@sdelsad/commodity-desk-daily 1.0.79 → 1.0.80

package/covered.md

@@ -32,3 +32,4 @@ Running log. Read before writing a new episode: avoid repeating material, and on
 - **Ep 28** (Fri) — *The Energy Link*: Ep 28 — The Energy Link: agriculture as an energy derivative, organised as three chains on three clocks - same session, weeks, a full season; unit moments on the barrel (158.99 litres, so a dollar a barrel is six tenths of a cent a litre, from the 42-gallon Pennsylvania cask) and on the energy-equivalence ratio (a litre of ethanol carries about two thirds of petrol's energy so hydrous competes at roughly 70 percent of the pump price, which is the attenuation in the first chain); CHAIN ONE crude into sugar, cleared same session, built as an explicit transfer function from the ep14 CONSECANA factors - one tonne of ATR makes 591.3 litres of hydrous or 952.8 kg of sugar so a litre of ethanol and 1.612 kg of sugar come from the same cane, equate the revenues and a dollar per litre of hydrous is 0.6205 per kg of sugar which at 2204.62 lb/t is 28.15 c/lb, hence every cent a litre on hydrous is 0.2815 of a cent a pound on the parity; the ten-dollar walk - 10 USD/bbl divided by 158.99 = 6.29 c/l of crude, one-for-one into petrol at import parity, times 0.70 = 4.40 c/l on hydrous, times 0.2815 = 1.24 c/lb of parity, so ten dollars of Brent is a cent and a quarter of sugar parity and a dollar is about an eighth of a cent; the year of crude applied to ep14's floor - Brent 65.96 to 103.86 is 37.90 dollars and 4.70 cents of parity, taking ep14's 12.60 parity to about 17.30 against the 18.66 print of late August, so headroom above parity collapsed from about 109 USD/t to under 30 and the floor came up to the price rather than the price coming down to the floor (ep14 switch-is-spent argument taken to its state-dependent form); the switch-not-coefficient point - far above parity a sugar rally pulls no extra Brazilian tonnes and can only ration demand so crude is noise, close to parity the mill is indifferent and sugar is a crude price in a different unit, so a five-year regression of sugar on crude averages a switch that was on with a switch that was off and describes no market that ever existed; TRADER/ANALYST dialogue on the distance to parity closing on long sugar being a crude position nobody asked for, plus the logistics asymmetry that the sugar route pays mill-to-port and the domestic ethanol route does not so a 19-point screen gap can already be the wrong side of the decision at the mill gate; the FX caveat with the sign reversed from ep27 - the floor is set in reais and read in cents a pound so a weaker real lowers the dollar parity with crude unchanged; CHAIN TWO gas into fertiliser into the cost curve into acres, one season, with the textbook cheap-gas-cheap-nitrogen version refuted by the sourced month - anhydrous ammonia 843 to 998 USD/short ton and urea 579 to 823 (plus 42.1 percent) between 20 Feb and 20 Mar 2026 while Henry Hub sat at 3.62 USD/MMBtu, so the chain conducts through the international marginal tonne (export capacity, a conflict premium on Middle East ammonia, Chinese export licensing) and not through the domestic gas bill, nitrogen being priced where the last clearing tonne is made and almost never where it is spread; the per-acre and per-bushel walk insisting on pounds of actual nitrogen rather than tons of product - urea at 46 percent N gives 0.6294 and 0.8946 USD per lb of N, 113.28 and 161.02 USD per acre at 180 lb N, a 47.74 dollar move, 26.5 cents a bushel at 180 bu/acre and 5.3 percent of a 500.50 December board, with the comparability trap that urea is 46 percent, UAN 32 and ammonia 82 so comparing 823 urea with 998 ammonia without dividing by the analysis names the wrong one as expensive; the soft-floor definition - a cost curve does not stop a price falling because the farmer's in-season marginal cost is near zero and sunk costs do not enter a marginal decision, so the floor bites at planting and not at harvest, and it bites as a relative price because nitrogen is a corn cost and barely a soybean cost, making a nitrogen shock an argument about which crop gets the acre rather than a bullish grain story, March fertiliser becoming June acres, October yield and a January balance sheet; CHAIN THREE bunkers into freight into the arb in weeks (ep10 mechanics assumed), fuel as the largest variable voyage cost so the move lands on differentials and origin basis before the flat price, with retail diesel at 6.38 USD/gal inside every truck and barge rate in the basis stack; the closing argument that the desk ignoring energy trades a lagged version of a market that has already printed since every link was knowable on the day crude moved, and that the commoner error among those who have noticed the link is one number for all three chains because the gearing is state-dependent in sugar, seasonal in fertiliser and nearly immediate in freight, so energy gives an ag desk a schedule of things that have already happened rather than a correlation to trade. Pulse: Thu 1 Oct settles Dec corn 500.50 -0.25, Nov beans 1289.25 -3.75, Dec Chi wheat 674.75 -1.00, Matif Dec 236 EUR/t unchanged, Brent 103.86 up 6.7 percent on the month from 97.38 and 57 percent on the year from 65.96; the Wednesday 30 Sep Grain Stocks print as the week's whole move with 1 Sep corn stocks at 2.095bn bu against a trade estimate near 1.92bn, above every figure in the survey and 35 percent above a year ago, corn -21.25 in the session and under its 50-day average for the first time since July on a fourth straight lower close, new-crop corn export sales 535,989 t the smallest of four weeks; Gulf corn basis firming about 10 cents while the board broke, used as the basis-versus-flat-price detail and later as the diesel-in-the-basis-stack hook; four-session Dec corn line chart 528.25 / 523.00 / 522.00 / 500.75 / 500.50; GEO rotation to EXPORT-CAPACITY REVISION with an inter-exchange consequence - SovEcon cutting Russian wheat exports 4.7 Mt to 36.7 Mt on Azov and Black Sea loading expected shut into 2027, Matif holding near 236 EUR while Chicago fell because the missing capacity is European-facing and the replacement buyer does not take soft red winter, plus reports of China looking at Australian wheat as flow substitution that moves a differential before a flat price (ep26 Black Sea lesson escalated to a quantified forecast revision and its two-screen split).
 - **Ep 29** (Mon) — *Life of a Trade, and the Desk Around It*: Ep 29 - Life of a trade and the desk around it: one 30,000 t cargo of #2 yellow corn sold CFR Veracruz at Dec +150 and bought FOB Gulf at Dec +92, hedged in two minutes with 30,000 t x 39.368 = 1,181,040 bu = 236 Chicago lots and 1,040 bu of unhedgeable residue; the five risks owned at the word done - flat price, origin, freight, buyer credit and execution - with four shed inside a week and execution carried to discharge; the planned margin worked twice, 58 c/bu gross less freight 48.26 and costs 6.35 = 3.39 c/bu, and 0.58 x 39.368 = 22.83 USD/t less 19.00 freight and 2.50 costs = 1.33 USD/t = 40,003 dollars against a cargo worth 7,650,300 at a CFR of 255.01 USD/t, so the merchant margin is half a percent of the thing being moved and that single ratio governs everything else; the central depth argument that one cargo exists on paper three separate times - a GAFTA sale contract, a charter party and a letter of credit - and that the money leaks in the gaps between them rather than on the price view, worked as three sized gaps; GAP ONE time, charter 3 days laytime all purposes at 18,000/day against a sale contract discharge rate of 8,000 t/day giving 3.75 days at 12,000/day, vessel loading 1 day and discharging 6, so 4 days owed to the owner = 72,000 against 2.25 days recoverable from the buyer = 27,000 leaving 45,000 uncovered which exceeds the whole planned margin, with the point that nothing went wrong at the port and two contracts simply measured the same seven days with a different allowance and a different daily rate; GAP TWO weight, outturn 29,880 t against a bill of lading 30,000 t, 120 t or four tenths of one percent, and the certificate-final clause deciding everything - final at load port and the loss is the buyer's, final at discharge and 120 x 255.01 = 30,601 dollars or three quarters of the margin; GAP THREE the bank, a documentary discrepancy delaying payment a fortnight on 7,650,300 at 6 percent costing 17,851 dollars or 45 percent of the margin (ep23 callback taken to a number); the punchline comparison that the 1,040 bu of unhedgeable residue on a 20-cent move is worth 208 dollars while the three gaps with no market against them are worth 45,000, 30,601 and 17,851, so the risk the industry measures is 0.5 percent of the risk it does not; the desk ecosystem mapped onto those numbers - trader decides the position, execution desk works the futures, analysts own the balance sheet, risk owns limits and marks, traffic or operations owns nomination surveyor laytime and documents, finance owns the letter of credit and working capital, legal owns the three contracts and the arbitration - with 208 dollars belonging to the trader and the other 93,452 belonging to traffic, finance and legal; TRADER/TRAFFIC dialogue at the moment of nomination in which traffic converts a laytime mismatch into a cent and a half a bushel before the cargo exists, against three and a third cents of margin; the problem cargo routed through five functions - traffic appoints the surveyor, legal says which certificate binds, finance decides whether to present or hold the documents, the trader decides whether to settle commercially because the buyer returns in February, and risk asks whether the book is still flat, which it is not because the house is short 120 t already hedged and already shipped; the closing asymmetry argument that every gap runs against the merchant by selection rather than luck because the owner drafts the charter to protect the ship and the buyer's bank drafts the credit to protect the buyer while the merchant is the only party sitting between two, so back-to-back is redefined as clause matching rather than price matching - same shipment period, tolerance, survey, demurrage rate and arbitration - and a trader who matches the price and not the paper has decorated the trade rather than hedged it. Pulse: Fri 2 Oct settles Dec corn 497.75 -4.50, Nov beans 1278.25 -5.75, Dec Chi wheat 683.00 +0.25, Dec meal 347.50 -5.80, Dec bean oil 68.62 +124 pts, ICE raw sugar 18.92 +1.7 percent at an eighteen-month high; corn a fourth consecutive lower weekly close on improved harvest conditions, soft export commitments and a near-record speculative long, with the weekly declines led by Kansas City down over 25 cents against Chicago and Minneapolis down 15 to 20; StoneX US corn 16.115bn bu at 182.1 bu/acre against USDA September 15.80bn and StoneX soybeans 4.648bn at 54.1 against 4.535bn, both private numbers above official three days before the Thursday 9 Oct WASDE; August crush 210m bu and ethanol grind 478m bu in line to firm with the board crush margin up 9.5 cents to 2.44 USD/bu; USDA flash sale 218,600 t corn to Mexico used as the bridge into the cargo lesson, and Tunisia buying 100,000 t durum from just over 358 to just under 384 USD/t; GEO escalation of the Black Sea thread onto a THIRD CLOCK - the Russian president rejecting the Black Sea truce proposals over the weekend so the corridor probability that had been draining out of the board since ep27 has nothing left to price, while Ukraine's agriculture ministry puts winter crop area down about 15 percent, winter grains about 17 and winter rapeseed about 7, naming farm economics rather than shelling, with the transmission distinguished explicitly by clock - a damaged berth reprices a differential this week, a cut export forecast reprices a season, lost winter area reprices next July through the farmer's margin rather than through loading capacity, so it belongs in the deferred months and not the prompt (ep26 capacity lesson and ep28 margin-into-acres chain joined from the price side).
 - **Ep 30** (Wed) — *The Houses, and the Trader's Checklist*: Ep 30 - The Houses, and the Trader's Checklist: the ABCD houses as information arbitrageurs (Bunge 1818, Louis Dreyfus 1851, Cargill 1865, ADM 1902) and the 1972 great grain robbery as the high-water mark of that edge, destroyed by satellites and a simultaneous USDA release so the margin migrated from knowing to owning; asset optionality as the reason the asset-heavy model won - elevator, crush plant, berth and time-chartered vessel each a right without an obligation that pays whether or not the view is good; Bunge-Viterra closed July 2025 at about 8bn of equity and cash plus nearly 10bn assumed debt, 100bn revenue, 230 Mt marketed, 56 crush plants and 26 port terminals, worked to 434.78 USD/t of revenue and a half-percent margin of 2.17 USD/t giving about 500m on the whole book; the arithmetic of a cent with 30,000,000 t x 39.368 = 1,181,040,000 bu so one cent is 11,810,400 a year against 1,968,400 on a 5 Mt book, consolidation explained with no theory of market power; ABCD-plus with COFCO buying Nidera and Noble Agri to own origination in Brazil and Argentina, and EUDR traceability from December 2026 splitting one commodity into compliant and non-compliant as an infrastructure problem; the seven-question checklist in order - name the risk, the counterparty's motive, the named-day downside, funding, exit, mandate, the cost of being right - with direction deliberately outside the first five; worked example 60,000 t Gulf corn bought Dec +72 sold CFR Alexandria Dec +158, freight 23.00 and ifa 2.20, giving 0.86 x 39.368 = 33.86/t less 25.20 = 8.66/t = 519,389, hedged with 2,362,080 bu = 472 lots, a 90c rally paying out 2,124,000 of variation margin or 4.1 times the entire margin while the physical gain stays unrealised to payment, so a merchant's credit line is the real position limit; the depth point that recovering a drawdown costs more than the drawdown (25 percent needs 33, 50 needs 100) and survival is a constraint on the optimisation rather than a style, with ep21 quantity-risk, ep22 stress-test, ep29 half-percent-margin callbacks. Pulse: Tue 6 Oct settles Dec corn 508.00 +10.75, Nov beans 1303.00 +22.25, Dec Chi wheat 704.25 +12.00, Dec meal 354.80 +7.70, Dec bean oil 69.91 +56 pts, read against Fri 2 Oct so wheat +21.25, beans +24.75, corn +10.25 and the first session to interrupt four straight lower weekly closes in corn; wheat leading for a third session with a four-session indexed corn-versus-wheat chart 500.50/497.75/497.25/508.00 against 674.75/683.00/692.25/704.25; the Thursday 9 Oct WASDE escalated from ep29's private-above-official framing to the DISPERSION itself at a 173 to 182 bu/acre trade range plus flood quality discounts of 3-5 percent on some bean lots that no official yield number will carry; crop progress corn 23 percent harvested against a 27 percent average and beans 25 against 33 with Iowa the laggard; GEO escalation of the Black Sea thread from ep28's forecast revision and ep29's refused truce to REALISED TONNAGE - Ukrainian September exports -36.6 percent year on year after renewed strikes on Odesa while Russia shipped a record month with Baltic ports alone clearing more than 1 Mt, so the world's tonnes were replaced from a different berth under a different flag and the cost landed on the importer, evidenced by Saudi Arabia buying 683,000 t (148,000 t above the 535,000 t sought) at 334.60 to 348.00 USD/t C&F averaging about 339, some 70 USD/t above July, for Nov-Dec shipment in eleven cargoes filled by Cargill, Bunge, Olam, CHS, Soufflet and Aston - used as the bridge into the houses lesson.
+- **Ep 31** (Fri) — *Cocoa: Two Terminals, the Ratio, and the State That Sells the Crop*: Ep 31 - Cocoa: Two Terminals, the Ratio, and the State That Sells the Crop: cocoa's two terminal markets as a unit moment - ICE New York in USD/t and ICE London cocoa No. 7 in GBP/t, both 10 t a lot - and the discovery that the New York minus London spread is largely a sterling position, worked from the 6 Oct Dec closes of 5,762 USD and 4,250 GBP at cable 1.3276 giving a London equivalent of 5,642.30 and a New York premium of 119.70/t, which is 2.1 percent of the price of cocoa while one cent of cable is 42.50/t of spread so 2.8 cents of sterling erases the whole premium and a 1 percent cable move is 56 dollars or nearly half the spread, with a TRADER/BROKER dialogue in which the trader refuses the spread until the rate is named and the European grinder who buys against London and sells butter in dollars identified as carrying the same unwritten cross-hedge leak (ep22 callback); the state as seller - Ivory Coast's Coffee-Cocoa Council forward-sells the crop in tranches before harvest, nearly 1 Mt of the 2026/27 main crop sold by June 2026, and sets the guaranteed farmgate price from the average of those completed sales rather than from the screen, cut on 1 September from 2,800 to 1,200 CFA/kg or about 2.05 USD/kg and 2,050 USD/t against a 5,762 terminal, with three consequences worked - the farmgate as a moving average of last year's forward curve so the supply response arrives a year after the prices that caused it, price risk becoming performance risk because the beans become worth more than the contract that bought them and the holder has every reason not to deliver at last year's price (ep23 callback), and the differential at an administered origin being an allocation plus a queue so the watched variables are contract allocation and port arrivals, evidenced by 1-13 September deliveries of 26,000 t down 45.8 percent year on year; cocoa's crush - 20 percent shell and loss leaving 800 kg of nib pressed into 400 kg butter and 400 kg cake, butter and powder quoted as ratios against the terminal rather than in dollars, combined ratio as the grinder's margin gauge, worked at 2.10 and 0.85 for a combined 2.95 giving 4,840.08 of butter plus 1,959.08 of powder against 5,762 of beans and 600 of processing for a margin of 437.16/t; and the depth point that the break-even combined ratio is not the constant 2.8 everybody quotes but (1 + c/P) divided by 0.4, so pure conversion breaks even at 2.50 and the fixed dollar processing cost lifts it to 2.76 at 5,762 beans, 2.86 at 4,200 and 3.00 at 3,000 - cheap beans require a BETTER ratio to break even, the margin falling from 437 to 156 on a 27 percent fall in beans and going negative at 3,000, read against Q2 grind of Europe -4.6 percent, North America +7.7 and Asia +25 as capacity migrating to where the ratio clears (ep8 crush contrast: the crush is quoted in dollars so the squeeze is visible, the combined ratio is a multiple so a grinder watching a flat 2.95 loses two thirds of his margin without his number moving). Pulse: Thu 8 Oct settles Dec corn 500.25 -1.75, Nov beans 1287.50 -10.00, Dec Chi wheat 683.25 -3.25, Dec meal 357.60 -8.20, Dec bean oil 67.92 -24 pts, with the WEEK against Fri 2 Oct netting to nothing - corn +2.50, beans +9.25, Chicago wheat +0.25 after a range of more than 21 cents, only meal doing real work at +10.10 while oil lost 70 points so the product spread moved and the beans barely did, and wheat's 29.5 cent three-session rally fully round-tripping with 21 cents given back in two; the WASDE thread escalated from ep29's private-above-official and ep30's trade range to the PRINT ITSELF landing today at 11:00 CT with the trade average corn yield at 177.8 bu/acre against USDA September 178.5 and a poll range of 173.2 to 182.1, nine bushels wide which on roughly 88.5m harvested acres is close to 800m bushels of production, plus the point that a yield estimate measures bushels and not grade so this crop's mould and sprouting reaches the market as a discount in the basis rather than as ending stocks; GEO escalation of the Black Sea thread to a FOURTH mechanism, the DOLLAR rather than the corridor - record Treasury yields and a firmer dollar pulled wheat down Wednesday with nothing happening to supply, and Russian wheat offered in dollars around 204/t FOB means a stronger dollar converts an unchanged offer into more roubles at the seller's bank so his willingness to offer rises while the importer's local-currency cost rises in the same session, making the seller keener and the buyer poorer simultaneously, which is why a closed corridor has stopped reaching the flat price at all - tonnes leaving via northern ports and rail with IKAR September exports near 2.5 Mt, SovEcon trimming the crop to 87.5 Mt and Jordan, the Philippines, Thailand and Morocco all buying this week (ep26 capacity, ep29 refused truce, ep30 realised tonnage joined from the currency side).

