The slow way, made survivable Chapter 8 of 23 Svenska

The map · Part One · The method · Chapter 8 of 23

Pre-registration: write the belief before you see the number

Calculate what the price assumes first and whatever you write afterwards lands suspiciously close to it. Every human does this. So the belief is written first, timestamped, and a view written after the number was seen is void. About 10 minutes · on the 10-minute route.

The last chapter ended with a question: do I believe the growth the price is assuming? Here is the problem with answering it.

The natural order, which is wrong

You calculate that NAME A's price assumes 11 percent growth a year. Now you sit down to write what you think NAME A will actually grow. You have read the reports. You think about it carefully. You write down 10 percent, with reasons.

The reasons are real. The number is not. It is 11 with a small discount for modesty. If the price had assumed 6 percent, you would have written 5 or 7, with equally real reasons. The number you saw first became the centre of the range you thought you were choosing freely.

This is not a character flaw. It is one of the most reliably reproduced findings in the study of judgement. In a 1974 experiment, people watched a wheel of fortune stop at a number, then estimated what share of African countries belonged to the United Nations. The wheel was rigged to stop at 10 or at 65. The people who saw 10 guessed about 25 percent; the people who saw 65 guessed about 45. They knew the wheel was random. It did not help.11Tversky and Kahneman, 'Judgment under Uncertainty: Heuristics and Biases', Science 185 (1974). The effect is called anchoring. A later study gave estate agents a tour of a house and a listing price; the agents' valuations tracked the listing price they had been shown, and they denied being influenced by it.22Northcraft and Neale, 'Experts, Amateurs, and Real Estate', Organizational Behavior and Human Decision Processes 39 (1987).

Add to this the tendency to read evidence in favour of the belief you already hold, documented in hundreds of studies under the name confirmation bias,33Nickerson, 'Confirmation Bias: A Ubiquitous Phenomenon in Many Guises', Review of General Psychology 2 (1998). and the picture is clear. A person who has seen the market's number and then writes their own is not writing their own.

The rule

The rule, in force since 2026-08-25

Before the growth implied by the price is calculated, the owner writes three growth rates, a low case, a base case and a high case, each with its reasons, into a file that is timestamped and kept under version control. Only then is the implied number solved. If the implied number was seen first, the view is void for that company in that round, and cannot be used.

The two orderings of pre-registrationOrdering A, marked with a cross: solve the growth the price assumes, then write your own view; a dashed arrow shows the view drifting toward the number just seen; the result is VOID. Ordering B, marked with a tick: write your view first with a timestamp, then solve, then compare. ordering A: look, then writesolve the growththe price assumeswrite your owngrowth viewthe view drifts toward the number just seenVOID for this namein this cycleordering B: write, then lookwrite your growthview, with reasons,timestampedsolve the growththe price assumescompare: is the marketasking more thanyou believe?the timestamp is the proof of order
The same two steps in two orders. In A, the market's number is seen first and pulls the view toward it. In B, the view is on record before the number exists, and the record's timestamp is the proof.

The ordering is the whole mechanism. Nothing about the arithmetic changes. What changes is that the owner's belief exists as a dated record before the number that could have bent it. When the two are compared, the comparison means something: either the market is asking for more growth than the owner believes, or less, and the owner's side of that comparison was not manufactured to fit.

The record is enforced by the version-control history, which cannot be edited without leaving a trace. A superseded view is not deleted; it is marked as superseded and left in the file, so that the sequence of what the owner believed, and when, is preserved.44Ruling E28: 'A view written after the implied growth has been seen is VOID and may not be used for that name in that cycle.' Ruling E95: a superseded view is marked, kept, and refused by the code. The view files record what figures had already been printed before the view was written, which is what makes a late view detectable.

Where this comes from: the clinical trial

Science had exactly this problem and solved it the same way. A researcher who collects data first and then decides what hypothesis to test can always find something that looks significant, by choosing after the fact which outcome to report. The practices have names: p-hacking, trying analyses until one works; HARKing, hypothesising after the results are known; outcome switching, quietly reporting a different outcome from the one the trial was designed to measure.55Simmons, Nelson and Simonsohn, 'False-Positive Psychology', Psychological Science 22 (2011); Kerr, 'HARKing', Personality and Social Psychology Review 2 (1998); Chan et al., JAMA 291 (2004), finding that 62 percent of trials had at least one primary outcome changed, introduced or omitted against the protocol.

