@sdelsad/commodity-desk-daily 1.0.50 → 1.0.51

package/ep09.md

@@ -0,0 +1,258 @@
+## Market pulse
+
+**Everything in Chicago rallied, and the crusher still had a worse day than the day before.**
+
+| Commodity | Contract | Price | Change |
+|---|---|---|---|
+| Corn | Sep (CBOT) | 473 c/bu | +9¾¢ |
+| Corn | Dec (CBOT) | 498 c/bu | +10¢ |
+| Soybeans | Sep (CBOT) | 1222¼ c/bu | +21½¢ |
+| Soybeans | Nov (CBOT) | 1237¾ c/bu | +20½¢ |
+| Soymeal | Sep (CBOT) | — | +2.0% |
+| Soyoil | Sep (CBOT) | — | +0.25% |
+| Wheat SRW | Sep (CBOT) | 680¼ c/bu | +15¾¢ |
+| Wheat HRW | Sep (KC) | 762 c/bu | +18¼¢ |
+| Crude palm oil | Sep (BMD) | RM 4,648/t | +RM 32 |
+
+Wednesday was a broad buying day, and the trigger was boots in fields. Every August the Pro Farmer crop tour sends scouts across the corn belt to count ears and pods, and this week's early results were reported as less than stellar. USDA still carries a corn yield of 180.7 bu/ac; after two days of the tour the trade has started to ask whether that is generous. Flooding in the eastern belt and positioning ahead of the weekly export sales report did the rest. Corn added ten cents in December, beans twenty and a half in November, and wheat took the largest percentage move of the three.
+
+The move worth reading was inside the bean complex. Meal led, up about 2%. Oil managed a quarter of a percent. Beans rose more than either product in percentage terms. That combination matters for a crushing plant, and Q3 works it through: a day on which every price on the screen is green can still be a losing day for a plant, because the crush is a difference between prices, not a price.
+
+**The policy read: this morning the risk is a fuel rule, not a war.** Indonesia's B50 programme (50% biodiesel in the diesel pool) came into full effect in July, with the 2026 biodiesel allocation set at 16.75 million kilolitres. To fund the subsidy that makes it work, Jakarta raised the crude palm oil export levy from 10% to 12.5%. The oil exists and the mills are running; it has simply been made expensive to leave the country. That is a supply withdrawal decided in a ministry, and it lands first on the palm–soyoil spread, then on every vegetable oil that competes with either.
+
+```chart
+{"type":"line","unit":"RM per tonne","title":"Palm pays you to wait",
+ "x":["Sep 26","Oct 26","Nov 26","Dec 26"],
+ "series":[{"name":"BMD crude palm oil","values":[4648,4787,4888,4960]}],
+ "caption":"The curve rises 312 ringgit from September to December, about 76 dollars a tonne. In the language of episode 3 palm is in carry: the market is paying to hold oil into the low-production quarter rather than sell it now.",
+ "source":"MDEX crude palm oil futures, Wednesday 19 August 2026, closing quotes."}
+```
+
+## Key takeaways
+
+- The four big vegetable oils (palm, soybean, rapeseed, sunflower) are one market with four tickers, because the buyers want a liquid with a melting point and a price, not a crop, and they can change the recipe. The spread between two oils is the switch that moves demand from one to the other.
+- Palm is the volume leader because a hectare of oil palm yields seven or eight times as much oil as a hectare of soybeans. Its benchmark, FCPO in Kuala Lumpur, is priced in ringgit, so every foreign hedger picks up a currency position for free.
+- One cent per pound is $22.05 per tonne, because a tonne is 2,204.6 pounds. That single factor turns a Chicago soybean oil quote into something you can hold next to a palm quote.
+- A mandate creates a standing bid: demand that does not respond to price. A food buyer walks away when oil gets expensive; a blender under a legal obligation pays.
+- Indonesia's 16.75 million kilolitre allocation is about 15 million tonnes of palm oil consumed at home, close to a third of the crop, and about the size of India's entire annual vegetable oil import demand.
+- Indonesia pays the blending subsidy out of the export levy, and the mandate's purpose is to shrink the exports the levy is collected on. The levy rate, not the mandate headline, is the honest indicator of whether the programme is affordable.