package/ep31.md

@@ -0,0 +1,214 @@
+# Cocoa: Two Terminals, the Ratio, and the State That Sells the Crop
+
+## Market pulse
+
+**The week in Chicago netted to nothing, and the October WASDE lands today at 11:00 CT.**
+
+| Commodity | Contract | Settle (8 Oct) | Change | Change on the week |
+|---|---|---|---|---|
+| Corn | Dec 26 | 500.25 ¢/bu | −1¾¢ | +2½¢ |
+| Soybeans | Nov 26 | 1287.50 ¢/bu | −10¢ | +9¼¢ |
+| Chicago wheat | Dec 26 | 683.25 ¢/bu | −3¼¢ | +¼¢ |
+| Soybean meal | Dec 26 | $357.60/st | −$8.20 | +$10.10 |
+| Soybean oil | Dec 26 | 67.92 ¢/lb | −24 pts | −70 pts |
+
+Wheat is the market that moved and did not go anywhere. It rallied 29½¢ in three sessions and gave 21¢ of it back in two, finishing the week a quarter of a cent higher. Only soybean meal did real work, up $10.10 while bean oil lost 70 points — the product spread moved, the beans barely did.
+
+The October **WASDE** prints at 11:00 CT today. The trade average for the US corn yield is **177.8 bu/acre** against USDA's September 178.5, and the poll range runs **173.2 to 182.1**. The average is a nine-tenths cut; the range is nine bushels wide, which on roughly 88½ million harvested acres is close to 800 million bushels of production. The number will land inside the range and still surprise most of the desks reading it. One thing the report cannot carry: this crop has mould and sprouting in it in places, and a yield estimate measures bushels, not grade. A bushel that fails on damage does not leave the balance sheet — it moves to a discount, and discounts live in the basis, not in ending stocks.
+
+**The geopolitical read.** This week's wheat move came through the dollar rather than the corridor. Record Treasury yields and a firmer dollar pulled the board lower on Wednesday with nothing happening to supply, and the transmission runs both ways at once. Russian wheat is offered in dollars, around **$204/t FOB**. A stronger dollar converts that unchanged offer into more roubles at the seller's bank, so his willingness to keep offering rises — while the importer's local-currency cost rises in the same session. A strong dollar makes the Russian seller keener and the buyer poorer simultaneously, which is why a closed Black Sea corridor has stopped reaching the flat price at all. The tonnes leave through northern ports and by rail: IKAR puts September exports near 2.5 Mt, SovEcon has trimmed the crop to 87.5 Mt, and Jordan, the Philippines, Thailand and Morocco all bought this week.
+
+```chart
+{"type":"bar","unit":"% change, 2–8 October",
+ "title":"The week netted to nothing",
+ "caption":"Four of the five markets finished the week within a percent of where they started. Only soybean meal did real work, and it did it against bean oil rather than against beans.",
+ "source":"CBOT settlements, 2 and 8 October 2026 (Brownfield Ag News; USDA AMS via IndexBox)",
+ "x":["Dec corn","Nov beans","Dec Chi wheat","Dec meal","Dec bean oil"],
+ "series":[{"name":"Week","values":[0.5,0.7,0.0,2.9,-1.0]}]}
+```
+
+Cocoa, today's subject, has no WASDE at all. No crop survey, no balance sheet released at a fixed minute. Its most important number is set by a state marketing board — and on 1 September that board cut it by 57%.
+
+## Key takeaways
+
+- Cocoa trades on two terminals in two currencies: ICE New York in dollars per tonne and ICE London (cocoa No. 7) in sterling per tonne, both 10 t a lot.
+- The New York minus London spread is roughly 2% of the price of cocoa, and one cent of sterling is $42.50/t of it. Less than three cents of cable erases the whole premium with no cocoa trading. Never quote that spread without naming the exchange rate.
+- Ivory Coast's crop is forward-sold by a state body before it is harvested, and the farmgate price is set from the average of those completed sales — not from the screen. The farmer's price therefore lags the terminal by a season in both directions.
+- When the terminal runs far above the level at which a pre-sold crop was priced, the beans become worth more than the contract that bought them, and the holder of the physical has every reason not to deliver. Price risk becomes performance risk: the hedge is perfect and the cargo does not load.
+- At an administered origin the differential is not a merchant's price. It is an administered number plus an allocation, so the variables worth watching are contract allocation and port arrivals.
+- Butter and powder are quoted as multiples of the terminal bean price. Add the two ratios and you have the **combined ratio**, the grinder's margin gauge.
+- The break-even combined ratio is not a constant. Pure conversion breaks even at 2.50 because a tonne of beans yields 800 kg of product; processing cost lifts it above that, and processing cost is fixed in dollars.
+- So cheap beans require a *better* combined ratio to break even, not a worse one. A collapsing bean price does not rescue the grind — it raises the bar, and capacity migrates to wherever the ratio clears.
+
+## Vocabulary
+
+| Term | What it means |
+|---|---|
+| **terminal market** | the futures exchange whose quoted price serves as the pricing reference for the physical trade, against which origin differentials and processing ratios are quoted |
+| **cocoa No. 7** | the ICE London cocoa futures contract, quoted in pounds sterling per tonne in lots of 10 tonnes, historically the terminal for West African cocoa |
+| **cable** | desk shorthand for the sterling–dollar exchange rate, the number that converts a London cocoa price into a New York one |
+| **nib** | the cocoa bean after the shell is removed, roughly 800 kg per tonne of beans, and the material that is actually pressed |
+| **press cake** | the solid residue left after cocoa butter is pressed out of the nib, milled into cocoa powder and typically retaining 10–12% fat |
+| **butter ratio** | the price of cocoa butter expressed as a multiple of the terminal bean price rather than in dollars per tonne |
+| **powder ratio** | the price of cocoa powder or press cake expressed as a multiple of the terminal bean price |
+| **combined ratio** | the butter ratio plus the powder ratio, the single number a grinder watches because it states the processing margin as a multiple of the bean cost |
+| **grind** | the volume of beans processed into butter and powder, reported quarterly by region and the market's best read on actual chocolate demand |
+| **main crop** | the larger of West Africa's two harvests, running from October, as distinct from the smaller mid crop that follows from around April |
+| **farmgate price** | the price paid to the grower at the farm, which in Ivory Coast and Ghana is a guaranteed minimum fixed by a state body for a whole season rather than discovered daily |
+| **Coffee-Cocoa Council** | the Ivorian state body that forward-sells the crop in tranches before harvest and sets the season's guaranteed farmgate price from the average of those completed sales |
+| **performance risk** | the risk that a counterparty who has priced a contract correctly simply does not deliver against it, which replaces price risk when the physical becomes worth more than the paper |
+
+## Quiz
+
+**Q1.** A grinder presses one tonne of cocoa beans. Shell and loss take 20%, leaving 800 kg of nib, which presses into 400 kg of cocoa butter and 400 kg of cake that is milled into powder. Processing costs $600 per tonne of beans and is fixed in dollars. The December New York terminal is $5,762/t, the butter ratio is 2.10 and the powder ratio is 0.85. (i) What is the margin per tonne of beans? (ii) The bean price falls to $4,200/t and both ratios are unchanged — what is the margin now? (iii) What combined ratio would the grinder need at $4,200 beans to earn the same dollar margin as in (i)?
+
+**Q2.** December New York cocoa is $5,762/t and December London cocoa is £4,250/t, with sterling at 1.3276. Sterling now moves to 1.3450 and neither futures price changes. What is the New York minus London spread, in dollars per tonne?
+
+**Q3.** Episode 30 put one cent a bushel across a 30 Mt corn book at $11.8m a year. What is one cent a bushel worth across a 12 Mt corn book?
+
+**Q4.** Episode 28 put ten dollars a barrel of Brent at about 1¼ cents a pound on the Brazilian hydrous ethanol parity. Hydrous parity sits at 17.30 c/lb, Brent now falls $16 a barrel and the real is unchanged. Where does the parity go?
+
+**Q5. (conversion drill)** A ten-day forecast puts 85 mm of rain on the Ivorian cocoa belt. How many inches is that?
+
+## SOLUTIONS (spoilers)
+
+**A1.** The whole question is whether a cheaper bean helps a grinder. It does not.
+
+**(i) At $5,762/t.** Butter is 2.10 × 5,762 = $12,100.20/t, and you have 400 kg of it: $4,840.08. Powder is 0.85 × 5,762 = $4,897.70/t, and you have 400 kg: $1,959.08. Product revenue is $6,799.16 per tonne of beans.
+
+| | $/t of beans |
+|---|---|
+| Butter, 400 kg at $12,100.20/t | +4,840.08 |
+| Powder, 400 kg at $4,897.70/t | +1,959.08 |
+| Beans | −5,762.00 |
+| Processing | −600.00 |
+| **Margin** | **+437.16** |
+
+**(ii) At $4,200/t.** Nothing about the ratios changes, so revenue is 2.95 × 0.4 × 4,200 = $4,956.00. Costs are 4,200 + 600 = $4,800. The margin is **$156.00/t**. The bean price fell 27% and the margin fell 64%.
+
+Why: revenue scales with the bean price, and so does the bean cost — but processing does not. At a lower bean price the same $600 is a larger share of the trade. Formally, break-even sits where the combined ratio equals (1 + c/P) ÷ 0.4. At P = 5,762 and c = 600 that is 1.1041 ÷ 0.4 = **2.76**; at P = 4,200 it is 1.1429 ÷ 0.4 = **2.86**. The grinder's actual combined ratio of 2.95 is unchanged, so the cushion above break-even shrank from 0.19 to 0.09.
+
+**(iii)** To earn $437.16 at $4,200 beans, revenue must be 4,200 + 600 + 437.16 = $5,237.16. Revenue is combined ratio × 0.4 × 4,200 = combined ratio × 1,680. So the required combined ratio is 5,237.16 ÷ 1,680 = **3.12**.
+
+**The trap.** "Beans are cheaper, so the grinder is better off" is the instinct, and it is backwards. A grinder is long the ratio and short a fixed dollar conversion cost. Falling beans shrink the dollar value of the ratio spread while leaving the cost alone. This is the same structure as a crusher (episode 8), with one difference that matters: the crush is quoted in dollars per bushel, so the squeeze is visible on the screen. The cocoa combined ratio is quoted as a multiple, so a grinder watching a flat 2.95 can lose two thirds of his margin without the number he watches moving at all.
+
+**A2.** £4,250 at 1.3450 is $5,716.25. The spread is 5,762.00 − 5,716.25 = **$45.75/t**.
+