The fix was pre-registration: write down the hypothesis and the outcome you will measure, in a public registry, before collecting the data. The evidence that it matters is stark. A study of large, expensive heart-disease trials funded by one American agency found that 57 percent of trials published before 2000 reported a significant benefit. After 2000, when registration in advance became required, that fell to 8 percent. Same kind of trial, same kind of funding. The difference was that the answer could no longer be chosen after the fact.66Kaplan and Irvin, 'Likelihood of Null Effects of Large NHLBI Clinical Trials Has Increased over Time', PLOS ONE 10 (2015): '17 of 30 studies (57%) published prior to 2000 showed a significant benefit' against '2 among the 25 (8%) trials published after 2000'; 'Pre-registration in clinical trials.gov was strongly associated with the trend toward null findings.'

This project borrows the mechanism whole. The hypothesis is the growth view. The data is one number, the growth implied by the price. The registry is a file under version control. The penalty for looking first is that the view does not count.

What it costs, said plainly

The method's estimate of value now rests on a growth assumption that is a stated opinion and will never be verified. The ruling that introduced it says so in those words. Earlier rulings had raised the evidence bar on figures that move the answer by less than one percent; this one moves it by thirty. The project's answer is that verification effort should follow how much a number matters, not how easy it is to check, and that the one number that matters most is the one that can only be protected by ordering, because it cannot be checked at all.77From the framework's §5 as amended by E28: 'Verification effort follows sensitivity, not availability.'

Where this comes from
  • Anchoring: Tversky and Kahneman, Science 185 (1974), the wheel-of-fortune experiment. Northcraft and Neale (1987) on estate agents.
  • Confirmation bias: Nickerson (1998); the original demonstration is Wason (1960).
  • Pre-registration in science: Simmons, Nelson and Simonsohn (2011); Kerr (1998); Chan et al. (2004); Kaplan and Irvin (2015); Nosek et al., 'The preregistration revolution', PNAS 115 (2018).
  • This project's rule: ruling E28 (2026-08-25) and E95 (superseded views), reference/growth-views/.

Where this differs from its source: a scientific pre-registration is public and the hypothesis is tested against many data points. Here the registry is private, the "data" is one number, and the penalty is that the view is void rather than that a paper is rejected. The mechanism is the same: the belief must exist, dated, before the thing that could bend it.

Sources for this chapter

  1. Tversky and Kahneman (1974), Judgment under Uncertainty, Science
    https://www.science.org/doi/10.1126/science.185.4157.1124
  2. Anchoring effect (the wheel experiment described), Wikipedia
    https://en.wikipedia.org/wiki/Anchoring_effect
  3. Northcraft and Neale (1987), Experts, Amateurs, and Real Estate
    https://www.sciencedirect.com/science/article/abs/pii/074959788790046X
  4. Nickerson (1998), Confirmation Bias
    https://journals.sagepub.com/doi/abs/10.1037/1089-2680.2.2.175
  5. Simmons, Nelson and Simonsohn (2011), False-Positive Psychology
    https://journals.sagepub.com/doi/10.1177/0956797611417632
  6. Kerr (1998), HARKing
    https://journals.sagepub.com/doi/10.1207/s15327957pspr0203_4
  7. Chan et al. (2004), Selective reporting of outcomes in randomized trials, JAMA
    https://jamanetwork.com/journals/jama/fullarticle/198809
  8. Kaplan and Irvin (2015), Likelihood of Null Effects of Large NHLBI Clinical Trials Has Increased over Time, PLOS ONE
    https://doi.org/10.1371/journal.pone.0132382
  9. Nosek et al. (2018), The preregistration revolution, PNAS
    https://www.pnas.org/doi/abs/10.1073/pnas.1708274114
  10. The pre-registration rule — read from this repository: reference/FRAMEWORK-EDITS.md E28, E95; reference/growth-views/

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.