What does Liuyao Multiple Motion, Hidden Spirits and Timing conclude, in numbers, about “Practice: One Cast with Three Moving Lines, the Ten Entries It Switches On, Its Timing, and a Pre-registered Forecasting Test”?

Answer: Chapter 8 demonstrates the look-up and the timing on a business cast with three moving lines that switches on ten entries, and gives a pre-registered forecasting test: with one day of tolerance each, the timing rules that apply to that cast cover 0.9286 of the following two weeks; comparing timing distributions by the logarithmic score needs 39 casts if the reading is right every time, and 203387 if only half the casts carry real information. Marked boxes separate published evidence, runnable Python calculations, classical sources and the author's hypotheses.

Scope of the numbers: All calculations are generated by the programs printed in the book; it is not investment, legal, psychological or marital advice. Marked boxes separate published evidence, runnable Python calculations, classical sources and the author's hypotheses.

中文版

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: 203387, 39

(0) calendar chance of a timing call, long horizon
branches named t0 t = 1 t = 2
10.0833 0.2500 0.4167
20.1667 0.4545 0.6818
30.2500 0.6182 0.8409
40.3333 0.7455 0.9293
(1) hits against calendar chance, exact binomial
one branch, exact day chance 0.0833 true 0.15: N141
one branch, exact day chance 0.0833 true 0.20: N56
one branch, +-1 day chance 0.2500 true 0.35: N129
five named days of 14 chance 0.3571 true 0.50: N77
(2) log-score gain per call, reading vs rival
f1.00: mean +0.3117 sd 0.7736 N = 39
f0.50: mean +0.0045 sd 0.8167 N = 203387
f0.25: mean -0.1491 sd 0.7940 N = never
Show the full program
# Chapter 8. p09: Calendar chance, and how many pre-registered calls a test needs.
# (0) A branch returns every 12 days, so a rule that names a branch names
# every twelfth day; with a tolerance of t days either side it covers
# (2t + 1)/12 of the calendar, and several named branches cover their
# union (averaged here over all sets of m distinct branches). (1) Hit
# counting against that chance rate c: smallest N for which a one-sided
# exact binomial test at alpha 0.05 reaches power 0.80 when the true hit
# rate is p1 (scan upward from N = 10). (2) Log-score gain of the timing
# distribution of p08 over its rival. With mean mu and standard deviation
# sd per call, a one-sided z-test needs N = ((1.6449 + 0.8416) sd / mu)^2;
# the truth is the reading's distribution for a fraction f of calls and
# the rival's for the rest, f = 1, 0.5, 0.25.
from fractions import Fraction as F
from itertools import combinations
from math import ceil, comb, log, sqrt

def cover(branches, t):
    return F(len({(b + s) % 12 for b in branches for s in range(-t, t + 1)}), 12)

def tail(n, k, p):
    return sum(comb(n, x) * p ** x * (1 - p) ** (n - x) for x in range(k, n + 1))

def n_needed(c, p1):
    n = 10
    while True:
        k = 0
        while tail(n, k, c) > 0.05:
            k += 1
        if tail(n, k, p1) >= 0.80:
            return n
        n += 1

if __name__ == "__main__":
    print("   (0) calendar chance of a timing call, long horizon")
    print("       branches named   t = 0    t = 1    t = 2")
    for m in (1, 2, 3, 4):
        sets = list(combinations(range(12), m))
        print("       {}                {}".format(m, "   ".join("{:.4f}".format(
            float(sum(cover(s, t) for s in sets) / len(sets))) for t in (0, 1, 2))))
    print("   (1) hits against calendar chance, exact binomial")
    for name, c, p1 in (("one branch, exact day", F(1, 12), 0.15),
                        ("one branch, exact day", F(1, 12), 0.20),
                        ("one branch, +-1 day", F(1, 4), 0.35),
                        ("five named days of 14", F(5, 14), 0.50)):
        print("       {:22s} chance {:.4f} true {:.2f}: N = {}".format(
            name, float(c), p1, n_needed(float(c), p1)))
    r = {d: F(1, 5) / 9 for d in range(1, 15)}
    r.update({1: F(1, 5), 4: F(3, 20), 6: F(1, 10), 10: F(1, 10),
              13: F(1, 20), "none": F(1, 5)})
    b = {d: F(4, 5) / 14 for d in range(1, 15)}
    b["none"] = F(1, 5)
    g = {o: log(r[o]) - log(b[o]) for o in r}
    print("   (2) log-score gain per call, reading vs rival")
    for f in (1.0, 0.5, 0.25):
        truth = {o: f * float(r[o]) + (1 - f) * float(b[o]) for o in r}
        mu = sum(truth[o] * g[o] for o in r)
        sd = sqrt(sum(truth[o] * (g[o] - mu) ** 2 for o in r))
        n = ceil(((1.6449 + 0.8416) * sd / mu) ** 2) if mu > 0 else "never"
        print("       f = {:.2f}: mean {:+.4f} sd {:.4f}  N = {}".format(
            f, mu, sd, n))

Single source and citation

Fan, Y. (2026). The Liuyao Codex of Multiple Moving Lines, Hidden Motion, Hidden Spirits and Timing (Version 1.0.0), Chapter 8: Practice: One Cast with Three Moving Lines, the Ten Entries It Switches On, Its Timing, and a Pre-registered Forecasting Test. 5D Library. DOI pending

DOI pending · monograph page goes live after publication · 5DL-B-125 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026125,
  author    = {Fan, Yafu},
  title     = {The Liuyao Codex of Multiple Moving Lines, Hidden Motion, Hidden Spirits and Timing},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

Other chapter answers in this book