+It was $119.70 at 1.3276 (4,250 × 1.3276 = $5,642.30). So 1.74 cents of sterling took $73.95/t out of the spread — 4,250 × 0.0174 = 73.95 — with no cocoa trading. The sensitivity is simply the London price in pounds: every cent of cable is £42.50, so $42.50 a tonne of spread. A desk that marks this spread as a cocoa position is marking a currency position and filing it under the wrong risk.
+
+**A3.** 12,000,000 t × 39.368 = 472,416,000 bu. At one cent a bushel that is **$4,724,160** a year.
+
+The point of episode 30's arithmetic was that the edge is fixed per tonne, so the only lever is tonnes. A cent is worth $11.8m on 30 Mt and under $4.8m on 12 Mt — the same skill, the same people to pay for it, and a fraction of the return. That is the mechanical case for consolidation, with no theory of market power required.
+
+**A4.** $16 is 1.6 lots of ten dollars, so 1.6 × 1.25 = 2.00 c/lb. The parity falls from 17.30 to **15.30 c/lb**.
+
+Two cautions that episode 28 made explicit. First, the parity is the floor, not the price: with the ICE No. 11 screen at 18.60 the mill is already well above indifference, so pushing the floor down 2 cents changes nothing about this week's crystal decision. Second, the gearing is a switch and not a coefficient — the $16 would matter a great deal if sugar were trading at 17.50 and nothing at all at 25.00.
+
+**A5.** 85 ÷ 25.4 = **3.3 inches**.
+
+Fast method: 85 ÷ 100 = 0.85, then × 4 = 3.4 inches. Close enough to repeat out loud without pausing, which is the point. Rain forecasts for West Africa arrive in millimetres and for the US Midwest in inches, often in the same sentence.
+
+## The written edition
+
+### A number set by committee
+
+On 1 September the Ivorian agriculture ministry fixed the guaranteed farmgate price for the 2026/27 main crop at **1,200 CFA francs a kilo** — about $2.05/kg, so roughly $2,050/t. The year before it was 2,800 CFA/kg. That is a cut of 57.1% in the price paid to the people who grow roughly two fifths of the world's cocoa.
+
+The world price did not fall 57%. The December New York contract closed at $5,762/t on 6 October. The two numbers are not in conflict, because they are not measuring the same thing — and understanding why is most of what a desk needs to know about trading West African cocoa.
+
+### Two terminals, and a currency hiding inside the spread
+
+Start with the screens, because cocoa has two.
+
+The Intercontinental Exchange runs a cocoa contract in **New York**, quoted in US dollars per tonne, 10 tonnes to a lot, tick $1/t. It runs another in **London** — cocoa No. 7 — quoted in pounds sterling per tonne, also 10 tonnes to a lot, tick £1/t. Same bean, two currencies, two deliverable specifications. London grew up as the terminal for West African cocoa, which is most of the world's cocoa; New York takes a wider set of origins. Both survived because the physical market is genuinely split, and because a European grinder's costs are not denominated in dollars.
+
+Now put the two on the same axis. Monday's closes:
+
+| | Level | In dollars |
+|---|---|---|
+| ICE New York, Dec 26 | $5,762.00/t | 5,762.00 |
+| ICE London, Dec 26 | £4,250.00/t | 5,642.30 |
+| Sterling | 1.3276 | |
+| **New York premium** | | **+119.70** |
+
+$119.70 a tonne. Now move sterling one cent and leave both cocoa prices exactly where they are. One cent on £4,250 is **$42.50 a tonne** of spread. The whole New York premium is 2.8 cents of cable.
+
+Put it the other way round: that spread is 2.1% of the price of cocoa, and a 1% move in sterling is $56 — nearly half the spread. A desk quoting the New York–London spread without naming the exchange rate is quoting a cross-rate and calling it a cocoa view.
+
+You hear this in the way the spread actually gets worked:
+
+> **TRADER:** Where's December New York, London?
+> **BROKER:** Hundred and twenty over.
+> **TRADER:** Over at what cable?
+> **BROKER:** One thirty-two seventy-six.
+> **TRADER:** Then show me eighty over at one thirty-four and I'm a buyer.
+
+The trader refused the number until the rate was named, because "a hundred and twenty over" implies a different sterling price at every cable. A European grinder who buys beans against London and sells cocoa butter in dollars carries exactly that position, whether or not anyone wrote it down — which is a cross-hedge leak of the kind episode 22 catalogued, hiding inside a book that nets to flat.
+
+```chart
+{"type":"bar","unit":"$/t",
+ "title":"Three prices for one bean",
+ "caption":"The two terminals disagree by about two percent, and most of that gap is an exchange rate. The price paid to the Ivorian farmer is a different number entirely, because it was fixed from sales made a year earlier.",
+ "source":"ICE New York and ICE London December closes, 6 October 2026 (CocoaIntel); Ivorian 2026/27 guaranteed farmgate price, Ministry of Agriculture, 1 September 2026, converted at the rate given in the announcement",
+ "x":["ICE New York Dec","ICE London Dec","Ivorian farmgate"],
+ "series":[{"name":"$/t","values":[5762,5642,2050]}]}
+```
+
+### The state that sells the crop
+
+In Ivory Coast the crop is not sold by farmers to merchants at a discovered price. The **Coffee-Cocoa Council** sells it forward in tranches through the year *before* it is harvested. By June 2026 it had already forward-sold close to a million tonnes of the 2026/27 main crop. The guaranteed farmgate price is then set from the average of those completed sales.
+
+Not from today's screen. From sales that are already done.
+
+Three consequences, and each one changes how the origin is traded.
+
+**One: the farmgate is a moving average of last year's forward curve.** The farmer's price lags the terminal by a season, in both directions. It did not rise with the spike, and it is falling after it. Which means the supply response to high prices arrives roughly a year after the high prices — precisely when they are gone. The textbook cobweb, written into policy.
+
+**Two: price risk becomes performance risk.** When the terminal runs far above the level at which the crop was pre-sold, the beans become worth more than the contract that bought them. The exporter holding a cheap contract cannot get loaded, because whoever is sitting on the physical has every incentive not to deliver it at last year's price. The hedge is perfect, the paper is correct, and the beans do not arrive. This is the failure mode episode 23 described in the abstract, and an administered origin manufactures it structurally rather than occasionally.
+
+**Three: the differential is an allocation, not a price.** At an origin where the state fixes the farmgate and pre-sells the tonnage, the Ivorian differential is an administered number plus a queue. So the variable a desk watches is not a basis quote. It is contract allocation, and arrivals at the ports — deliveries from 1 to 13 September ran 26,000 t, down 45.8% year on year on a marketing year that starts 1 September. That is the number with information in it.
+
+### The grind, and the ratio that is not a constant
+
+A tonne of beans is not sold as beans. It is ground.
+
+Shell and loss take about a fifth, leaving 800 kg of nib. Press the nib and you get roughly 400 kg of **cocoa butter** and 400 kg of **press cake**, milled into cocoa powder. And here is the convention that makes cocoa feel foreign at first: butter and powder are not quoted in dollars. They are quoted as **ratios** — multiples of the terminal bean price. A butter ratio of 2.10 means butter is 2.10 times the screen. Add the butter ratio to the powder ratio and you have the **combined ratio**, which is the grinder's margin in one number.
+
+Work it at Monday's terminal, with a butter ratio of 2.10 and a powder ratio of 0.85 — a combined ratio of 2.95:
+
+```chart
+{"type":"waterfall","unit":"$/t of beans",
+ "title":"A grinder's margin, one tonne",
+ "caption":"Four hundred and thirty-seven dollars of margin sits between two numbers thirteen times its size, which is why a grinder hedges the ratio rather than the bean.",
+ "source":"Worked example, episode 31. Bean price is the ICE New York December close of 6 October 2026; ratios and the processing cost are the example's assumptions.",
+ "steps":[{"label":"Butter, 400 kg","value":4840,"kind":"base"},
+          {"label":"Powder, 400 kg","value":1959},
+          {"label":"Beans","value":-5762},
+          {"label":"Processing","value":-600},
+          {"label":"Margin","kind":"total"}]}
+```
+
+$437.16 a tonne, from a trade that turns over more than $12,500 of product and bean value. That ratio of margin to turnover is why grinders hedge the ratio and not the bean.
+
+Now the part most people get wrong.
+
+Everybody on a cocoa desk quotes a break-even combined ratio as though it were a fixed property of the industry — around 2.8. It is not a constant. The yield arithmetic alone sets a floor: 800 kg of product per tonne of beans means pure conversion breaks even at 1 ÷ 0.4 = **2.50**. What lifts break-even above 2.50 is the processing cost, and the processing cost is **fixed in dollars**.
+
+| Bean price | Processing as % of beans | Break-even combined ratio | Margin at a 2.95 ratio |
+|---|---|---|---|
+| $8,000/t | 7.5% | 2.69 | $840/t |
+| $5,762/t | 10.4% | 2.76 | $437/t |
+| $4,200/t | 14.3% | 2.86 | $156/t |
+| $3,000/t | 20.0% | 3.00 | −$60/t |
+
+Read the third column again. **Cheap beans require a better combined ratio to break even, not a worse one.** A collapsing bean price does not rescue the grind margin; it raises the bar. At $3,000 beans the industry's quoted "healthy" 2.95 is loss-making.
+
+That is visible in the volumes. Second-quarter grind was down 4.6% in Europe, up 7.7% in North America and up 25% in Asia. Capacity is not being rescued by cheaper beans. It is migrating to wherever the ratio clears, which is to say wherever conversion cost is low enough that the fixed dollar term stops dominating.
+
+### The thing to carry away
+
+Three mechanisms, one shape.
+
+The New York–London spread looks like a cocoa position and is mostly a currency position. The Ivorian farmgate looks like a price and is an average of last year's sales. The combined ratio looks like a margin and is a margin only at one bean price.
+
+In each case a number that appears to measure cocoa is actually measuring something else, and the error is the same error: taking a quoted convention for the thing it is quoted against. When somebody other than the market sets a price — an exchange rate, a state committee, an industry rule of thumb — the risk does not disappear. It changes form, and it turns up in a column nobody is watching.