+- Oil and meal are joined at the bushel: 11 lb of oil always comes with 44 lb of meal. A billion pounds of new oil demand drags in four billion pounds of meal that no fuel policy asked for, which is why the crush margin captures far less of an oil rally than the oil chart implies. A fuel policy is always a protein policy.
+- Vegetable oil balance sheets now carry an energy term. When gasoil trades above the oils, refiners blend them into diesel for profit with no mandate at all; when crude falls, that demand disappears within a week, and no crop model can forecast it.
+
+## Vocabulary
+
+| Term | Meaning |
+|---|---|
+| **Vegetable oil** | The liquid fat pressed from an oilseed (soybean, rapeseed, sunflower) or from the fruit of the oil palm; the other product of crushing a seed is the protein meal |
+| **CPO** | Crude palm oil, the unrefined oil pressed from the flesh of the oil palm fruit and the benchmark grade traded internationally |
+| **FCPO** | The Bursa Malaysia Derivatives crude palm oil futures contract: 25 tonnes per lot, quoted in Malaysian ringgit per tonne, tick RM 1 (RM 25 per lot) |
+| **Substitution spread** | The price gap between two competing vegetable oils. Wide enough, and buyers reformulate toward the cheaper one; that switching is what makes four oils one market |
+| **Reformulation** | Changing a food recipe to use more of one oil and less of another, within the limits set by melting point and shelf life |
+| **Biodiesel (FAME)** | Fatty acid methyl ester: the fuel made by reacting a vegetable oil with methanol, which runs in an ordinary diesel engine |
+| **Blending mandate** | A law requiring a set share of the diesel sold in a country to be biodiesel. Indonesia's B50 means 50% |
+| **Standing bid** | Demand that is present regardless of price, because it is created by legal obligation rather than by choice |
+| **Blending subsidy** | The payment a government makes to fuel companies to cover the gap between the cost of biodiesel and the cheaper fossil diesel it replaces |
+| **Export levy** | A tax charged on each tonne of a commodity leaving the country; in Indonesia it both discourages exports and funds the blending subsidy |
+| **Kilolitre** | One thousand litres, one cubic metre. Asian governments state fuel mandates in kilolitres; convert to tonnes with the fuel's density (about 0.88 t per m³ for biodiesel) |
+| **RFS** | The US Renewable Fuel Standard, the rule setting the minimum volume of renewable fuel that must be blended into American transport fuel each year |
+| **RIN** | Renewable identification number, the compliance credit created with each gallon of renewable fuel. Biodiesel earns 1.5 RINs per physical gallon, so a volume stated in RINs is not a volume of fuel |
+| **Oil share** | Oil's share of the value of a crushed bushel; the "oil share trade" is long soybean oil against short soybean meal |
+| **Gasoil** | The traded wholesale price of diesel, and the reference against which blending vegetable oil into fuel is judged profitable or not |
+| **Discretionary blending** | Blending vegetable oil into diesel purely because it is cheaper than gasoil, with no mandate and no subsidy behind it |
+
+## Quiz
+
+**Q1.** September palm settled at RM 4,648/t with the ringgit at 4.08 to the dollar, and Chicago soybean oil was quoted near 69.00 c/lb. A European refiner uses 40,000 t of soybean oil a year and his recipes allow him to replace up to 30% of that volume with palm.
+
+- Express both oils in dollars per tonne and give the substitution spread.
+- What is the annual saving if he switches the full 30% at today's spread?
+- Name the two new exposures the switch creates that he did not have while buying soybean oil, and say why each one arises.
+
+**Q2.** A rule change is expected to add 1.4 billion pounds a year to US soybean oil demand. Using the ep 8 yields of 11 lb of oil and 44 lb of meal per bushel, how many short tons of soybean meal does the extra crush produce?
+
+**Q3.** *(Ep 8)* Take the board crush formula from ep 8 (meal × 0.022 + oil × 0.11 − beans). Start from meal at $325.00/short ton, oil at 69.00 c/lb and November beans at 1217¼. Apply Wednesday's moves: meal +2.0%, oil +0.25%, beans +20½¢. By how many cents per bushel did the board crush change?
