关于「尾数格点:把档位价落到当地客人认得的数字上」,《购买力平价与多币种尾数定价》给出的定量结论是什么?
结论:格点是一个周期加一组尾数,它离档位价最远有多远,大致由周期除以价格决定:同样的 x9,在令吉一百多的价位上就近误差约 6.0%,在新元四十多的价位上是 14.7%。八类候选格点各有用处:小数尾几乎不动价格,个位 9 与双 9 给出划算信号,整数价适合礼品与收藏,门槛下方格点是唯一改变首位读数的一类,吉利 8 与回避 4 是约束而不是证据,日元的 80 与 98 在百元周期上几乎不偏离。取整先看底线,再看打平所需的销量:令吉往下取一格要多卖 16.0%,新元要多卖 40.7%。左位数效应随首位数字递减,在 20、30、50 这样的门槛上最值钱。回避 4 的代价集中在日元四千段,偏离会到 10.5%。流水线把这些规则串起来,给出带理由的价格,牌卡在五个市场折成美元从 22.60 到 43.99,这个价差留给第五章去检查。
数据口径:书中的汇率与价格水平都是写明的样本假设,不是实时数据;税费、标价与价格歧视的法规只写一般框架,不是法律或税务意见。 第八章让一个虚构牌卡品牌从 49 美元出发走完四步,最终价格是美国 49.00 美元、马来西亚 RM 149.00(被套利带从 RM 129.90 提上来)、新加坡 S$ 59.90、台湾 NT$ 1,080、日本 ¥5,580;典藏礼盒五地美元等值价的最大最小之比只有 1.1344,日常牌卡是 1.4536;模拟的汇率路径让日本的毛利率跌到 0.5433,最先触发复核;在虚构流量下,单个市场要分辨一成的毛利差距需要 97 到 257 周,所以测价只测大的决定,五地合并分析。
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: 100, 14.7, 30, 6.0, 80, 98
| SAMPLE bands (assumed). S = period, |E| = endings per period, n = points in band | ||||||||
| ccy | band | class | S |E| | n | near | up | | no4 near | |
| USD | 30-60 | x.99 | 1 | 1 | 30 | 1.6% | 3.3% | | 12.1% |
| USD | 30-60 | x9 | 10 | 1 | 3 | 14.7% | 34.5% | | 20.4% |
| USD | 30-60 | x9.99 | 10 | 1 | 3 | 14.3% | 33.3% | | 20.0% |
| USD | 30-60 | x0 | 10 | 1 | 4 | 14.3% | 33.3% | | 25.0% |
| MYR | 80-160 | x.90 | 1 | 1 | 80 | 0.6% | 1.3% | | 3.8% |
| MYR | 80-160 | x9 | 10 | 1 | 8 | 6.0% | 12.7% | | 6.7% |
| MYR | 80-160 | x9.90 | 10 | 1 | 8 | 5.9% | 12.5% | | 6.7% |
| MYR | 80-160 | x8 | 10 | 1 | 8 | 6.0% | 12.8% | | 6.8% |
| MYR | 80-160 | x0 | 10 | 1 | 9 | 5.9% | 12.5% | | 7.1% |
| SGD | 35-70 | x.90 | 1 | 1 | 35 | 1.4% | 2.9% | | 12.1% |
| SGD | 35-70 | x9 | 10 | 1 | 4 | 14.7% | 34.5% | | 20.4% |
| SGD | 35-70 | x9.90 | 10 | 1 | 4 | 14.3% | 33.4% | | 20.0% |
| SGD | 35-70 | x0 | 10 | 1 | 4 | 14.3% | 33.3% | | 25.0% |
| TWD | 700-1,400 | x9 | 10 | 1 | 70 | 0.7% | 1.4% | | 3.8% |
| TWD | 700-1,400 | x99 | 100 | 1 | 7 | 6.7% | 14.3% | | 6.7% |
| TWD | 700-1,400 | x88 | 100 | 1 | 7 | 6.8% | 14.5% | | 6.8% |
| TWD | 700-1,400 | x80/88/90/99 | 100 | 4 | 28 | 5.5% | 11.6% | | 6.1% |
| TWD | 700-1,400 | x00 | 100 | 1 | 8 | 6.7% | 14.3% | | 7.1% |
| JPY | 3,500-7,000 | x80 | 100 | 1 | 35 | 1.4% | 2.9% | | 12.1% |
| JPY | 3,500-7,000 | x98 | 100 | 1 | 35 | 1.4% | 2.9% | | 12.1% |
| JPY | 3,500-7,000 | x980 | 1,000 | 1 | 4 | 14.4% | 33.6% | | 20.1% |
| JPY | 3,500-7,000 | x000 | 1,000 | 1 | 4 | 14.3% | 33.3% | | 25.0% |
| JPY | 3,500-7,000 | x500 | 500 | 1 | 8 | 6.7% | 14.3% | | 17.6% |
| rule of thumb: for one ending per period S at price p, near ~ S / (2p) and up ~ S / p | ||||||||