package/ep31.script.txt

@@ -0,0 +1,102 @@
+The world's largest cocoa producer just cut the price it pays its farmers by fifty-seven percent. ||| 0.5
+The market did not fall fifty-seven percent. Nothing like it. ||| 0.5
+The cut came out of a committee, from sales already made a year ago, at prices nobody can see on a screen. ||| 0.7
+This is Soft Commodity Trading, episode thirty-one. Cocoa: two terminals, one ratio, and the state that sells the crop. ||| 0.8
+First, the tape. Thursday was quiet and slightly lower across Chicago. ||| 0.4
+December Chicago wheat settled at six eighty-three and a quarter, down three and a quarter. ||| 0.35
+December corn settled at five hundred and a quarter cents, down one and three quarters. November beans twelve eighty-seven and a half, down ten. ||| 0.35
+Take the week as a whole, because the week is the story. ||| 0.4
+Corn finished two and a half cents higher than last Friday. Beans up nine and a quarter. ||| 0.35
+Chicago wheat up a quarter of a cent, after a range of more than twenty-one. ||| 0.5
+Wheat rallied twenty-nine and a half cents in three sessions and gave back twenty-one in two. A full round trip, ending nowhere. ||| 0.6
+Only meal did real work, up ten dollars ten on the week while oil lost seventy points. ||| 0.6
+The October WASDE lands today at eleven Chicago time. ||| 0.4
+The trade average for the corn yield is one hundred seventy-seven point eight bushels an acre, against the U S D A's September number of one hundred seventy-eight and a half. ||| 0.4
+The poll range runs from one hundred seventy-three point two to one hundred eighty-two point one. ||| 0.5
+That range is the point. The average is a nine-tenths cut; the range is nine bushels wide. ||| 0.5
+On about eighty-eight and a half million harvested acres, nine bushels is close to eight hundred million bushels of production. ||| 0.5
+So the print will land inside the range and still surprise most of the desks reading it. ||| 0.6
+And there is a piece of this crop the report cannot carry. ||| 0.4
+The new crop has mould and sprouting in it in places, and a yield estimate measures bushels, not grade. ||| 0.5
+A bushel that fails on damage does not leave the balance sheet. It moves to a discount, and discounts live in the basis. ||| 0.7
+The geopolitics this week came through the dollar, not the corridor. ||| 0.5
+Record Treasury yields and a firmer dollar pulled wheat down on Wednesday with nothing happening to supply. ||| 0.4
+Here is the mechanism. Russian wheat is offered in dollars, around two hundred and four a tonne free on board. ||| 0.4
+When the dollar strengthens, that unchanged dollar offer converts into more roubles at the seller's bank. ||| 0.4
+His incentive to keep offering rises. The importer's local-currency cost rises at the same moment. ||| 0.5
+A strong dollar makes the Russian seller more willing and the buyer less able, simultaneously. ||| 0.5
+Which is why a closed corridor has stopped mattering to the flat price. The tonnes leave through northern ports and by rail. ||| 0.4
+IKAR puts September exports near two and a half million tonnes. SovEcon has trimmed the crop to eighty-seven and a half million. ||| 0.4
+Jordan, the Philippines, Thailand and Morocco all bought this week. The currency is doing the pricing. ||| 0.7
+Now cocoa, which has no WASDE at all. ||| 0.5
+No crop survey, no balance sheet released at a fixed minute. The most important number in cocoa is set by a state marketing board. ||| 0.6
+Start with the screens, because cocoa has two. ||| 0.5
+The Intercontinental Exchange runs a cocoa contract in New York, quoted in U S dollars a tonne, ten tonnes a lot. ||| 0.4
+It runs another in London, called cocoa number seven, quoted in pounds sterling a tonne, also ten tonnes a lot. ||| 0.5
+Same bean, two currencies, two deliverable specifications. ||| 0.4
+London grew up as the terminal for West African cocoa, which is most of the world's cocoa. New York takes a wider set of origins. ||| 0.5
+Both survived because the physical market is genuinely split, and because a European grinder's costs are not in dollars. ||| 0.6
+So what is the spread between them? It is not what it looks like. ||| 0.5
+On Monday's closes, December New York was five thousand seven hundred sixty-two dollars a tonne. ||| 0.4
+December London, four thousand two hundred fifty pounds, with sterling at one point three two seven six. ||| 0.4
+Converted, that London price is five thousand six hundred forty-two dollars and thirty cents. ||| 0.4
+So New York sits at a premium of one hundred nineteen dollars and seventy cents. ||| 0.6
+Now move sterling one cent and leave both cocoa prices exactly where they are. ||| 0.4
+One cent on four thousand two hundred fifty pounds is forty-two dollars fifty a tonne. ||| 0.5
+Less than three cents of sterling wipes out the whole New York premium, with no cocoa trading at all. ||| 0.7
+Put it the other way. That spread is two percent of the price of cocoa, and a one percent move in sterling is half of it. ||| 0.6
+So a desk quoting the spread without naming the exchange rate is quoting a cross-rate and calling it a cocoa view, and you hear it in how the spread gets worked. ||| 0.5
+TRADER: Where's December New York, London? ||| 0.25
+BROKER: Hundred and twenty over. ||| 0.25
+TRADER: Over at what cable? ||| 0.3
+BROKER: One thirty-two seventy-six. ||| 0.25
+TRADER: Then show me eighty over at one thirty-four and I'm a buyer. ||| 0.6
+He refused the spread until the broker named it, because a hundred and twenty over means a different pound price at every cable. ||| 0.5
+A European grinder buying beans against London and selling butter in dollars carries that same position, whether or not anybody wrote it down. ||| 0.7
+In Ivory Coast the crop is not sold by farmers to merchants at a discovered price. ||| 0.4
+A state body, the Coffee-Cocoa Council, sells the crop forward in tranches through the year before it is harvested. ||| 0.4
+By June it had already forward-sold close to a million tonnes of the twenty twenty-six, twenty twenty-seven main crop. ||| 0.5
+Then it sets a guaranteed farmgate price from the average of those completed sales. Not from today's screen. ||| 0.6
+On the first of September it set that price at one thousand two hundred C F A francs a kilo. ||| 0.4
+The year before it was two thousand eight hundred. That is the fifty-seven percent cut. ||| 0.6
+Twelve hundred C F A a kilo is about two dollars and five cents a kilo, so two thousand and fifty dollars a tonne. ||| 0.4
+Against a December terminal near five thousand seven hundred and sixty. ||| 0.6
+Three things follow, and all three change how you trade the origin. ||| 0.5
+First, that farmgate price is a moving average of last year's forward curve. ||| 0.4
+So the farmer's price lags the screen by a season, in both directions. It did not rise with the spike, and it is falling after it. ||| 0.5
+The production response to high prices arrives a year after the high prices, which is exactly when they are gone. ||| 0.7
+Second, and this is the one that costs money. ||| 0.4
+When the terminal runs far above the level at which the crop was pre-sold, the beans become worth more than the contract that bought them. ||| 0.4
+The exporter holding a cheap contract cannot get loaded, because whoever is sitting on the beans has every reason not to deliver them at last year's price. ||| 0.4
+Price risk has quietly become performance risk. Your hedge is perfect and your beans never arrive. ||| 0.6
+Third, the Ivorian differential is not a merchant's price. It is an administered number plus an allocation. ||| 0.4
+So the variable a desk watches at that origin is not a basis quote. It is contract allocation, and arrivals at the ports. ||| 0.5
+Arrivals to the thirteenth of September ran twenty-six thousand tonnes, down forty-six percent on the year. That is the number with information in it. ||| 0.7
+A tonne of beans is not sold as beans. It is ground. ||| 0.4
+Now cocoa's version of the crush. Shell and loss take about a fifth of a tonne of beans, leaving eight hundred kilos of nib. ||| 0.4
+Press the nib and you get roughly four hundred kilos of cocoa butter and four hundred kilos of cake, which is milled into powder. ||| 0.5
+And here is the convention that makes cocoa feel strange. Butter and powder are not quoted in dollars, but as a multiple of the terminal bean price. ||| 0.5
+Add the butter ratio to the powder ratio and you have the combined ratio. That one number is the grinder's margin gauge. ||| 0.6
+Work it. Beans at five thousand seven hundred sixty-two. Butter two point one zero, powder zero point eight five, a combined ratio of two point nine five. ||| 0.4
+Butter is twelve thousand one hundred a tonne, and you have four hundred kilos. Four thousand eight hundred and forty dollars. ||| 0.4
+Powder is four thousand eight hundred ninety-eight a tonne, four hundred kilos, nineteen hundred and fifty-nine dollars. ||| 0.4
+Six thousand seven hundred ninety-nine dollars of product, from five thousand seven hundred sixty-two dollars of beans. ||| 0.4
+Take six hundred dollars of processing, and the margin is four hundred and thirty-seven dollars a tonne. ||| 0.6
+Four hundred and thirty-seven, between two numbers thirteen times its size. That is why a grinder hedges the ratio and not the bean. ||| 0.7
+Now the part most people get wrong. ||| 0.5
+Everybody quotes a break-even combined ratio as though it were a constant. Around two point eight. ||| 0.5
+It is not a constant. You get eight hundred kilos of product from a tonne of beans, so pure conversion breaks even at two point five. ||| 0.5
+Processing cost is what lifts it above two point five, and processing cost is fixed in dollars. ||| 0.5
+Six hundred dollars on a five thousand seven hundred sixty dollar bean is ten point four percent, and break-even is two point seven six. ||| 0.5
+Put the same six hundred dollars on a three thousand dollar bean, and it is twenty percent. Break-even goes to three point zero zero. ||| 0.6
+Cheap beans require a better combined ratio to break even, not a worse one. ||| 0.6
+So a collapsing bean price does not rescue the grind margin. It raises the bar. ||| 0.5
+And it shows in the grind. European second-quarter grind down four point six percent. Asia up twenty-five. North America up seven point seven. ||| 0.5
+Capacity is not being rescued by cheap beans. It is migrating to where the ratio clears. ||| 0.7
+What to take away. ||| 0.5
+The New York minus London cocoa spread is mostly a sterling position. Never quote it without naming the exchange rate. ||| 0.5
+An administered farmgate price turns a price move into a performance problem. Where a state pre-sells the crop, watch allocation and arrivals, not basis. ||| 0.5
+The grinder's margin is a ratio against the terminal, and the ratio needed to break even rises as beans get cheaper. ||| 0.5
+Underneath all three: when somebody other than the market sets a price, the risk does not disappear. It changes form. ||| 0.7
+Monday, the rule that splits one commodity into two: the European deforestation regulation, and what mandatory traceability does to a fungible market. ||| 0.5
+Four questions in the notes, with full solutions. The hard one is the combined ratio at a different bean price. Work it before you look. ||| 0.8