+
+**Q4.** *(Ep 6)* Ep 6 said corn is a demand story and that ethanol is one of the two buyers that walks away at a price. On Wednesday US ethanol production fell to a one-month low of 1.089 m barrels a day and ethanol stocks rose 1%, yet December corn rallied ten cents. Why are those two facts not in conflict?
+
+**Q5.** *Conversion drill.* A Mato Grosso soybean field yields 3.72 t/ha and an Illinois field is quoted at 58.5 bu/ac. Which is the higher yield? Soybeans convert at 1 bu/ac ≈ 0.0673 t/ha.
+
+## SOLUTIONS (spoilers)
+
+**A1.** Both sides first, in dollars per tonne.
+
+| Oil | Quote | Conversion | $/t |
+|---|---|---|---|
+| Palm, Sep BMD | RM 4,648/t | ÷ 4.08 RM per $ | $1,139 |
+| Soybean oil, CBOT | 69.00 c/lb | × 22.05 | $1,521 |
+| **Substitution spread** | | | **$382/t** |
+
+The palm step divides ringgit by ringgit-per-dollar to get dollars. The soybean oil step uses the bridge factor: a tonne is 2,204.6 lb, so one cent on every pound is $22.05 on the tonne.
+
+The saving. Thirty percent of 40,000 t is 12,000 t. At $382/t that is **about $4.6 m a year**, which is why formulation teams exist.
+
+The two new exposures.
+
+The first is **currency**. FCPO is priced and settled in ringgit. If the refiner hedges his palm purchases by selling FCPO futures, the hedge gains and loses in ringgit while his costs are in euros or dollars. On 12,000 t at RM 4,648 that is a ringgit position of about RM 56 m that nothing in the physical trade asked for. Buying soybean oil in Chicago, he had dollar exposure, but dollars are the currency he already manages.
+
+The second is **policy**. Palm's export price contains an Indonesian export levy that a ministry can change by decree, as it just did from 10% to 12.5%. Soybean oil carries US policy risk too, through the biofuel rules, but a refiner buying soyoil in Rotterdam is not paying a tax that moves on a fortnight's notice.
+
+Two further answers deserve half credit because they are real: **basis risk**, since CIF Rotterdam palm does not track the Kuala Lumpur futures perfectly, and **specification risk**, since palm and soybean oil have different melting points and oxidative stability, so "reformulate" is not a switch you can flip and unflip weekly.
+
+**A2.** The arithmetic runs through the bushel, and the proportions are fixed by the bean.
+
+| Step | Working | Result |
+|---|---|---|
+| Incremental oil demand | given | 1,400 m lb |
+| Oil per bushel | ÷ 11 lb | 127.3 m bu of extra crush |
+| Meal produced | × 44 lb | 5,600 m lb |
+| In short tons | ÷ 2,000 | **2.8 m short tons of meal** |
+
+**2.8 million short tons of soybean meal**, produced by a rule that was about fuel. It has to be fed to an animal, somewhere, at some price, and the price is what adjusts. This is why an oil-demand headline is a bearish meal headline, and why a trader who likes the oil story buys oil against meal rather than buying the whole crush.
+
+**A3.** Build the change leg by leg. Only the changes matter, not the levels.
+
+| Leg | Move | Multiplier | ¢/bu effect |
+|---|---|---|---|
+| Meal | +2.0% of $325.00 = +$6.50/st | × 0.022 | +14.3 |
+| Oil | +0.25% of 69.00 = +0.1725 c/lb | × 0.11 | +1.9 |
+| Beans | +20.5 c/bu | −1 | −20.5 |
+| **Board crush** | | | **−4.3** |
+
+The board crush **fell about 4.3 c/bu** on a day when all three prices rose. Wednesday was a bad day for a plant.
+
+The trap: the crush is a difference between large numbers, so its sign is set by relative moves, not by direction. Meal was the strongest product in percentage terms and still could not carry the bean move, because meal's contribution is scaled by 0.022 and the bean's by one. A 2% meal rally is worth 14 cents of crush; a 20-cent bean rally costs 20. A screen of green numbers tells you nothing about the margin.
