The Ledger · Near — one to five years

From The Rational Prophet (Book Four of The Laws of Game Theory)

The bet

The bet

As European defence spending rises, new arms agreements signed with the United States government, measured against what the same states spend on equipment, hold at or above their average over the five years published before this book went to press.

Win

The share, averaged over the years first published inside the window, is at or above its five-year average at baseline, while equipment spending rises.

Lose

The share falls below that average while spending rises.

Expiry

Two to three years. If equipment spending does not rise, the bet scores VOID.

Check it

The US Defense Security Cooperation Agency's annual Historical Sales Book, for the value of agreements signed with each state; NATO's annual report on defence expenditure, for what each spends on equipment. Both free.

In protocol form

Word for word from Appendix D of the book, the Prediction Protocol, where this bet is entered as The procurement. References to sections (§) are to that appendix.

Universe

the member states of the European Union that are NATO allies, plus the United Kingdom and Norway — the states for which NATO publishes equipment spending. Austria, Cyprus, Ireland and Malta are outside it because NATO does not report them. Finland and Sweden are counted in every baseline year, including the years before they joined the alliance, because the universe is fixed at publication and NATO publishes their figures for those years. The incumbent supplier is the United States.

Source

for the numerator, the Historical Sales Book of the US Defense Security Cooperation Agency, annual and free: the total authorised value of foreign military sales agreements implemented with each state in each US fiscal year, including those paid for with US military financing. For the denominator, NATO's annual Defence Expenditure of NATO Countries: each ally's defence expenditure in current US dollars, multiplied by the share it spends on equipment, for the calendar year that ends in the same year as the US fiscal year. From its 2026 edition NATO calls this series core defence expenditure; it is the same series under a new name, and it is the one used. The arms transfers database is not used: it measures transfers in trend indicator values, a proxy for capability, not the value of signed agreements.

Definitions

the share for a year is the sum of agreement values over the universe divided by the equipment spending of the same states. The baseline years are read from the editions of both sources current at publication. A year that first appears after publication is read from the first edition of each source that publishes it; later revisions of a year already read are ignored. If NATO publishes no figure for a state in a year, its equipment spending is carried at its latest published figure and its agreements are not counted for that year: missing evidence counts against the bet.

Coverage

the numerator holds only agreements between governments. Direct contracts between US companies and European ministries are not in it, because no free source publishes their signed value; the US government's annual report on export licences counts licences, not contracts. So the share is lower than the true US share of what Europe signs, and a shift from one channel to the other would move it without any change in dependence. That risk is stated rather than removed. The share is also an index rather than a market share: the numerator is what was signed in a year, the denominator what was spent.

Calculation

the average share over the years first published inside the window, against the average of the five most recent years published at publication.

Baseline

the share for each of those five years and their average.

Note

if the universe's equipment spending in the latest year published at expiry is not higher than in the latest year published at publication, the bet is VOID: the rearmament it depends on did not happen.

Baseline at publication

Reading date of record: 10 October 2026

What is recorded

US government-to-government agreements as a share of equipment spending: five most recent years published, and their average, from the editions current at publication

Value at publication

2020: 14.2% · 2021: 12.0% · 2022: 24.5% · 2023: 26.7% · 2024: 33.1% · average 22.1%. Agreements: total authorised value of implemented FMS and FMF cases, US fiscal years 2020–2024, Historical Sales Book, fiscal year 2024 edition; equipment spending: core defence expenditure in current US dollars times the equipment share, calendar years 2020–2024, NATO edition of July 2026; 25 states, Finland and Sweden included in every year. Equipment spending of the same states in the latest year published, 2026 (estimate): $231.6 billion (read 10 October 2026)

Author’s probability 85% that this bet scores HIT. Registered on 10 October 2026.

Sources

• DSCA Historical Sales Book
• NATO, Defence Expenditure of NATO Countries (2025)

Not investment advice. The bet describes what the book expects to happen; nothing here is a recommendation to buy, sell or hold anything.

Last reviewed 10 October 2026