package/feed.xml

@@ -18,6 +18,18 @@
       <title>Soft Commodity Trading</title>
       <link>https://storage.googleapis.com/podcast-audio-2647223968/index.html</link>
     </image>
+    <item>
+      <title>Ep 31 — Cocoa: Two Terminals, the Ratio, and the State That Sells the Crop</title>
+      <link>https://storage.googleapis.com/podcast-audio-2647223968/commodity-desk-daily/ep31.html</link>
+      <description><![CDATA[<p>Cocoa trades on two terminals in two currencies, and most of the spread between them is an exchange rate rather than cocoa. Ivory Coast has just cut its guaranteed farmgate price by 57% from sales made a year earlier, and a grinder's combined ratio needs to be higher, not lower, to break even as beans get cheaper.</p><p><a href="https://storage.googleapis.com/podcast-audio-2647223968/commodity-desk-daily/ep31.html">Read this episode, with the charts, the glossary and the quiz &rarr;</a></p>]]></description>
+      <itunes:summary>Cocoa trades on two terminals in two currencies, and most of the spread between them is an exchange rate rather than cocoa. Ivory Coast has just cut its guaranteed farmgate price by 57% from sales made a year earlier, and a grinder's combined ratio needs to be higher, not lower, to break even as beans get cheaper.
+
+Read this episode: https://storage.googleapis.com/podcast-audio-2647223968/commodity-desk-daily/ep31.html</itunes:summary>
+      <enclosure url="https://storage.googleapis.com/podcast-audio-2647223968/commodity-desk-daily/ep31.mp3" length="0" type="audio/mpeg"/>
+      <guid isPermaLink="false">https://storage.googleapis.com/podcast-audio-2647223968/commodity-desk-daily/ep31.mp3</guid>
+      <pubDate>Fri, 09 Oct 2026 04:40:00 GMT</pubDate>
+      <itunes:duration>0</itunes:duration>
+    </item>
     <item>
       <title>Ep 30 — The Houses, and the Trader's Checklist</title>
       <link>https://storage.googleapis.com/podcast-audio-2647223968/commodity-desk-daily/ep30.html</link>

package/glossary.md

@@ -49,7 +49,9 @@ Units, conventions and desk expressions, accumulated as the show introduces them
 - **bunkers** — the vessel's fuel, priced separately from the hire and carried by the owner on a voyage charter and by the charterer on a time charter  _(ep 10)_
 - **bushel** — volume measure standardized into weight, 60 lb for soybeans and wheat, 56 lb for corn  _(ep 1)_
 - **bushels per tonne** — about 36.7 for soybeans and wheat, 39.4 for corn  _(ep 1)_
+- **butter ratio** — the price of cocoa butter expressed as a multiple of the terminal bean price rather than in dollars per tonne  _(ep 31)_
 - **buyer's call** — a price-to-be-fixed contract in which the buyer holds the right to choose the moment of fixation  _(ep 13)_
+- **cable** — desk shorthand for the sterling-dollar exchange rate, the number that converts a London cocoa price into a New York one  _(ep 31)_
 - **calendar spread** — the price difference between two months of the same contract, traded as one instrument at one price  _(ep 3)_
 - **cancelling date** — the last day of the laycan, after which the counterparty may cancel  _(ep 4)_
 - **capacity utilisation** — the share of storage capacity actually occupied, the best leading indicator of what harvest basis is about to do  _(ep 11)_
@@ -68,8 +70,11 @@ Units, conventions and desk expressions, accumulated as the show introduces them
 - **CIF** — cost insurance and freight, CFR plus the seller buys the marine insurance the buyer would claim on  _(ep 4)_
 - **circle** — a string that returns to an earlier seller, after which the parties settle the price differences in cash and the physical never moves  _(ep 23)_
 - **citrus greening** — huanglongbing, the bacterial disease that permanently reduces an infected orange tree's yield and cannot be cured  _(ep 15)_
+- **cocoa No. 7** — the ICE London cocoa futures contract, quoted in pounds sterling per tonne in lots of 10 tonnes, historically the terminal for West African cocoa  _(ep 31)_
 - **Coffee C (KC)** — the ICE arabica futures contract, 37,500 lb quoted in US cents per pound with a 0.05 cent tick worth 18.75 dollars  _(ep 12)_
+- **Coffee-Cocoa Council** — the Ivorian state body that forward-sells the cocoa crop in tranches before harvest and sets the season's guaranteed farmgate price from the average of those completed sales  _(ep 31)_
 - **collar (fence)** — buying a put and selling a call against the same position so the price is bounded on both sides, the standard hedging structure around unpriced physical  _(ep 17)_
+- **combined ratio** — the butter ratio plus the powder ratio, the single number a grinder watches because it states the processing margin as a multiple of the bean cost, and whose break-even level rises as the bean price falls because conversion cost is fixed in dollars  _(ep 31)_
 - **Commercial (COT category)** — merchants, processors, elevators and end users who hold futures against physical grain, and who are normally short when the speculative crowd is long  _(ep 24)_
 - **commercial buyer** — a privately accountable buyer such as a crusher or feed mill whose bid is derived from the margin it can compute on its own products, and which stops buying when that margin closes  _(ep 25)_
 - **Commitments of Traders (COT)** — the weekly CFTC report that splits open interest in each futures market by category of holder, taken as of Tuesday's close and published the following Friday afternoon  _(ep 24)_
@@ -178,6 +183,7 @@ Units, conventions and desk expressions, accumulated as the show introduces them
 - **Grain Stocks** — the quarterly USDA survey of physical inventories, from which the feed and residual line is backed out  _(ep 7)_
 - **great grain robbery** — trade shorthand for the 1972 Soviet purchase of roughly a quarter of the US wheat crop, executed quietly through several houses at once, and the high-water mark of information as a merchant's edge  _(ep 30)_
 - **green coffee** — unroasted milled coffee beans, the form in which all internationally traded coffee moves  _(ep 13)_
+- **grind** — the volume of cocoa beans processed into butter and powder, reported quarterly by region and the market's best read on actual chocolate demand  _(ep 31)_
 - **gross processing margin** — the industry name for product value minus raw material cost, the crush stated as a margin  _(ep 8)_
 - **Handysize** — the smallest mainstream dry bulk class at roughly 10,000 to 40,000 dwt, geared and able to work berths larger ships cannot reach  _(ep 10)_
 - **hard red spring (HRS)** — the 13.5 percent plus Minneapolis wheat bought to lift the protein of a grist  _(ep 5)_
@@ -228,6 +234,7 @@ Units, conventions and desk expressions, accumulated as the show introduces them
 - **long ton** — 2,240 lb, the imperial weight unit the sugar No. 11 contract is still sized in at 50 long tons a lot  _(ep 14)_
 - **loss limit** — a cap expressed as a value-at-risk number or as a stress-scenario loss, breached when a book is too large for the balance sheet rather than when it is wrong  _(ep 22)_
 - **lot** — one futures contract, 5,000 bushels for Chicago grains, the unit desks count positions in  _(ep 1)_
+- **main crop** — the larger of West Africa's two cocoa harvests, running from October, as distinct from the smaller mid crop that follows from around April  _(ep 31)_
 - **managed money** — speculative funds reported as non-commercial in exchange positioning data, which trade direction rather than physical  _(ep 13)_
 - **margin financing** — the cost of funding variation margin paid out on a losing futures leg while the offsetting gain on the physical is still unrealised  _(ep 21)_
 - **mark to market** — repricing every open line at the day's settlement, physical included, so a book's value moves daily on grain nobody has yet agreed to buy  _(ep 21)_
@@ -247,6 +254,7 @@ Units, conventions and desk expressions, accumulated as the show introduces them
 - **net position** — physical long less physical short plus futures, for one month and one location, the only number that actually describes exposure  _(ep 21)_
 - **netback** — the value of a cargo at an upstream point, obtained by taking a downstream price and subtracting every cost in between, the standard way an export desk turns a destination price into an origin bid  _(ep 20)_
 - **new crop** — the marketing year about to begin, priced by the contract months that follow the coming harvest  _(ep 7)_
+- **nib** — the cocoa bean after the shell is removed, roughly 800 kg per tonne of beans, and the material that is actually pressed  _(ep 31)_
 - **Nino-3.4** — the box of equatorial Pacific between 5 degrees north and 5 degrees south and 170 to 120 degrees west whose sea-surface temperature anomaly is the standard measure of ENSO strength  _(ep 27)_
 - **No. 11** — the ICE raw cane sugar futures contract, 112,000 lb quoted in US cents per pound FOB at origin, and the world price of raw sugar  _(ep 14)_
 - **No. 5** — the ICE London white sugar futures contract, 50 tonnes quoted in US dollars per tonne delivered, and the world price of refined sugar  _(ep 14)_
@@ -279,6 +287,7 @@ Units, conventions and desk expressions, accumulated as the show introduces them
 - **pass-fail specification** — a contract term that cannot be met on average, such as contamination, infestation or an unapproved genetic event, where blending increases the affected tonnage instead of diluting it  _(ep 11)_
 - **percent of full carry** — a calendar spread expressed as a fraction of the interest and storage cost of holding the grain to the later month, the standard way a desk reads how badly a market wants storage  _(ep 16)_
 - **performance bond** — a guarantee posted by the winner of a tender against actually shipping to the contracted terms, usually a low single-digit percentage of contract value and carrying a real financing cost  _(ep 20)_
+- **performance risk** — the risk that a counterparty who has priced a contract correctly simply does not deliver against it, which replaces price risk when the physical becomes worth more than the paper  _(ep 31)_
 - **physical (cash)** — real cargoes under contract with specs and load windows, as opposed to paper  _(ep 2)_
 - **plant crush** — what a physical plant actually earns, the board crush adjusted for bean, meal and oil basis and net of conversion cost  _(ep 8)_
 - **planted acres** — area sown, the number that moves on farmer decisions and USDA area surveys  _(ep 6)_
@@ -289,7 +298,9 @@ Units, conventions and desk expressions, accumulated as the show introduces them
 - **position limit** — a cap on exposure in tonnes or lots, set by commodity and by delivery month  _(ep 22)_
 - **position sheet** — a desk's record of net exposure by futures month and by location, kept in lots with long positive and short negative, in which physical and paper appear as the same row  _(ep 21)_
 - **posted bid** — the price an elevator displays to growers for immediate delivery, quoted as a differential to a named futures month and used to manage the delivery queue as much as to set a price  _(ep 19)_
+- **powder ratio** — the price of cocoa powder or press cake expressed as a multiple of the terminal bean price  _(ep 31)_
 - **pre-advice** — the notice a seller owes the buyer naming the performing vessel and the expected dates, a stated number of days before arrival  _(ep 29)_
+- **press cake** — the solid residue left after cocoa butter is pressed out of the nib, milled into cocoa powder and typically retaining 10 to 12 percent fat  _(ep 31)_
 - **price assessment** — a published daily price built by surveying brokers and exporters, used where no futures contract exists  _(ep 5)_
 - **price-later deadline** — the date by which an unpriced farmer contract must be fixed, after which the buyer prices it at the market  _(ep 19)_
 - **price-to-be-fixed (PTBF)** — a physical contract where quantity, quality, shipment and differential are agreed now and the futures price is set later  _(ep 13)_
@@ -361,6 +372,7 @@ Units, conventions and desk expressions, accumulated as the show introduces them
 - **tender** — a formal published invitation to offer in which an importer states a quantity, specification, delivery period and terms, and invites sellers to submit sealed price offers by a stated deadline  _(ep 20)_
 - **tender validity** — the period after the bid deadline during which a submitted offer stays firm and the buyer may accept it, usually a few hours, during which the seller carries the market risk and the buyer holds the choice  _(ep 20)_
 - **terminal elevator** — large storage at a port, river or rail hub whose business is blending, load-out speed and access rather than farm origination  _(ep 11)_
+- **terminal market** — the futures exchange whose quoted price serves as the pricing reference for the physical trade, against which origin differentials and processing ratios are quoted  _(ep 31)_
 - **test weight** — the density measure telling a miller how much flour comes out of a tonne  _(ep 5)_
 - **thin market** — a market in which the price obtainable depends materially on the size being traded, whatever a single lot is worth in notional terms  _(ep 15)_
 - **throughput** — the volume moved through a facility in a period, the number that actually pays for a fixed asset because capacity earns nothing standing still  _(ep 11)_

