The Rational Prophet
25 Game-Theory Predictions About Power, Money and War — and How to Prove Them Wrong
Book Four of The Laws of Game Theory
Nostradamus wrote in riddles. He had to. A prophecy vague enough can never be wrong, and a prophecy that can never be wrong can never be useful.
This book makes the opposite bet. Every prediction in it is numbered, tied to a public source, given a threshold that decides it, and written so that it can fail. They are all collected in an appendix — the Prediction Ledger — where you are invited to do something no reader of Nostradamus could ever do: keep score.
There is no claimed accuracy rate anywhere in the book.
Why thirty years can be easier than three
It sounds backwards, and it is the argument the whole book rests on.
Over three years, noise dominates: personalities, accidents, elections, luck. Over thirty years, noise averages out and structure is what remains. Everyone who will be forty in 2060 has already been born. Most of the debt that will shape the 2030s has already been issued. The geography that will decide the resource conflicts of the 2040s has been fixed since the ice age.
That holds for the slow variables, not for everything. A specific event that needs several things to go one way before a deadline is not easier to call at long range, and a shock can still break a trend. That is why every bet in the book carries a window and the observation that would refute it.
The prophet who works from structure is not seeing the future. He is reading the present carefully.
How the book is built
Part I sets out the method honestly, including what it cannot do — and scores the predictions made in the earlier volumes of this series, failures first. A forecaster who will not grade his own record is asking for something he will not give.
Part II takes the method into the archive: five centuries of empire cycles, a century of monetary breakdowns, and the clustering of great wars around power transitions. A framework that cannot explain the past has no licence to describe the future.
Parts III, IV and V play the three great games forward: Power, Money and War. Part VI puts the whole arc in order — which things break first, and what the arrangement that follows looks like.
What is in the ledger
The entries are sorted into three bands: near (one to five years), medium (two to fifteen), and far (twenty to thirty-five).
Each entry is built the same way. It names the game and the players who can actually choose, and what each is optimising for. It explains why the outcome follows. It names the consensus claim it is betting against — and the rival explanation that would account for the same evidence, with the observation that separates the two.
Then it commits. The bet in one sentence. Check it — the public dataset, named. Win and lose, as thresholds rather than impressions. Expiry. And the baseline, calculated and printed at publication so it cannot be moved afterwards.
Every one of them carries a losing condition. Not one of them can be quietly reinterpreted.
Pre-registered on the Open Science Framework
The bets, their definitions, the baselines read on publication day and the author’s probability for each are also pre-registered on the Open Science Framework, an independent archive the author cannot edit afterwards: osf.io/7ev8k, DOI 10.17605/OSF.IO/7EV8K.
A sample
GP-12 — The debt is repaid by repression, not default.
The bet: The debt cycle resolves through negative real returns to savers rather than through default or a growth miracle.
Check it: Ten-year sovereign yields from FRED and the OECD; the headline consumer price index of each national statistics office.
Loses if: Sustained positive real yields alongside a falling debt ratio, or a large-scale formal restructuring.
GP-19 — Taiwan's status changes without a great-power war.
The bet: The island's status changes — by absorption, neutralisation or settled autonomy — without sustained combat between the two great powers.
Check it: Conflict event data plus official confirmation.
Wins if: The status changes, and no engagement between their forces exceeds seven consecutive days or a hundred combat deaths on either side.
Loses if: Either threshold is crossed, or the status is unchanged at expiry. Isolated incidents below it do not count.
GP-13 — Retirement past seventy, with no transition.
The bet: At least three G7 states legislate a retirement age above seventy, with no transitional arrangement for anyone under forty-five at the time.
Check it: The OECD's pension survey.
Loses if: Fewer than three G7 states enact it inside the band.
Who it is for
Readers who want a structural, incentive-first framework for the coming decades rather than commentary on the last one — and who would rather have a forecast that can be scored than one that can always be reinterpreted.
Buy on Amazon Paperback Read the weekly analysis → Next: Book Five →
Not investment advice. The chapters on money describe what states, central banks and creditors have historically done under fiscal pressure, and what it cost the parties who misread it. Nothing in the book is a recommendation to buy, sell or hold anything.
Last reviewed 19 October 2026