+
+**A4.** They are statements about different things, on different timescales.
+
+The ethanol print is a *demand* datapoint about last week's grind: production at a one-month low with stocks up 1% is a mildly negative corn signal, worth a fraction of a cent. Wednesday's rally was a *supply* datapoint about this year's crop: scouts in the field reporting worse-than-expected conditions against a USDA yield of 180.7 bu/ac that the trade already suspected. In August the supply side of the corn balance sheet has a far larger range of outcomes than the demand side: a two-bushel yield change moves the crop by roughly 180 million bushels, while a soft week of ethanol grind moves demand by a few million. The market prices the larger uncertainty first.
+
+The ep 6 point still bites, later. If ethanol's weakness turns out to be margin-driven rather than seasonal, that is a buyer walking away at a price, and a rally built on supply will be capped by the demand it destroys. The tell is the ethanol margin, not the production number.
+
+**A5.** *Conversion drill.* Soybeans convert at 1 bu/ac ≈ 0.0673 t/ha, or divide by 15 as a fast method.
+
+| Field | Given | Converted |
+|---|---|---|
+| Illinois | 58.5 bu/ac | × 0.0673 = **3.94 t/ha** |
+| Mato Grosso | 3.72 t/ha | ÷ 0.0673 = **55.3 bu/ac** |
+
+**Illinois is the higher yield**, by about 0.22 t/ha or 3.2 bu/ac, roughly 6%. The fast method gets you there too: 58.5 ÷ 15 = 3.9 t/ha and 3.72 × 15 = 55.8 bu/ac. Close enough to answer the question in a phone call.
+
+## The episode, in writing
+
+### Where vegetable oil comes from
+
+Start from something ep 8 established. When you crush a soybean you get two products: oil, the liquid fat, and meal, the protein that is left behind and goes to animal feed. Every oilseed does the same. Rapeseed gives rapeseed oil and rapeseed meal; sunflower seed gives sunflower oil and sunflower meal. The proportions are fixed by the seed, and that fact will matter a great deal later in this edition.
+
+Palm is the odd one out. Palm oil is not pressed from a seed but from the flesh of a fruit that grows on a tree, in a narrow band around the equator. Indonesia and Malaysia grow most of it. And it is by far the biggest of the four, for one reason: a hectare of oil palm produces around 3.5 to 4 tonnes of oil a year, where a hectare of soybeans produces about half a tonne. Seven or eight times the yield is why a crop from one strip of the tropics supplies more vegetable oil than any other single source.
+
+### One market with four tickers
+
+Palm, soybean oil, rapeseed oil and sunflower oil are grown on different continents, harvested on different calendars and traded on different exchanges. They are nonetheless a single market, and the reason is that almost nobody in the chain wants the crop.
+
+A refiner in Rotterdam making margarine, a bottler in Mumbai selling cooking oil, a snack manufacturer in Jakarta frying crisps: each is buying a liquid with a melting point, a shelf life and a price. Within limits, the recipe is a choice. A margarine can carry more palm and less soybean oil, or the reverse. So when one oil becomes expensive relative to another, buyers **reformulate** toward the cheaper one, and that extra demand pulls the cheap oil up and lets the dear one fall.
+
+The vocabulary: the difference between the prices of two oils is the **substitution spread**. When it is narrow nobody bothers to switch. When it is wide enough, recipes change. That switching behaviour is what welds four crops into one system, and it means the spread between two oils is more informative than either price on its own. The spread is the switch.
+
+### Putting the two prices in one unit
+
+To read the spread you need palm and soybean oil in the same unit, and they are not quoted that way.
+
+Palm's benchmark is the **FCPO** contract on Bursa Malaysia Derivatives in Kuala Lumpur.
+
+**The unit moment.** One FCPO lot is 25 tonnes of crude palm oil, quoted in Malaysian ringgit per tonne, with a minimum tick of RM 1, so RM 25 per lot. It is quoted in ringgit because it settles against Malaysian physical delivery. That is a detail with teeth: it is the only major vegetable oil benchmark denominated in neither dollars nor euros, so every non-Malaysian hedger acquires a currency position that no part of the underlying trade asked for.