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@sdelsad/commodity-desk-daily",
-  "version": "1.0.79",
-  "description": "Soft Commodity Trading - Ep 30: The Houses, and the Trader's Checklist",
+  "version": "1.0.80",
+  "description": "Soft Commodity Trading - Ep 31: Cocoa: Two Terminals, the Ratio, and the State That Sells the Crop",
   "license": "CC-BY-4.0",
   "keywords": [
     "podcast",

package/ep30.md

@@ -1,242 +0,0 @@
-# The Houses, and the Trader's Checklist
-
-## Market pulse
-
-**The grain market bounced on Tuesday, and wheat led it for a third straight session — a WASDE two days out with a nine-bushel yield range, and a Black Sea in which the tonnes are not disappearing so much as changing flag.**
-
-| Contract | Month | Settle | Change |
-|---|---|---|---|
-| CBOT corn | Dec | 508 c/bu | +10¾ |
-| CBOT soybeans | Nov | 1303 c/bu | +22¼ |
-| CBOT wheat | Dec | 704¼ c/bu | +12 |
-| CBOT soybean meal | Dec | $354.80/short ton | +7.70 |
-| CBOT soybean oil | Dec | 69.91 c/lb | +56 pts |
-
-Read it against Friday rather than against Monday. Chicago wheat is **21¼ c** higher than its 2 October close, corn **10¼ c** higher, soybeans **24¾ c** higher. Corn had closed lower four weeks running; this is the first session that interrupted it.
-
-Two forces are behind the bounce. The first is the **October WASDE on Thursday**, into which the trade's corn yield estimates run from 173 to 182 bu/acre. That is a nine-bushel spread on the national yield, and a nine-bushel spread is not a forecast — it is an admission that nobody knows. Flood-related quality problems, mould and sprouting, are being discounted in the cash market at 3–5% on some soybean lots but will not appear in an official yield number at all. A short position into a distribution that wide is an expensive thing to hold for two days.
-
-The second is harvest pace. Corn was 23% harvested against a 27% five-year average and soybeans 25% against 33%, with Iowa the laggard and more rain forecast.
-
-**The geopolitical read — the export-capacity story stops being a forecast and starts being a tonnage.** Episode 28 followed a Russian export forecast being cut on expected loading outages, and episode 29 followed a refused truce. September's actual numbers now exist. Ukrainian grain exports fell **36.6% year on year** after renewed strikes on Odesa port infrastructure, while Russia shipped a record month, its Baltic ports alone clearing more than 1 Mt. That combination is the reason the flat price is only now responding: the world's tonnes were largely replaced, just from a different berth under a different flag.
-
-Where the cost actually landed is visible in the week's tender. Saudi Arabia bought **683,000 t** of wheat — 148,000 t more than it asked for — in a range from $334.60 to $348.00/t C&F, averaging about **$339/t**, roughly $70/t above what it paid in July, for November–December shipment in eleven cargoes. The importer pays for the rerouting before the exchange does. And the firms that filled it — Cargill, Bunge, Olam, CHS, Soufflet, Aston — are today's subject.
-
-```chart
-{"type":"line","unit":"indexed, 1 October = 100","mode":"index","title":"Wheat led the bounce, corn followed",
- "caption":"Chicago wheat has gained ground in every session since 1 October while corn spent three of them going nowhere. When the lead comes from wheat rather than corn, the story is supply geography rather than the US balance sheet.",
- "source":"CBOT settlements, 1 to 6 October 2026",
- "x":["1 Oct","2 Oct","5 Oct","6 Oct"],
- "series":[{"name":"Dec corn","values":[500.50,497.75,497.25,508.00]},
-           {"name":"Dec Chicago wheat","values":[674.75,683.00,692.25,704.25]}]}
-```
-
-## Key takeaways
-
-- The ABCD houses were built as arbitrageurs of information. Satellites and a simultaneous USDA release destroyed that edge, and the margin migrated from knowing to owning.
-- An elevator, a crush plant and a berth are each a right without an obligation. That option pays whether or not the owner's view is any good, which is why the asset-heavy model won.
-- At roughly $435 of revenue per tonne and a half-percent trading margin, the whole prize on 230 Mt is about $500m. The edge is fixed per tonne, so the only lever that moves it is tonnes.
-- One cent a bushel across a 30 Mt corn book is $11.8m a year. The same cent across 5 Mt is under $2m, which does not pay for the people who find it. That is consolidation, with no moralising required.
-- A good desk asks about funding and exit before it asks about direction. The natural first question — where is the price going — is not in the first five.
-- A hedge nets on paper and does not net in cash. Variation margin goes out daily against a physical gain that is realised on payment, and the cash out can be several times the margin on the trade.
-- A merchant's credit line is the real position limit.
-- Recovering a drawdown costs more than the drawdown. Down 25% needs +33%; down 50% needs +100%. Survival is a constraint on the optimisation, not a style of trading.
-
-## Vocabulary
-
-| Term | What it means |
-|---|---|
-| **grain major** | a house that both originates crop from farmers and merchandises it to end users at scale, owning the elevators, terminals, vessels and plants in between |
-| **ABCD-plus** | the present landscape, in which the four historic majors are joined by COFCO, by Viterra inside Bunge, and by Olam, Wilmar and the large farmer co-operatives |
-| **asset optionality** | the value an elevator, plant or berth carries as a right without an obligation — store or ship, crush or idle, load your cargo or a third party's — which accrues whether or not the owner's market view turns out to be correct |
-| **tolling** | processing a third party's commodity for a fee without ever owning it, converting a volatile trading margin into a capacity rent |
-| **merchandising margin** | the gross per-tonne earned between the origin purchase and the destination sale, before the cost of the assets and the working capital used to earn it |
-| **trading mandate** | the written permission defining what a desk may trade — which commodities, origins, instruments, months and size — and the first document a proposed trade is checked against |
-| **drawdown** | the peak-to-trough fall in the value of a book or a firm, whose recovery always requires a larger percentage gain than the percentage lost |
-| **great grain robbery** | trade shorthand for the 1972 Soviet purchase of roughly a quarter of the US wheat crop, executed quietly through several houses at once, and the high-water mark of information as a merchant's edge |
-
-## Quiz
-
-**Q1.** A house buys 60,000 t of US Gulf corn FOB at December futures plus 72 c/bu and has already sold it CFR Alexandria at December plus 158 c/bu. Freight is $23.00/t; insurance, finance and agency together are $2.20/t. December corn is 508.00 when both legs are priced, and the house hedges the long physical by selling December futures. Between pricing and shipment the board rallies 90 c to 598.00. (i) What is the planned margin on the cargo in dollars? (ii) How much variation margin does the house pay out on the hedge over that rally, and where does the offsetting gain sit? (iii) The treasury line has $8m of headroom. Does the trade survive?
-
-**Q2.** A house moves 30 Mt of corn a year. A new execution process saves, on average, a quarter of a cent per bushel on the basis at which cargoes are bought. What is that worth a year, in dollars?
-
-**Q3.** Episode 29 worked a cargo on a planned margin of $8.00/t. Take the same planned margin on 25,000 t, and a single demurrage claim of $45,000 landing at discharge. By what percentage does the realised margin fall?
-
-**Q4.** Episode 27 showed that a weakening Brazilian real lifts the farm-gate price in reais with nothing happening in Chicago. The real now strengthens instead, from 5.40 to 5.10 to the dollar. In which direction does that push Brazilian farmer selling?
-
-**Q5. (conversion drill)** Take Matif December milling wheat offered at €236.00 a tonne and a euro worth 1.1740 dollars. What is that offer in US cents per bushel?
-
-## SOLUTIONS (spoilers)
-
-**A1.** Work the planned margin first, then the cash, and keep the two apart — that separation is the whole question.
-
-*(i) The planned margin.* The trade is a basis trade: the flat price cancels, and what is left is the differential less the costs of moving the cargo.
-
-| Line | Value |
-|---|---|
-| Sale differential | December +158 c/bu |
-| Purchase differential | December +72 c/bu |
-| Gross basis margin | 86 c/bu |
-| Gross basis margin per tonne | 0.86 × 39.368 = $33.86/t |
-| Freight | −$23.00/t |
-| Insurance, finance, agency | −$2.20/t |
-| Net margin | $8.66/t |
-| On 60,000 t | **$519,389** |
-
-A tonne of corn is 39.368 bushels, so a differential quoted in cents a bushel becomes dollars a tonne by multiplying, not dividing. Exactly: $8.65648/t × 60,000 t = $519,388.80.
-
-*(ii) The variation margin.* The hedge is sized off the bushels, not the tonnes. 60,000 t × 39.368 = 2,362,080 bu, which at 5,000 bu to a Chicago lot is 472.4 lots — so the house sells **472 lots**, leaving 2,080 bu unhedged as quantity risk (episode 21).
-
-A 90 c rally against a short of 472 lots costs:
-
-> 472 lots × 5,000 bu × $0.90 = **$2,124,000**
-
-That is paid in cash, through the clearing house, on the day the board moves — and again the next day if it keeps moving. The offsetting gain is real but it is *unrealised*: the physical corn in the house's hands is worth 90 c/bu more, and that value is converted into cash only when the vessel loads, the documents are presented and the buyer pays. The residual 2,080 bu gains $1,872, which is noise.
-
-*(iii) Does it survive?* Yes — $2.124m against $8m of headroom leaves $5.876m. But look at what the number says. The house pays out **$2.124m of cash in order to earn $519,389**. The cash is **4.1 times the entire margin on the trade**, it leaves first, and it comes back last.
-
-**The trap** is the word *netting*. The position nets — the house is flat price-neutral and the P&L is unchanged. The *cash* does not net, because the clearing house settles daily and the buyer in Alexandria does not. A correct, fully hedged, profitable trade can therefore exhaust a credit line and force the hedge to be lifted at exactly the wrong moment, which converts a basis trade into an outright short. This is the mechanism behind every famous hedging failure, and it is why a merchant's credit line, not its risk policy, is the real position limit. Episode 21 priced the carrying cost of this; the point here is the size of the cash, not its cost.
-
-**A2.** 30 Mt is 30,000,000 t × 39.368 = 1,181,040,000 bu. A quarter of a cent is $0.0025, so:
-
-> 1,181,040,000 bu × $0.0025 = **$2,952,600 a year**
-
