5D Library › Prediction Market Microstructure, Bregman Information Projection and Cross-Book Arbitrage Bounds · 中文
Chapter 8. A Four-Step Check and a Pre-Registered Test of the Lock-Up Discount
Four questions before reading a prediction-market price
This chapter turns the book into four questions that can be put to any prediction-market price — including one you are thinking of trading. They are not investment advice and not a trading strategy.
1. Whose statistic is this price, and which one? A price is some quantile or weighted mean of the belief distribution, depending on traders’ budgets and attitudes to risk (Chapter 1). Longshots usually trade above their probability and favourites below. Ask who is trading and how much they disagree.
2. How long until resolution, and what does capital cost? A contract that pays no interest before resolution has a no-arbitrage band that widens with the horizon and the cost of capital (Chapter 3, p06). A “sure thing” a year from resolution sitting at 0.92 does not mean the market sees an 8 % chance of failure.
3. How much of the displayed arbitrage can be executed? Prices that do not sum to 1, or related contracts that disagree, are only ceilings (Chapters 4 and 5). Count the cost of walking the book, fees, fixed costs, leg risk and the opportunity cost of the lock-up; and read the settlement rules to be sure a “set” really pays 1.
4. How many independent events stand behind the evidence of bias? A thousand contracts may come from twenty events (Chapter 6). Cluster standard errors by event; before correcting a bias, make sure there are enough independent resolved observations, or the correction may do more harm than the bias.
A pre-registered test of the lock-up discount
Chapter 3’s model makes a prediction that data could refute: on platforms where capital has a cost, favourites far from resolution are systematically underpriced. The test below is written before any data are looked at.
Hypothesis · not established
Protocol. Sample: resolved binary contracts from any public exchange archive, priced exactly 90 days before resolution, with a price in [0.85, 0.97]; at most 5 per event. Primary estimand : the mean of outcome (0 or 1) minus price; positive means favourites are underpriced. Smallest effect of interest (SESOI): 0.01. Standard errors clustered by event. The decision rule is in the program; three gates make a result uninterpretable rather than negative: fewer than 90 % of the planned events found; more than 2 % of contracts with disputed settlement; the snapshot-timing rule changed after outcomes were seen. The test recommends no trade and is to be run and analysed by people independent of any trading decision.
