The manual
Around 1340 Francesco Pegolotti of the Bardi bank compiled La Pratica della Mercatura, a working manual for the overland China trade, from returning merchants’ reports. He never made the trip; he wrote at a desk in Florence. The numbers are the useful part.
| Element | What the record shows |
|---|---|
| Cargo, worked example | 25,000 gold florins |
| Transport, Black Sea to Beijing | 60 to 80 sommi of silver, 1 to 2 percent of cargo value |
| Financing, the commenda | investor’s capital against the merchant’s year of travel, profit split 75/25 in capital’s favour; Genoa’s notaries recorded more than eight thousand |
| The relay | oasis to oasis; a dozen hands took a dozen cuts, so the hundredfold markup belonged to twelve different people |
| The conversions | sell the linen at Urgench, carry the proceeds east as silver, surrender the silver at the Chinese border for the Khan’s paper currency, spend the paper on silk at fixed rates |
| Net return, a good year | near 10 percent, after the duties, the drivers, the interpreter and capital’s 75, for more than a year of mortal risk |
| The licence, the ortoq | a Mongol prince advances capital; the merchant carries a paiza, a stamped tablet that worked as passport, credit line and armed escort in one |
| The fine print | at first the prince ate losses up to his invested capital, an equity structure; by the late 1200s the same silver and the same paiza had become a loan, the debt personal and unlimited on failure |
| The tail | 1218, the governor of Otrar declared the richest convoy he had ever seen to be spies and executed nearly all of them |
| The end | plague on the same roads as the silk, the Yuan dynasty fell, Italian cities planted their own mulberry trees; within a generation the route was an asset class that no longer existed |
Almost nobody held a fortune of that size, so the trade ran on other people’s capital. The camels never made anyone rich; the paperwork did. What the manual documents is a financing structure and a relay of intermediaries wrapped around a cargo. The caravan was a bank in motion.
The licence and the fine print
Two clauses did the work: who holds the licence, and who bears the loss. The ortoq flip from equity to unlimited personal liability changed the trade without changing the asset, the route or the cargo.
The 2026 reading
Read as an underwriting document, the manual yields ten structural lessons. Five sit on joints my Forecast Wing had already priced, across fourteen pre-registered forecasts with binding resolution criteria; the sixth, the dragoman, is not priced yet.
| Lesson | The 2026 joint | Priced at |
|---|---|---|
| The relay | booking the full markup to one layer counts returns of seventy-five to one before the relay takes its cuts | F13, F14 |
| The paiza | export licences, sovereign compute deals and government equity stakes are the stamped tablet; the slate’s thesis, intelligence abundant and permission scarce | F1, F12, F5 |
| The fine print | who eats the loss on a financed GPU (graphics processing unit) fleet; tariffs financed as working capital; whether capital spending (capex) keeps compounding | F2, F7 |
| The caravansaries | watering stops sat one camel-day apart and spacing set throughput; the binding constraint now is electrical, interconnection queues and turbine lead times, read at the PJM Interconnection’s capacity auction | F8 |
| The dragoman | the integration layer between models and domains; “a good interpreter’s wages will cost you less than what a bad one loses you” | not yet |
| The mulberry trees | the parties being disintermediated are planting now; whether the Chinese fab sector’s trees mature | F5 |
One caveat on the caravansaries, from a quant who once owned Citadel’s power model. A queued megawatt is not a price. New load moves the power price only when it exceeds the marginal unit’s headroom, so a 500 MW data centre against a one-gigawatt plant at the margin moves nothing, and an arrangement that runs off-grid never reaches the auction at all. The constraint is real; reading announced megawatts as price pressure is a separate claim, and this note does not make it (added 21 August 2026).
What this does not claim
This is an essay, not a statistical result. Nothing here passed the pass/fail tests this site’s pipeline applies to trading results, and no historical analogy resolves a forecast; the fourteen forecasts resolve on their own binding criteria between 2027 and 2030. Five structural features of a trade seven centuries old correspond to joints I had priced before reading the manual. That is weak evidence that the slate is testing the structure that bears the load in this cycle, and it is exactly as weak as stated.
Pegolotti wrote that the road to Cathay was “perfectly safe, whether by day or by night,” and he never left Florence. Most guides to the AI trade are written at desks.
References
- Pegolotti, F. B. (c. 1340). La Pratica della Mercatura. Ed. Allan Evans, Mediaeval Academy of America, 1936.
- Lopez, R. S. & Raymond, I. W. (1955). Medieval Trade in the Mediterranean World. Columbia University Press (commenda contracts).
- Allsen, T. T. (1989). Mongolian princes and their merchant partners, 1200–1260. Asia Major 2(2) (the ortoq system).
- Richthofen, F. von (1877). China: Ergebnisse eigener Reisen (coinage of “Seidenstrasse”).
- Sims-Williams, N. (2001). The Sogdian Ancient Letters (the Dunhuang mailbag).
- Dimopoulos, V. (2026). The Forecast Wing: fourteen pre-registered forecasts. theriskmuseum.com/forecasts.html.