+
+Converting is one division. September palm closed at RM 4,648/t; the ringgit was near 4.08 to the dollar; 4,648 ÷ 4.08 = **$1,139/t**.
+
+Soybean oil trades in Chicago, quoted in cents per pound, 60,000 lb to a contract. The bridge to dollars per tonne is one number: **1 c/lb = $22.05/t**. The reason is simply that a tonne is 2,204.6 lb, so one cent on every pound is 2,204.6 cents, or $22.05, on the tonne. Multiply the cents-per-pound price by 22.05 and you have dollars per tonne. Mid-August soybean oil near 69.00 c/lb is 69 × 22.05 = **$1,521/t**.
+
+```chart
+{"type":"bar","unit":"US dollars per tonne",
+ "title":"Two oils, one unit",
+ "x":["Crude palm oil, Sep BMD","Soybean oil, CBOT"],
+ "series":[{"name":"Price in $/t","values":[1139,1521]}],
+ "caption":"RM 4,648 divided by 4.08, and 69 cents a pound times 22.05. Once both are in dollars per tonne the substitution spread is visible: soybean oil is about 382 dollars a tonne dearer than palm.",
+ "source":"MDEX and CBOT quotes, Wednesday 19 August 2026; ringgit at 4.08 per dollar."}
+```
+
+| Market | Quote | In $/t |
+|---|---|---|
+| Palm, Sep BMD | RM 4,648/t at 4.08 RM/$ | $1,139 |
+| Soybean oil, CBOT, mid-August | 69.00 c/lb | $1,521 |
+| **Substitution spread** | | **$382/t** |
+
+Nearly four hundred dollars a tonne between two liquids that a refiner can, within limits, use interchangeably. For a European refiner using 40,000 t of soybean oil a year with room to switch 30% of it, that is 12,000 t × $382 = about $4.6 m a year, which is why food companies employ people whose only job is reformulation. And that gap is not a mispricing waiting to be arbitraged. It is the price of two different national policies, and the rest of this edition is about how those policies get there.
+
+### How the switch sounds
+
+> **REFINER:** What are you showing me on September palm, CIF Rotterdam?
+> **BROKER:** Call it forty over the board. Soft. Nobody wants September.
+> **REFINER:** And the soy?
+> **BROKER:** Soy is not competing. It is bid by the fuel guys, not by you.
+> **REFINER:** Then I take palm and I reformulate.
+
+"Forty over the board" means the physical oil, delivered to Rotterdam, costs $40/t more than the Kuala Lumpur futures price: the basis, exactly as in the grain episodes. "Soft" means the seller will negotiate down, because there is more September oil around than buyers for it. And the refiner never argued about the level. He asked for the spread, and when the spread was wide enough he changed his recipe.
+
+The switch is not free. The refiner who moves to palm picks up two exposures he did not have. **Currency**, because palm is priced in ringgit and a futures hedge moves with the ringgit while his costs are in euros. **Policy**, because palm's export price contains an Indonesian levy that a ministry can change overnight, as it just did. Q1 sizes both.
+
+### Reading the palm curve
+
+Wednesday's palm board also carries a shape worth reading (the chart in the market pulse), and it is a chance to apply the curve grammar from ep 3 to a market that behaves nothing like Chicago wheat. The curve rises RM 312 from September to December, roughly $76/t, or about 6.7% over three months. On the ep 3 framing that is a **carry market**: the board is paying you to hold oil rather than sell it prompt. And the carry is front-loaded: the Sep–Oct spread alone is RM 139, then RM 101, then RM 72. A front-loaded carry is not a general statement that oil is abundant. It says oil is abundant *now*, and that the market expects palm's seasonal production decline into the northern winter to tighten things later.
+
+### A mandate is a standing bid
+
+Two definitions first. If you react a vegetable oil with methanol you get a fuel that runs in an ordinary diesel engine: **biodiesel**, or FAME by its chemical name. A **blending mandate** is a law that says a set share of the diesel sold in a country must be biodiesel. Indonesia's B50 means 50%, and the fuel companies are the ones who must comply.