-Roughly $3m from a quarter of a cent — and no view, no position and no risk taken. That is the shape of the whole industry: the edge is tiny, fixed per tonne and unglamorous, which means the only variable that matters is how many tonnes you run it across. Run the same process over a 5 Mt book and it earns $492,100, which will not cover the team that built it.
-
-**A3.** Planned margin is $8.00/t × 25,000 t = $200,000. Demurrage of $45,000 is a cost of execution, so the realised margin is $155,000.
-
-> $45,000 ÷ $200,000 = **22.5%**
-
-Nearly a quarter of the cargo's earnings, removed by a laytime clock. The point from episode 29 is not that demurrage is expensive in absolute terms — $45,000 is three days on a modest Panamax rate — but that it is large *relative to a merchant margin of half a percent of cargo value*. To recover $45,000 by being right about the market, the trader would need the basis to move almost 2 c/bu in his favour on the whole parcel. The laytime clause was agreed before the price was, by someone who was not the trader.
-
-**A4.** It slows it down — and can stop it.
-
-The Brazilian farmer sells a dollar-denominated commodity and lives on reais. His farm-gate price is the port value in dollars, converted at the spot rate, less inland freight and elevation. Strengthen the real from 5.40 to 5.10 and the dollar leg buys fewer reais, so the price he sees falls even with Chicago frozen.
-
-Take a port value of $450/t. Grain and oilseeds are quoted to the Brazilian farmer in 60-kg *sacas*, 16⅔ to the tonne:
-
-| | At 5.40 | At 5.10 |
-|---|---|---|
-| Port value, R$/t | 2,430 | 2,295 |
-| Per 60-kg saca | R$145.80 | R$137.70 |
-
-That is R$8.10 a bag, or about 5.6%, taken out of his price by the currency alone. To produce the same fall from the board would take about 68 c/bu of Chicago. A producer who was a willing seller at R$145 is not a willing seller at R$137, so pricing stops, the origin basis has to firm to pull grain out of the farm, and the exporter who assumed a supply curve discovers he has a currency position. The relationship is the single most useful one in agricultural trading, and it runs the other way too: a weakening real is what unlocks Brazilian selling and collapses the basis.
-
-**A5.** Convert the currency first, then the unit.
-
-| Step | Arithmetic | Result |
-|---|---|---|
-| Euro to dollar | 236.00 × 1.1740 | $277.064 per tonne |
-| Tonne to bushel | 277.064 divided by 36.744 bu per tonne | $7.5404 per bushel |
-| To cents | × 100 | **754.0 c/bu** |
-
-Wheat and soybeans both run 36.744 bushels to the tonne, corn 39.368, because the test weight differs — 60 lb a bushel for wheat and beans, 56 for corn. Note the direction: converting a *quantity* of tonnes into bushels is a multiplication, but converting a *price* per tonne into a price per bushel is a division, because you are spreading the same money over more units.
-
-The payoff: Dec Chicago settled 704¼. The Matif offer is therefore about **50 c/bu over Chicago**, which is the transatlantic wheat spread expressed in the only unit in which the two can be compared at all.
-
-## The written edition
-
-### The merchants were an information business
-
-The four houses the trade calls the **ABCD** are older than the market structure they operate in. Bunge was founded in Amsterdam in 1818, Louis Dreyfus in Alsace in 1851, Cargill on a grain flat in Iowa in 1865, Archer Daniels Midland in 1902. All four predate the regulatory architecture of the futures exchanges they now hedge on.
-
-They were built to arbitrage **space and information**. A merchant who learned that the Odessa crop was short before Chicago learned it owned the trade outright. The high-water mark of that model was 1972, when Soviet buyers quietly bought roughly a quarter of the American wheat crop through several houses at once and the market discovered it afterwards — the *great grain robbery*, which is remembered as a scandal but was really a demonstration of what a pure information edge was worth.
-
-That edge is gone. Everybody has the satellite. Everybody receives the USDA release in the same second. Private crop services publish estimates that are frequently better than the official ones, and they publish them to subscribers rather than to one buyer.
-
-### Where the margin went
-
-It went into the assets, and the reason is not sentimental.
-
-**An elevator is an option.** So is a crush plant, and so is a berth. Each is a right without a matching obligation:
-
-| Asset | The right it confers |
-|---|---|
-| Terminal elevator | hold the grain, or ship it, whichever the spread pays |
-| Crush plant | run beans, or idle, whichever the board crush pays |
-| Port berth | load your own cargo, or a third party's for a fee |
-| Vessel on time charter | carry your cargo, or relet the ship |
-
-A desk that rents capacity has to pay to exercise. A desk that owns it has already paid, and holds the choice. That is **asset optionality**, and it is the property that makes an asset-heavy book earn money in years when the trading view is worthless. It is also the reason the same firms keep buying each other.
-
-Bunge closed its merger with Viterra in July 2025: roughly $8bn of equity and cash, with close to $10bn of Viterra debt assumed on top. The combined firm turns over more than $100bn, markets more than 230 Mt a year, and operates 56 crush plants, 26 port terminals and more than 350 grain storage facilities across more than 40 countries.
-
-### The arithmetic of a cent
-
-Those figures look like a story about size. They are a story about **thinness**.
-
-| Line | Arithmetic | Result |
-|---|---|---|
-| Revenue per tonne | $100bn across 230 Mt | ≈ $435/t |
-| Trading margin at half a percent | 0.5% of $435 | $2.17/t |
-| On the whole book | $2.17 × 230 Mt | ≈ $500m |
-
-Half a percent is the figure episode 29 worked on a single cargo. It holds at the firm level too, and it explains everything about the industry's shape: the margin is **fixed per tonne and does not improve with cleverness**, so the only variable available is tonnes.
-
-Say the same thing in the unit a trader actually feels. Take a 30 Mt corn book:
-
-> 30,000,000 t × 39.368 = 1,181,040,000 bu
-> One cent a bushel = **$11,810,400 a year**
-
-One cent. Not a forecast, not a position — a cent of execution. A cent better on the basis at which cargoes are bought, a day of demurrage avoided, a roll worked rather than hit. And the same cent on a 5 Mt book is $1,968,400, which will not fund the analysts, the traffic desk and the systems that produce it.
-
-```chart
-{"type":"bar","unit":"$ million a year","title":"What one cent a bushel is worth",
- "caption":"The same execution improvement earns six times as much on a 30 Mt book as on a 5 Mt one, and the cost of producing it barely changes. This is why the industry consolidates, and no theory of market power is needed to explain it.",
- "source":"Worked example, episode 30, at 39.368 bu per tonne of corn",
- "x":["5 Mt book","15 Mt book","30 Mt book"],
- "series":[{"name":"Value of 1 c/bu","values":[1.97,5.91,11.81]}]}
-```
-
-### ABCD-plus, and what traceability does to it
-
-Two forces have redrawn the map since the acronym was coined.
-
-The first is **COFCO**. China is the largest importer of soybeans on earth, and it chose not to remain a price-taker: COFCO bought Nidera and Noble Agri and assembled a merchant with origination assets in Brazil and Argentina. A buyer that owns the elevator at the origin no longer has to accept whatever basis the houses show it. The acronym became **ABCD-plus**, which now also takes in Olam, Wilmar and the large farmer co-operatives.
-
-The second is **traceability**. From December 2026 the EU deforestation regulation requires documented proof of the land a shipment came from. That turns one commodity into two commodities with two differentials — compliant and non-compliant — and the capability to prove provenance is an infrastructure problem, not a trading one. It requires segregated storage, audited chains of custody and plot-level data from the farm.
-
-Both forces push the same way: toward firms that own the chain, away from firms that merely have a view.
-
-### The checklist, in order
-
-A desk is offered something. The order in which it thinks about the offer is close to the reverse of the order that feels natural. *Where is the price going* is not in the first five questions.
-
-1. **What is the trade, in one sentence, and which risk am I taking?** Flat price, basis, spread, freight, quality, credit, currency. If you cannot name one, you are carrying all seven.
-2. **Who is on the other side, and why is he doing this?** A counterparty's motive is information. If you cannot explain it, the edge may be his.
-3. **What is the downside in dollars, under a scenario that happened?** A named day, not a statistic (episode 22).
-4. **Can I fund it?** Below.
-5. **Can I get out?** Exit is priced at entry, in the liquidity of the hedge month and the existence of a washout counterparty — never on the day you need it.
-6. **Is it inside the mandate?** Size, origin, counterparty, month.
-7. **What does it cost to be right?** Carry, storage, demurrage, the roll. A view that is right six months late pays for all six months.
-
-### Question four, worked
-
-Question four is where correct trades die. The arithmetic is in A1, and the summary is this:
-
-| | Amount |
-|---|---|
-| Margin earned on the cargo | $519,389 |
-| Cash paid to the clearing house on a 90 c rally | $2,124,000 |
-| Ratio | 4.1× |
-| When the cash leaves | daily, from the day the board moves |
-| When the margin arrives | on payment, after documents |
-
-The hedge is correct. The position is flat price-neutral. The P&L is unaffected. And the firm can still fail, because **variation margin settles daily and a buyer in Alexandria does not**.
-
-### What a decade costs
-
-What separates a trader who lasts ten years from one who has a great year?
-
-The great year comes from being large at the right moment. There is nothing mysterious in it and no way to repeat it on demand. The decade comes from the opposite tail: from never having the year that ends the firm.
-
-```chart
-{"type":"bar","unit":"% gain required to return to the prior peak","title":"Recovering a drawdown costs more than the drawdown",
- "caption":"A 25% loss needs a 33% gain to undo it and a 50% loss needs 100%. The asymmetry is why survival is treated as a constraint on the optimisation rather than as a conservative style of trading.",
- "source":"Arithmetic of drawdown recovery",
- "x":["−10%","−25%","−40%","−50%"],
- "series":[{"name":"Gain needed to get level","values":[11.1,33.3,66.7,100.0]}]}
-```
-
-A trader who maximises expected profit without that constraint is maximising a quantity whose variance removes him from the sample. And the firms that have been in this business for a hundred and fifty years are, almost without exception, the ones that treated the constraint as the strategy.

