The slow way, made survivable Chapter 12 of 23 Svenska

The map · Part One · The method · Chapter 12 of 23

The shadow book

Every refusal is measured against the market afterwards. A system that only measures what it bought cannot learn, and a process that only refuses, and never scores its refusals, cannot be shown to be wrong. About 7 minutes.

By now the shape of the method should be clear: it mostly says no. Of the names it has read closely, it has, at the time of writing, refused every one. That raises an uncomfortable question. How would anyone know if the method were simply too strict? A process that never buys never has a bad purchase to point to. It also never has a good one. Its record is spotless and says nothing.

The owner's ruling put it this way: "This framework has evaluated names and refused all of them, and nothing in it measures what a refusal cost or saved. A process that only refuses, and never scores its refusals, cannot be falsified."11Ruling E114, 2026-09-08.

What is recorded

The shadow book is a table with one row per refusal. Each row records six facts and nothing else: the company, the date of the verdict, what the verdict was and whether it stands, the settled closing price on that date in the company's own currency, the value if one had been struck, and one line naming what decided it.

A refusal tracked forwardA timeline from the verdict day to a year later. A box lists what is recorded on the day: the date, the closing price, one line naming what decided it. A dashed box lists what is not recorded: a thesis, a target, an expectation. An orange line with circles follows the refused name, a dashed black line follows an index. verdict daya year laterrecorded that day:the date, the closing price,one line naming what decided itnot recorded:a thesis, a target,an expectation, an argumentthe refused namean index, same currency, same kind of returnread once a year, on purpose, so a scoreboard cannot bend a rule
Every refusal is measured afterwards against the market, from the day it was written. What is recorded is deliberately thin: a date, a price, one reason. A row that argues is a row that will be re-argued.

What is deliberately not recorded matters as much. No thesis. No expectation of what the price will do. No target. In the ruling's words, "a row that argues is a row that will be re-argued." The book is a record of decisions and prices, not a record of opinions about them.

What is measured

From the verdict date forward, the refused name's return is set beside the return of an index over the same period, in the same currency, and, this is the part that took an amendment, the same kind of return. A share's return includes its dividends; an index's may or may not. The first version of the rule let a name whose dividends the vendor did not record be compared as a bare price against an index that included dividends, which would have made every refusal look a little wiser than it was. The rule was amended one day after it was written: where dividends cannot be established, the row is DATA MISSING rather than a flattering comparison. The owner's reason: "a shadow book exists to be able to say the method does not work, and one that systematically flatters the refusals cannot do that."22Ruling E114 as amended 2026-09-09. The book stores the unadjusted close, a fact that never changes, and derives the total return beside it at reading time, saying so.

Four limits, stated by the ruling itself

  1. It measures what happened, never whether the verdict was right. A refused name that rose may have risen for exactly the reasons the verdict declined to bet on. The book cannot tell the difference, and does not pretend to.
  2. It is not a signal. Nothing in it triggers an alert, enters a name, or feeds a valuation. No other part of the program reads it. It is read once a year, on purpose, because a scoreboard glanced at weekly is a reason to change a rule for the wrong reason.
  3. Both sides are the same kind of return, as above, or the row is missing.
  4. A verdict written on a weekend is re-dated to the last weekday before it, and the row says so. The walk-back crosses weekends and nothing else, because a weekday with no price could be a holiday or a data outage, and the tool cannot tell those apart.

Where this comes from, and where it does not

Nobody in finance has a standard name for this. The nearest relatives are the decision journal, a habit recommended by Daniel Kahneman and popularised by writers on decision-making: write down what you decided, why, and what you expect, before the outcome; and the distinction, made in the poker literature by Annie Duke and in investing by Michael Mauboussin, between judging a decision by its process and judging it by its outcome, where the second is a mistake with a name, "resulting".33Duke, Thinking in Bets (2018); Mauboussin, More Than You Know, popularising a process-versus-outcome matrix originally from Russo and Schoemaker. The decision-journal advice is widely attributed to Kahneman; a primary interview could not be located and is not cited. The underlying psychology is hindsight: once an outcome is known, people believe they expected it.44Fischhoff, 'Hindsight ≠ Foresight', Journal of Experimental Psychology: Human Perception and Performance 1 (1975); Fischhoff and Beyth, 'I knew it would happen', Organizational Behavior and Human Performance 13 (1975).

The shadow book departs from the decision-journal idea in one way, on purpose. A journal records the expectation. This book refuses to, because an expectation on record is an argument waiting to be resumed. It records only what was decided, when, and at what price, and lets the market supply the rest a year later.

Decided
Every refusal is recorded with its date, price and one-line reason, and measured against an index afterwards. Nothing else about it is written down, nothing in the program reads the book, and it is read once a year.
Why
A method that never buys has a perfect record that proves nothing. The only way to find out whether the refusals were right is to measure them, and the only way to keep the measurement from bending the rules is to look at it rarely.
Cost
The book cannot say whether a verdict was right, only what the price did. That is a weaker claim than it looks like, and the ruling says so in its first limit.

Sources for this chapter

  1. Hindsight bias, Wikipedia (Fischhoff 1975; Fischhoff and Beyth 1975)
    https://en.wikipedia.org/wiki/Hindsight_bias
  2. The shadow book ruling and code — read from this repository: reference/FRAMEWORK-EDITS.md E114 (2026-09-08, amended 2026-09-09); vss/shadowbook.py

The slow way, made survivable. Built 2026-09-13 from the repository as it stood that day. NAME A and NAME B are fictional; no company, price or holding appears on this site. Not financial advice: one private investor's method, shared as it is.