+
+Now think about how an ordinary buyer behaves. Every demand line on a balance sheet assumes that when the price rises, somebody stops buying: the food company reformulates, the importer waits, the farmer feeds less. Demand curves slope downward.
+
+A mandate breaks that. A blender legally required to put 50% biodiesel into the pool does not walk away at a higher palm price. He pays, or he does not sell fuel. In market language a bid is an offer to buy, and a **standing bid** is an offer to buy that is always there, at whatever the price is. That is what a mandate is: a political decision converted into a standing bid for a crop, and it is the single most important structural fact about vegetable oils today.
+
+Work the Indonesian number slowly, because the arithmetic is the lesson.
+
+| Step | Working | Result |
+|---|---|---|
+| 2026 biodiesel allocation | stated by Jakarta | 16.75 m kL |
+| Volume to weight | × 0.88 t per m³ (a kilolitre is a cubic metre; biodiesel is lighter than water) | 14.74 m t of fuel |
+| Fuel to feedstock | × ~1.03 t CPO per t of biodiesel (some weight is lost in the reaction) | **≈ 15.2 m t of CPO** |
+
+Fifteen million tonnes of palm oil, consumed inside the country that produced it. Indonesia produces somewhere around 47 million tonnes a year, so that is close to a third of the crop. For scale, it is roughly what India, the world's largest vegetable oil importer, buys from the whole world in a year. One cabinet decision, sized like the world's largest importer.
+
+### Who pays, and the loop in it
+
+Biodiesel costs more to make than fossil diesel, so the government pays the fuel companies the difference. That payment is the **blending subsidy**. The money comes from the **export levy**, a tax on every tonne of palm oil leaving the country, which was raised from 10% to 12.5% this year.
+
+Here is the loop. The mandate's purpose is to keep palm oil at home, so it reduces exports. Fewer exports means less levy collected. Less levy means a smaller pot to pay the subsidy from. The more successful the programme, the smaller the pot that pays for it. That is not a stable arrangement, and it is exactly why the rate just went up: the pot was running short. So the levy rate, not the mandate headline, is the honest indicator of whether the programme is affordable. Every time it rises, the programme is telling you it is under strain. Watch the rate.
+
+### Why a fuel policy is a protein policy
+
+The United States sets its volumes through the Renewable Fuel Standard, and the final rule for 2026 put biomass-based diesel at 8.86 bn gallons. There is a trap in the American numbers, so this edition skips sizing them: some RFS volumes are stated in physical gallons and some in **RIN** gallons, and biodiesel generates 1.5 RINs per physical gallon. Read the wrong basis and your feedstock estimate is 50% too large. This mistake is made in public, by people who should know better.
+
+The safe method is to work with the change rather than the total. Suppose a rule adds one billion pounds of annual soybean oil demand.
+
+Go back to ep 8. A 60 lb bushel of soybeans yields about 11 lb of oil and 44 lb of meal, in proportions fixed by the bean. You cannot ask a soybean for more oil and less meal. So one billion pounds of oil requires 1,000 ÷ 11 = about **91 million bushels** of additional crush. And those same bushels yield 91 × 44 = about **4.0 billion pounds of meal**, or 2.0 million short tons.
+
+```chart
+{"type":"bar","unit":"million lb per year",
+ "title":"One billion of oil, four of meal",
+ "x":["Soybean oil","Soybean meal"],
+ "series":[{"name":"Produced by 91 m bu of extra crush","values":[1000,4004]}],
+ "caption":"The mandate asks for the oil. The bushel delivers four times as much meal alongside it, about two million short tons that no fuel policy requested and no fuel industry can use.",
+ "source":"Worked example, episode 9, using the 11 lb oil and 44 lb meal yields per bushel from episode 8."}
+```
+
+That meal was not requested by anyone. It has to be fed to an animal, somewhere, at some price, and the only thing that makes a feeder take an extra two million tons is a lower price. So the chain runs: a fuel rule bids up oil, crushers raise run rates to make more oil, the extra meal floods the feed market, and the meal price falls.