Calculation
# Chapter 8. p16: A pre-registered test of the lock-up discount.
# Question: do long-dated favourites resolve YES more often than their
# price says? Sample: resolved binary contracts priced in [0.85, 0.97]
# exactly 90 days before resolution, five per event (the cap), from any
# public exchange archive. Estimand D = mean(outcome - price). The model
# of Chapter 3 predicts about +0.035; the smallest effect of interest
# (SESOI) is 0.01. Outcomes within an event share a shock (Gaussian
# copula, rho = 0.3), so standard errors are clustered by event.
# Part A: power at true D = +0.02 (half the model's prediction) and the
# false-positive rate at D = 0. Part B: the decision rule applied to one
# simulated archive.
import math
import random
from statistics import NormalDist
M, RHO, SESOI = 5, 0.3, 0.01
PHI = NormalDist()
def archive(rng, events, d):
rows = []
for e in range(events):
z = rng.gauss(0, 1)
for _ in range(M):
price = rng.uniform(0.85, 0.97)
p = min(price + d, 0.999)
u = PHI.cdf(math.sqrt(RHO) * z + math.sqrt(1 - RHO)
* rng.gauss(0, 1))
rows.append((e, price, 1 if u < p else 0))
return rows
def estimate(rows):
n = len(rows)
d = sum(y - p for _, p, y in rows) / n
by = {}
for e, p, y in rows:
by[e] = by.get(e, 0.0) + (y - p - d)
se = math.sqrt(sum(v * v for v in by.values())) / n
return d, se
def decide(d, se, planned, got, disputed):
lo, hi = d - 1.96 * se, d + 1.96 * se
if got < 0.9 * planned:
return "UNINTERPRETABLE: fewer than 90 % of planned events"
if disputed > 0.02:
return "UNINTERPRETABLE: over 2 % of contracts disputed"
if lo > 0 and d >= SESOI:
return "SUPPORTED: favourites underpriced by at least the SESOI"
if lo > -SESOI and hi < SESOI:
return "EQUIVALENT: no lock-up discount of practical size"
if hi < 0:
return "REVERSED: favourites overpriced"
return "INCONCLUSIVE: interval too wide to decide"
if __name__ == "__main__":
rng = random.Random(59016)
print("A. events power at D=+0.02 false positives at D=0")
for ev in (100, 200, 400, 600, 800):
pw = fp = 0
for _ in range(400):
d, se = estimate(archive(rng, ev, 0.02))
pw += (d - 1.96 * se > 0 and d >= SESOI)
d, se = estimate(archive(rng, ev, 0.0))
fp += (d - 1.96 * se > 0 and d >= SESOI)
print(" {:6d} {:5.3f} {:5.3f}".format(
ev, pw / 400, fp / 400))
d, se = estimate(archive(rng, 600, 0.02))
print("\nB. one archive of 600 events: D = {:+.4f}, 95 % CI"
" [{:+.4f}, {:+.4f}]".format(d, d - 1.96 * se, d + 1.96 * se))
print(" 600 of 600 events found:")
print(" " + decide(d, se, 600, 600, 0.0))
print(" 520 of 600 events found:")
print(" " + decide(d, se, 600, 520, 0.0))
print("\nReading: 600 events (3,000 contracts) give 0.95 power at")
print("half the effect the model predicts; every branch of")
print("the rule, including the ones that would embarrass the model,")
print("is written down before any archive is opened.")
# A. events power at D=+0.02 false positives at D=0
# 100 0.347 0.043
# 200 0.570 0.037
# 400 0.823 0.022
# 600 0.950 0.030
# 800 0.978 0.020
#
# B. one archive of 600 events: D = +0.0235, 95 % CI [+0.0123, +0.0347]
# 600 of 600 events found:
# SUPPORTED: favourites underpriced by at least the SESOI
# 520 of 600 events found:
# UNINTERPRETABLE: fewer than 90 % of planned events
#
# Reading: 600 events (3,000 contracts) give 0.95 power at
# half the effect the model predicts; every branch of
# the rule, including the ones that would embarrass the model,
# is written down before any archive is opened.Chapter 3’s model predicts of about +0.035. So that the test does not depend on the model being right, power is computed at half that effect (+0.02): only 0.347 with 100 events, 0.823 with 400, and 0.950 with 600 events (3,000 contracts); the false-positive rate at is below 5 % at every size. The planned sample is therefore 600 events.
Part B runs the decision rule on one simulated archive: , 95 % interval [+0.0123, +0.0347]; the lower end is above 0 and the estimate is at least the SESOI, so the verdict is “supported”. If the same archive had yielded only 520 events, under 90 % of plan, the verdict would be “uninterpretable” — whatever the estimate. The rule names every branch, including the two that would embarrass this book’s model: an interval inside ±0.01 (“equivalent”: no lock-up discount of practical size) and an interval entirely below zero (“reversed”: favourites overpriced).
Note
What a result would mean and what it would not. It describes favourites on the chosen platforms at the chosen horizon. A positive is consistent with a lock-up discount but does not exclude other explanations — for instance, that buyers of favourites happen to be more risk-averse; separating them needs the comparison in Chapter 3’s hypothesis: interest-bearing against non-interest-bearing contracts on one platform. Nor would a positive mean that buying such contracts pays: Chapter 3 showed their annual return need not beat the opportunity cost of capital.
What this book does not do
This book uses no data from any real exchange, does not assess the compliance or reliability of any platform, does not discuss how particular jurisdictions regulate prediction markets, and does not advise readers to take part in any prediction market or betting. All prices and order books in it are synthetic data generated by its programs.
Fan, Y. (2026). Prediction Market Microstructure, Bregman Information Projection and Cross-Book Arbitrage Bounds: Logarithmic Market Scoring Rules, Favorite-Longshot Calibration Bias, Lock-Up Opportunity Cost and Frank-Wolfe Combinatorial Pricing [预测市场微观结构、布雷格曼信息投影与跨盘口无套利边界] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.24090166 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).