package/ep30.script.txt

@@ -1,88 +0,0 @@
-Two dollars a tonne. ||| 0.5
-That is roughly what a grain merchant keeps on a cargo worth four hundred and thirty-five dollars a tonne. Half of one percent. ||| 0.5
-That is not a bad year. That is the business model, and almost everything about how these firms are built follows from it. ||| 0.6
-This is Soft Commodity Trading, episode thirty. The houses, and the trader's checklist. ||| 0.8
-First, the tape. ||| 0.4
-Tuesday brought the bounce the grain market had been refusing for a month. December corn settled five dollars and eight cents, up ten and three quarter cents. ||| 0.4
-November soybeans thirteen dollars and three cents, up twenty two and a quarter. December Chicago wheat seven dollars four and a quarter, up twelve cents. ||| 0.5
-Wheat led, and wheat has led for three sessions running. Against Friday's close it is up twenty one and a quarter cents, while corn is up ten and a quarter. ||| 0.5
-Two things are pushing. The W A S D E prints Thursday, and the trade's corn yield guesses run from one hundred seventy three to one hundred eighty two bushels an acre. ||| 0.4
-That is a nine bushel spread on a crop this size. Nobody wants to be short into a number that wide. ||| 0.6
-The second thing is the Black Sea, and this one is concrete. ||| 0.4
-Russia struck Odesa over the weekend. Ukraine's September grain exports came in down thirty six point six percent on the year. ||| 0.4
-Russia shipped a record month, its Baltic ports alone handling more than a million tonnes. ||| 0.5
-Here is the transmission, and it is not the one people expect. The tonnes are not missing from the world. They are missing from one origin and arriving from another. ||| 0.5
-So the flat price barely moves, and the cost lands on the importer instead. Saudi Arabia bought six hundred and eighty three thousand tonnes this week, at about three hundred and thirty nine dollars a tonne cost and freight. ||| 0.4
-That is some seventy dollars a tonne above what it paid in July. The buyer pays for the rerouting. ||| 0.6
-Now look at who filled that tender. Cargill. Bunge. Olam. C H S. Soufflet. Aston. ||| 0.5
-Six houses, one buyer, eleven cargoes. That is today's subject. ||| 0.7
-Start with the dates, because the dates explain the shape. ||| 0.4
-Bunge was founded in eighteen eighteen, in Amsterdam. Louis Dreyfus in eighteen fifty one, in Alsace. Cargill in eighteen sixty five, on a grain flat in Iowa. Archer Daniels Midland in nineteen oh two. ||| 0.5
-Four firms, all older than the exchanges they hedge on. The trade calls them the A B C D. ||| 0.6
-They were born as arbitrageurs of information. A merchant who knew the Odessa crop was short before Chicago knew it owned the trade. ||| 0.5
-That advantage peaked in nineteen seventy two, when Soviet buyers quietly bought roughly a quarter of the American wheat crop through several houses at once, and the market found out afterwards. ||| 0.6
-That trade cannot happen now. Everybody has the satellite. Everybody gets the U S D A print in the same second. ||| 0.5
-So where did the margin go? ||| 0.4
-It went into the assets. Not because owning elevators is romantic, but because an elevator is an option, and an option pays whether or not your view is any good. ||| 0.6
-Take that slowly. A terminal elevator can hold grain or ship it. A crush plant can run beans or sit idle. A berth can load your cargo or somebody else's, for a fee. ||| 0.5
-Each of those is a right without an obligation. The desk that rents capacity has to pay to exercise it. The desk that owns it already has. ||| 0.6
-That is also the answer to the question people ask about the mergers. ||| 0.4
-In July twenty twenty five, Bunge closed its merger with Viterra. Call it eight billion dollars of equity and cash, with nearly ten billion of assumed debt on top. ||| 0.5
-The combined firm turns over more than a hundred billion dollars, moves more than two hundred and thirty million tonnes a year, and runs fifty six crush plants and twenty six port terminals. ||| 0.6
-Now do the arithmetic that makes that make sense. ||| 0.5
-A hundred billion of revenue on two hundred and thirty million tonnes is about four hundred and thirty five dollars a tonne. Half a percent of that is two dollars and seventeen cents. ||| 0.5
-Two dollars seventeen a tonne, across two hundred and thirty million tonnes, is around five hundred million dollars. ||| 0.6
-That is the whole prize. Which is why the only lever that really moves it is tonnes. ||| 0.5
-Here is the same point in the unit a trader actually feels. ||| 0.4
-Say a house moves thirty million tonnes of corn in a year. A tonne of corn is thirty nine point three six eight bushels, so that is one point one eight billion bushels. ||| 0.5
-One cent a bushel on that book is eleven point eight million dollars. ||| 0.6
-One cent. Not a view, not a forecast. A cent of execution. A cent better on the basis, a day of demurrage avoided, a roll done properly. ||| 0.5
-At five million tonnes the same cent is under two million dollars, which does not pay for the people it takes to find it. ||| 0.5
-That is consolidation explained without any moralising. The edge is small and fixed per tonne, so the only way to make it a business is to have a great many tonnes. ||| 0.7
-Two things have redrawn the map since. ||| 0.4
-The first is that the A B C D became A B C D plus. Cofco built a merchant out of China's import need, buying Nidera and Noble Agri, so the largest buyer of soybeans on earth now owns origination in Brazil and Argentina. ||| 0.6
-The second is traceability. From December twenty twenty six, the European deforestation rules require proof of where a shipment's land was and what stood on it. ||| 0.5
-That turns one commodity into two, with two differentials, and proving provenance is an asset business rather than a trading one. ||| 0.6
-Both changes push the same way. Toward firms that own the chain, and away from firms that only have a view. ||| 0.7
-Which brings us to the capstone, and the capstone is a list of questions. ||| 0.5
-A desk gets offered something. There is an order in which a good desk thinks about it, and the order is close to the opposite of the one that feels natural. ||| 0.5
-The natural first question is: where is the price going? On a desk that survives, that question is not in the first five. ||| 0.6
-First. What is the trade, in one sentence, and which risk am I actually taking? Flat price, basis, spread, freight, quality, credit, currency. ||| 0.4
-If you cannot name one, you are carrying all seven. ||| 0.6
-Second. Who is on the other side, and why is he doing this? If you cannot explain his reason, the edge may be his and not yours. ||| 0.6
-Third. What is the downside in dollars, under a scenario that actually happened once? Not a statistic. A named day. ||| 0.6
-Fourth. Can I fund it? ||| 0.5
-Fifth. Can I get out? Exit is priced at entry, not on the day you need it. ||| 0.5
-Sixth. Is it inside the mandate? Size, origin, counterparty, month. ||| 0.5
-And seventh. What does it cost to be right? Carry, storage, demurrage, the roll. A view that is right six months late pays for every one of those months. ||| 0.7
-Question four is where correct trades actually die, and the conversation that kills them is short. ||| 0.5
-TRADER: Sixty thousand corn, Gulf, December plus seventy two. Sold already, Alexandria, plus one fifty eight. ||| 0.25
-RISK: Margin? ||| 0.25
-TRADER: Eight sixty six a tonne. Five nineteen on the cargo. ||| 0.25
-RISK: And if the board goes up ninety cents before you ship? ||| 0.25
-TRADER: I'm hedged. It nets. ||| 0.25
-RISK: It nets on paper. It does not net in cash. ||| 0.6
-The risk officer is right, and this is the most common way a correct trade turns into a dead firm. ||| 0.5
-Sixty thousand tonnes is two million three hundred and sixty two thousand bushels. At five thousand bushels to a Chicago lot, that is four hundred and seventy two lots short. ||| 0.5
-The board rallies ninety cents. Four hundred and seventy two lots, five thousand bushels each, ninety cents a bushel. Two point one two million dollars. ||| 0.5
-Paid in cash. Today. Tomorrow. Every day the board goes up. ||| 0.6
-The cargo is worth exactly that much more. But it is worth more on paper, and it stays on paper until the vessel sails, the documents are presented, and the buyer pays. ||| 0.5
-So the house pays out two point one two million dollars of real cash in order to earn five hundred and nineteen thousand. ||| 0.5
-The hedge is right. The trade is right. The cash is four times the prize, and it goes out first. ||| 0.7
-Which is why a merchant's credit line is not a back office detail. It is the real position limit. ||| 0.7
-Last thing, and it is the one worth keeping. ||| 0.5
-What separates a trader who lasts ten years from one who has a great year? ||| 0.5
-The great year comes from being big at the right moment. There is no mystery in it, and no way to repeat it on demand. ||| 0.5
-The decade comes from the other end of the same distribution. From never having the year that ends the firm. ||| 0.6
-The arithmetic is brutal and worth memorising. Lose ten percent and you need eleven percent back. Lose twenty five and you need thirty three. ||| 0.4
-Lose fifty percent, and you need one hundred percent just to be level again. ||| 0.6
-A drawdown is not symmetric with the gain that undoes it, and the gap widens fast. ||| 0.5
-So survival is not caution. It is a constraint on the optimisation. A trader who maximises expected profit without it is maximising a number whose variance removes him from the sample. ||| 0.7
-What to keep. ||| 0.4
-The houses stopped selling information and started selling assets, because an elevator is an option, and an option pays whether or not you are right. ||| 0.5
-The edge is a couple of dollars a tonne, and it does not scale with cleverness. It scales with tonnes. That is the whole story of the mergers. ||| 0.5
-The checklist runs funding and exit before direction, because a right trade that cannot be funded is a loss, and a right trade that cannot be exited is a hostage. ||| 0.5
-And a cent a bushel, repeated across a billion bushels, is a better business than a view. ||| 0.7
-That closes the thirty episode run. From here the show goes past the curriculum, into advanced ground and whatever the market is doing that morning. ||| 0.5
-Four questions in the written edition, with full solutions, and today's conversion drill runs from Matif euros into Chicago cents. ||| 0.5
-This has been Soft Commodity Trading. ||| 0.8