Does systematic equity positioning reach the 85th percentile of its two-year range?
- Market
- 46%
- Ours
- 71%
- Gap
- +25
Maxims of the House
I. Probability over prediction
No model is asked for certainty where none exists.
II. Disagreement is information
Models shall not agree merely because agreement is comfortable.
III. Calibration before reputation
Judged by outcomes, never by eloquence.
Q04 Positioning — market 46 · ours 71 · gap +25
Default Assumptions Are Inputs, Not Answers.
Being an account of a machine that prices the futures consensus forgot.
The engine begins with questions, never with price targets. It forecasts each one by independent agents, measures where it disagrees with the market, and acts only where that disagreement survives every scrutiny it can devise. Each forecast is sealed upon issue and kept, so that the record may be examined by anyone, at any time, and most of all when it was wrong.
Leading Intelligence
Does systematic equity positioning reach the 85th percentile of its two-year range?
Market 46%Ours 71%+25
Does aggregate dollar liquidity expand through October?
Market 42%Ours 63%+21
Does the 60-day equity–bond correlation turn positive?
Market 29%Ours 46%+17
Sealed on issue.
No forecast in these pages is revised after publication. 36 are already on record.
Machinery in Motion
The Engine Room
The Counting Wheels
Liquidity
Market0123456789+−0123456789+−
Ours0123456789+−0123456789+−
Gap0123456789+−0123456789+−0123456789+−
Question 1 of 10. The wheels turn when the gap is struck.
The Teleprinter
AWAITING CYCLEEach agent answers alone. The machine pools them, sets the result against the market, and seals it.
Specimen of Instruction
- 10FOR EACH Q IN OUTLOOK
- 20 LET P(I) = AGENT(I).FORECAST(Q)
- 30 LET F = POOL(P, ALPHA)
- 40 LET G = F - MARKET(Q)
- 50 IF NOT ADMISSIBLE(G) THEN 70
- 60 ALLOCATE Q, G
- 70 SEAL Q, F
- 80NEXT Q
Line 50 is where most ideas end. Of this month’s 10 questions, 3 pass it.
Part the First
The Method, in Seven Parts
I.
Observe
Markets and external information are collected and set in order. Nothing is yet interpreted.
II.
Decompose
Each month is broken into some ten questions answerable Yes or No, frozen before any forecast is made.
III.
Forecast
Independent agents forecast every question without sight of one another or of the market.
IV.
Disagree
Our probability is set against the market’s. The difference is the consensus gap.
V.
Allocate
Only gaps surviving confidence, diversity, liquidity, downside and cost become exposure.
VI.
Execute
Execution is programmatic. What the market charges us to act is taken as evidence.
VII.
Learn
Every forecast is scored and kept. The worst months are printed first.
“What must be true for the expectation to be correct?”
Part the Second · October 2026
Where We Disagree This Month
Does aggregate dollar liquidity expand through October?
- Market
- 42%
- Ours
- 63%
- Gap
- +21
Does the 60-day equity–bond correlation turn positive?
- Market
- 29%
- Ours
- 46%
- Gap
- +17
State of the Strategy
- Regime
- Constructive / Crowding
- Risk budget
- 0.58
- Conviction
- Moderate
Part the Third
The Record, Kept in Full
July MMXXVI was our worst month. It is printed in full, as every failure shall be.
| Questions resolved | 36 |
|---|---|
| Brier score, ours | 0.222 |
| Brier score, the market’s | 0.229 |
| Skill against the market | +2.8 per cent. |
Part the Fourth
Recent Dispatches
Models Fail Before Narratives Do
On regime change, model decay and the long half-life of a good story.
Execution Is Part of Intelligence
A forecast that cannot be acted upon is an opinion with decimal places.
Calibration Before Conviction
A post-mortem on July. Our probability was wrong. More importantly, it was badly calibrated.
The Consensus Gap
Measuring the distance between what everyone expects and what we believe — and why distance is never enough.