What does Likelihood-Ratio Screening of Luxury Goods conclude, in numbers, about “A Loss Matrix and Three Zones: Instant Buy, Referral and Decline”?
Answer: Chapter 4 derives three zones from a loss matrix: an instant buy below a posterior of 0.0539, a referral from 0.0539 to 0.8362, and a decline above; by Good's theorem the gross value of free information is never negative, a real test can change the decision only between 0.0017 and 0.9419, and after fees and seller walk-away the net value of a referral is positive only in the middle band.
Scope of the numbers: Brands are named only generically, the parameters are illustrative, all calculations are generated by the book's programs, and the book is not individual business advice. Chapter 3 gives the effective number of independent features (four features are worth only 2.007 under the illustrative matrix) and a one-number slope of 0.571.
Data table
The numbers below are printed verbatim by the chapter's Python program and can be reproduced by running it. standard-library Python onlyprinted values that match the answer: 0.0017, 0.0539, 0.8362, 0.9419
| q | no-info | VOI perfect | VOI real lab | net refer | zone |
|---|---|---|---|---|---|
| 0.0010 | 3574.4 | 22.0 | 0.0 | -1204.2 | buy |
| 0.0100 | 3344.0 | 220.0 | 182.2 | -999.4 | buy |
| 0.0300 | 2832.0 | 660.0 | 618.5 | -544.1 | buy |
| 0.0539 | 2220.2 | 1185.8 | 1139.9 | -0.1 | buy |
| 0.1000 | 1040.0 | 2200.0 | 2145.6 | 1049.2 | refer |
| 0.1406 | 0.6 | 3093.2 | 3031.3 | 1973.3 | refer |
| 0.3000 | 0.0 | 2520.0 | 2428.8 | 1521.6 | refer |
| 0.5000 | 0.0 | 1800.0 | 1672.0 | 954.0 | refer |
| 0.8362 | 0.0 | 589.7 | 399.8 | -0.1 | decline |
| 0.9000 | 0.0 | 360.0 | 158.4 | -181.2 | decline |
| 0.9700 | 0.0 | 108.0 | 0.0 | -379.9 | decline |
| 0.9950 | 0.0 | 18.0 | 0.0 | -450.8 | decline |
| real lab can change the action only for 0.0017 < q < 0.9419 | |||||
| outside that range its gross value is exactly 0.0 | |||||
| peak gross value 3031.9 at q* = 0.1406 |
Show the full program
# Chapter 4. p42: value of information along the posterior q = P(N).
# Gross value of a free, instant lab result (no fee, no walk-away) is
# never negative (Good 1967), and it is exactly zero wherever no lab
# result could change the action. Net value of referral = best of refer
# minus best of buy and decline, with fee 300 and walk-away 0.25.
# Loss matrix as p41 (price 20000, margin 3600, loss 22000), illustrative.
import math
M, BIG = 3600.0, 22000.0
FEE, WALK = 300.0, 0.25
A, B = 0.99, 0.01 # 鉴定通过率:G 物品与 N 物品(示意)
def act(q):
return max(M - q * (M + BIG), 0.0)
def gross_voi(q, a, b):
# 免费且即时的鉴定:看结果后再在买与不买之间选
out = []
for pg, pn in ((a, b), (1.0 - a, 1.0 - b)):
pr = (1.0 - q) * pg + q * pn
if pr <= 0.0:
continue
qq = q * pn / pr
out.append(pr * act(qq))
return math.fsum(out) - act(q)
def refer(q):
return -FEE + (1.0 - WALK) * (A * (1.0 - q) * M - B * q * BIG)
def flip_range(a, b):
# 鉴定结果能改变决定的 q 区间:通过后买、未通过后不买
qs = M / (M + BIG)
lo_odds = qs / (1 - qs) * (1 - a) / (1 - b)
hi_odds = qs / (1 - qs) * a / b
return lo_odds / (1 + lo_odds), hi_odds / (1 + hi_odds)
print(" q no-info VOI perfect VOI real lab net refer zone")
for q in [0.001, 0.01, 0.03, 0.0539, 0.10, 0.1406, 0.30, 0.50,
0.8362, 0.90, 0.97, 0.995]:
base = act(q)
vp = gross_voi(q, 1.0, 0.0)
vr = gross_voi(q, A, B)
net = refer(q) - base
zone = "refer" if net > 0 else ("buy" if base > 0 else "decline")
print("%6.4f %9.1f %12.1f %13.1f %10.1f %s"
% (q, base, vp, vr, net, zone))
lo, hi = flip_range(A, B)
print("real lab can change the action only for %.4f < q < %.4f"
% (lo, hi))
print("outside that range its gross value is exactly 0.0")
print("peak gross value %.1f at q* = %.4f"
% (gross_voi(M / (M + BIG), A, B), M / (M + BIG)))Single source and citation
Fan, Y. (2026). Multi-Feature Likelihood-Ratio Screening of Pre-Owned Luxury Goods: Fusing Stitch Pitch, Hardware Spectra and Leather Grain, and the Bayesian Boundary between Lab Referral and Instant Purchase (Version 1.0.0), Chapter 4: A Loss Matrix and Three Zones: Instant Buy, Referral and Decline. 5D Library. DOI pending
DOI pending · monograph page goes live after publication · 5DL-B-281 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026281,
author = {Fan, Yafu},
title = {Multi-Feature Likelihood-Ratio Screening of Pre-Owned Luxury Goods: Fusing Stitch Pitch, Hardware Spectra and Leather Grain, and the Bayesian Boundary between Lab Referral and Instant Purchase},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}Other chapter answers in this book
- Chapter 1 · Weight of Evidence: Single-Feature Likelihood Ratios, Decibans and the Sum of Four Features
- Chapter 2 · Double Counting: How Correlated Features Make Naive Bayes Overconfident
- Chapter 3 · Correcting for Correlation: the Effective Number of Independent Features, Exact Weights and a One-Number Slope
- Chapter 5 · Measuring More or Stopping: Cheap Evidence Is Worth Most at the Boundaries
- Chapter 6 · Channel Priors, Audits and Calibration: What Gets Referred Decides What a Store Can Learn
- Chapter 7 · East and West in the Sources: Suspecting the Neighbour of the Axe, and Peirce's Thermometer of Belief
- Chapter 8 · Closing the Loop: a Three-Zone Intake Procedure, the Refer-or-Buy Screener and a Pre-Registered Test