关于「过度自信遇上凯利公式:高估胜率的增长代价与破产概率」,《效用、真实与双账本》给出的定量结论是什么?
结论:第五章把过度自信接上凯利公式:真实胜率 0.55 的等额押注,高估胜率 0.05 就使长期增长率跌到 -0.000138。
数据口径:书中参数与数据均为示意或合成,全部计算由书中程序生成;本书不是医疗建议,不提供任何临床心理服务,不是个人投资建议,不承诺任何收益;功能性信念只用于自己的行动框架,不用于欺骗他人。 第六章建立双账本:行动框架可按效用编码,概率账以 Brier 分数校准,硬边界账不让步;框架改写概率账时,示意破产比例从 0.0925 升到 0.7055。
对照数据表
下表数字由本章 Python 程序原样打印,可重新运行复现。 程序只用 Python 标准库与结论吻合的打印值: -0.000138, 0.55
| true p 0.55, Kelly stake 0.100, best growth 0.005008 per round | ||||
| growth reaches zero at stake 0.1987 (1.99 x Kelly) | ||||
| dp | stake f | growth g(f) | g / best | half-stake growth |
| 0.0000 | 0.1000 | 0.005008 | 1.000 | 0.003753 |
| 0.0125 | 0.1250 | 0.004692 | 0.937 | 0.004301 |
| 0.0250 | 0.1500 | 0.003736 | 0.746 | 0.004694 |
| 0.0375 | 0.1750 | 0.002130 | 0.425 | 0.004930 |
| 0.0500 | 0.2000 | -0.000138 | -0.028 | 0.005008 |
| 0.0625 | 0.2250 | -0.003084 | -0.616 | 0.004929 |
| 0.0750 | 0.2500 | -0.006728 | -1.343 | 0.004692 |
| 0.0875 | 0.2750 | -0.011092 | -2.215 | 0.004294 |
| 0.1000 | 0.3000 | -0.016203 | -3.235 | 0.003736 |
查看完整验算程序
# Chapter 5. p51: growth rate when the win probability is overestimated
# A repeated even-money stake (Kelly 1956): win probability p = 0.55, a
# win returns the stake, a loss forfeits it. A bettor who believes the
# probability is p + dp stakes the Kelly fraction of that belief,
# f = 2 (p + dp) - 1. The true long-run growth of log wealth per round
# is g(f) = p ln(1 + f) + (1 - p) ln(1 - f). Also shown: the stake at
# which growth crosses zero (found by bisection) and half-Kelly on the
# inflated belief. Wealth units are abstract; illustrative only.
import math
P = 0.55
def growth(f):
return P * math.log(1 + f) + (1 - P) * math.log(1 - f)
f_star = 2 * P - 1
lo, hi = f_star, 0.999
for _ in range(80):
mid = (lo + hi) / 2
if growth(mid) > 0:
lo = mid
else:
hi = mid
print("true p %.2f, Kelly stake %.3f, best growth %.6f per round" % (
P, f_star, growth(f_star)))
print("growth reaches zero at stake %.4f (%.2f x Kelly)" % (
lo, lo / f_star))
print(" dp stake f growth g(f) g / best half-stake growth")
for k in range(0, 9):
dp = 0.0125 * k
f = min(0.999, 2 * (P + dp) - 1)
print("%5.4f %7.4f %12.6f %9.3f %17.6f" % (
dp, f, growth(f), growth(f) / growth(f_star), growth(f / 2)))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《效用优先于字面真实:演化适存度博弈、功能性信念重塑与物理硬边界母典》第 5 章「过度自信遇上凯利公式:高估胜率的增长代价与破产概率」(版本 1.0.0). 5D Library. DOI 待发布
DOI 待发布 · 专著页正式发布后上线(本站暂无专著页) · 5DL-B-427 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026427,
author = {Fan, Yafu},
title = {When Utility Outperforms Literal Truth: Evolutionary Fitness-Payoff Games, Functional Belief Engineering and Physical Guardrails},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}