| MYR x9 at 103.89: S/(2p) = 4.8%; JPY x80 at 4,445.89: S/(2p) = 1.1% |
查看完整验算程序
# Chapter 4. p40: Candidate ending lattices per currency: period, endings, density and worst snap error.
# A lattice class is a period S and a set of endings E: points k * S + e for integers k >= 0, e in E.
# near = worst error of snapping to the NEAREST point, max (b - a) / (a + b) over adjacent a < b;
# up = worst error of always snapping UP, max (b - a) / a. Both are fractions of the price.
# no4 = the same class after dropping every point whose integer part contains the digit 4.
# Bands are SAMPLE price bands around the chapter 1 tier prices (assumed, not market data).
from fractions import Fraction as F
CLASSES = {
"USD": [("x.99", 1, ["0.99"]), ("x9", 10, ["9"]), ("x9.99", 10, ["9.99"]), ("x0", 10, ["0"])],
"MYR": [("x.90", 1, ["0.90"]), ("x9", 10, ["9"]), ("x9.90", 10, ["9.90"]), ("x8", 10, ["8"]),
("x0", 10, ["0"])],
"SGD": [("x.90", 1, ["0.90"]), ("x9", 10, ["9"]), ("x9.90", 10, ["9.90"]), ("x0", 10, ["0"])],
"TWD": [("x9", 10, ["9"]), ("x99", 100, ["99"]), ("x88", 100, ["88"]),
("x80/88/90/99", 100, ["80", "88", "90", "99"]), ("x00", 100, ["0"])],
"JPY": [("x80", 100, ["80"]), ("x98", 100, ["98"]), ("x980", 1000, ["980"]), ("x000", 1000, ["0"]),
("x500", 500, ["0"])]}
BANDS = {"USD": (30, 60), "MYR": (80, 160), "SGD": (35, 70), "TWD": (700, 1400), "JPY": (3500, 7000)}
def points(period, ends, hi, no4=False):
out = []
for k in range(0, int(hi) // period + 2):
for e in ends:
p = k * period + F(e)
if p > 0 and not (no4 and "4" in str(int(p))):
out.append(p)
return sorted(out)
def worst(period, ends, lo, hi, no4=False):
pts = points(period, ends, 3 * hi, no4)
pairs = [(a, b) for a, b in zip(pts, pts[1:]) if b > lo and a < hi]
near = max(F(b - a) / (a + b) for a, b in pairs)
up = max(F(b - a) / a for a, b in pairs)
return near, up, sum(1 for p in pts if lo <= p <= hi)
if __name__ == "__main__":
print("SAMPLE bands (assumed). S = period, |E| = endings per period, n = points in band")
print("{:<4} {:>11} {:<13} {:>5} {:>3} {:>4} {:>7} {:>7} | {:>9}".format(
"ccy", "band", "class", "S", "|E|", "n", "near", "up", "no4 near"))
for ccy, rows in CLASSES.items():
lo, hi = BANDS[ccy]
for name, period, ends in rows:
near, up, n = worst(period, ends, lo, hi)
near4 = worst(period, ends, lo, hi, True)[0]
print("{:<4} {:>5,}-{:<5,} {:<13} {:>5,} {:>3} {:>4} {:>7.1%} {:>7.1%} | {:>9.1%}".format(
ccy, lo, hi, name, period, len(ends), n, float(near), float(up), float(near4)))
print("rule of thumb: for one ending per period S at price p, near ~ S / (2p) and up ~ S / p")
print(" MYR x9 at 103.89: S/(2p) = {:.1%}; JPY x80 at 4,445.89: S/(2p) = {:.1%}".format(
float(F(10) / (2 * F("103.89"))), float(F(100) / (2 * F("4445.89")))))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《跨国购买力平价畸变、多币种心理尾数映射与东南亚台日定价矩阵母典》第 4 章「尾数格点:把档位价落到当地客人认得的数字上」(版本 0.1.0). 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-134 · v0.1.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026134,
author = {Fan, Yafu},
title = {Purchasing-Power-Parity Distortions, Multi-Currency Psychological Price Endings and the Southeast Asia–Taiwan–Japan Pricing Matrix},
year = {2026},
publisher = {5D Library},
edition = {0.1.0},
note = {DOI pending}
}