344
#                     80                 0.215        0.477
#                    120                 0.215        0.613

In the Markov model a bull run ends within the next twelve months with chance 0.215 whether it is six months old or ten years old. With a Weibull run length of the same mean and shape 2, the chance rises from 0.087 at six months to 0.344 at 50 months and 0.613 at 120 months: “when a time reaches its extreme it turns”. Whether market regimes have this property is an empirical question; for business-cycle expansions in the United States the evidence of duration dependence has been mixed (Diebold and Rudebusch 1990). A bettor who believes Sima Qian should cut leverage as a bull run ages; a bettor who believes the Markov model should not. Part (b) shows how much the two beliefs differ: by a factor of almost three for a run of 120 months.

Feng Xuan’s three burrows

Source text

Zhanguo ce, “Qi ce” 4

狡兔有三窟,僅得免其死耳;今君有一窟,未得高枕而臥也。請為君復鑿二窟。

(刘向 1985)

“A cunning hare has three burrows, and even so only just escapes death; now you have one burrow, and cannot yet sleep on a high pillow. Let me dig two more for you.” Feng Xuan, retainer of Lord Mengchang, built three refuges for his master: the goodwill of the people of Xue, the favour of the king of Wei, and an ancestral temple of the Qi kings at Xue. Part (a) of p13 shows when three burrows are worth more than one. If each fails in a year with chance 0.10 and failures are independent, all three fail with chance 0.0010, a hundredth of the risk of one. If a common shock with chance 0.05 a year takes all three at once, all three fail with chance 0.0501, only half the risk of one. Diversification protects against independent failures, not against a shock that hits every refuge at once; in markets that shock is a bear regime, in which correlations rise.

Vince reconsidered

Chapter 5 checked Ralph Vince’s optimal f. Two clarifications follow from it. First, optimal f is not a different principle from Kelly but Kelly applied to a list of trades, with the bet scaled by the largest loss. Second, the largest loss is the most uncertain number in the list: a record of 30 trades missed it about one time in five (p10), and a trader who sizes by a largest loss that has not yet happened is overbetting without knowing it.

Hypothesis · not established

The author’s interpretation (not yet tested): across national equity markets, the chance that a bull market ends within a year rises with its age. What would show it wrong: in a long multi-country sample of bull and bear market dates, a test of duration dependence finds a flat or falling hazard of a bull market ending as it ages.

Summary

“When things flourish they decline” is a claim about a rising hazard, which the simplest regime models leave out; whether markets have it is still to be shown. Three burrows protect only against independent failures. Optimal f is Kelly in trader’s units, and its weakest input is the size of the worst trade.

References

Diebold, Francis X., and Glenn D. Rudebusch. 1990. “A Nonparametric Investigation of Duration Dependence in the American Business Cycle.” Journal of Political Economy 98 (3): 596 to 616.
刘向. 1985. 战国策. 上海古籍出版社.
司马迁. 1959. 史记. 中华书局.

Fan, Y. (2026). Regime-Switching Fractional Kelly, Maximum-Drawdown Constraints and Lifecycle Leverage Allocation: Cyclical Drift Estimation Errors, Ruin-Bounded Bet Sizing and Human-Capital Amortization [体制转换分数凯利、最大回撤约束与生命周期杠杆配置] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.24090256 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).