+
+Now connect that to the ep 8 board crush: meal × 0.022, plus oil × 0.11, minus the bean. An oil-driven demand shock lifts the oil term. It depresses the meal term, because of the flood. And it raises the bean term, because crushers compete for beans to run harder. Two of the three legs move against a trader who bought the whole crush on the oil headline, which is why the crush captures far less of the move than the oil chart suggests. The clean expression of the view is long oil against short meal, the **oil share** trade: it isolates the one thing the policy actually changes.
+
+The general principle is worth stating cleanly: **any demand shock arriving through one joint product must be released through the other.** Fixed proportions make the by-product the shock absorber, and the by-product market is where the price damage lands. A fuel policy is always, whether it intends to be or not, a protein policy.
+
+### Why balance sheets now have an energy term
+
+Palm and soybean oil each sit on two bids. The first is the mandate: contracted, price-insensitive, known in advance. The second is quieter.
+
+**Gasoil** is the traded wholesale price of diesel. If gasoil trades above the price of palm oil, a refiner can blend palm into diesel and make money on the blend itself, with no subsidy and no legal obligation. That is **discretionary blending**. Through the middle of this year it was live, because gasoil rallied roughly 30% in a fortnight and moved above both palm and soybean oil, so for a while fuel buyers were competing with food buyers for the same tonnes.
+
+That second bid is the energy term, and it has three properties that make it dangerous for a crop analyst. It is **large**: when it is on, it competes directly with food demand. It is **fast**: a per-cargo decision, not a programme, so it switches off within a week of a crude sell-off. And it is **unforecastable from agricultural data**: nothing in a crop model, a stocks report or a weather forecast tells you where gasoil will trade.
+
+The failure mode is symmetrical. Build a balance sheet with no energy line during a high-gasoil regime and you will understate demand, booking the missing tonnes to "residual". Add the line at the top of the cycle and carry it forward through a 30% crude decline, and you will overstate demand by exactly the amount you were previously missing. Both errors feel like diligence at the time.
+
+The discipline is to treat the energy term as a **regime**, not a level: state the gasoil condition under which the line is live, size it, and set it to zero the moment the condition fails. That is a harder forecast than a crop, and it is now unavoidable, because a soybean is no longer only a food.

package/ep09.script.txt

@@ -0,0 +1,124 @@
+Indonesia will burn about a third of its own palm oil crop this year. Not export it. Burn it, as fuel, in trucks and buses. ||| 0.6
+That is not an energy story. It is the largest single demand decision in the vegetable oil market, and it is made in a ministry, not on a trading screen. ||| 0.7
+This is Soft Commodity Trading, episode nine. Today, vegetable oils and biofuels. ||| 0.5
+This is a reworked version of the episode. The first cut packed too much into twelve minutes, so this one is longer, and it takes the ideas one at a time. ||| 0.6
+Here is the map for today. Three ideas. ||| 0.4
+First, the four vegetable oils are really one market, and a single number, the spread between two oils, is the switch that moves demand from one to the other. ||| 0.5
+Second, what a biofuel mandate does to demand. It creates a buyer who does not walk away when the price rises. ||| 0.5
+Third, and this is the one that separates a desk from a headline, why a fuel policy always turns out to be a policy about animal feed. ||| 0.7
+First, the tape. ||| 0.5
+Wednesday was a buying day across the board in Chicago. December corn settled four ninety-eight, up ten cents. November soybeans finished twelve thirty-seven and a quarter, up twenty and a half. ||| 0.4
+Chicago September wheat added fifteen and three quarter cents to six eighty. Kansas City September gained eighteen and a quarter to seven sixty-two. ||| 0.5
+The trigger was the Pro Farmer crop tour. Every August, scouts walk fields across the corn belt, count ears, and report what they see. This week the early results were described as less than stellar. ||| 0.5
+The U S D A still carries a corn yield of one eighty point seven bushels an acre. After two days of the tour, the trade is starting to ask whether that number is too generous. ||| 0.5
+Inside the bean complex, meal was the leader, up about two percent. Oil added a quarter of a percent and no more. Keep that in mind. It comes back at the end. ||| 0.6
+And the policy news this morning is a fuel rule. ||| 0.4
+Indonesia's B fifty programme came into full effect in July. B fifty means that fifty percent of the diesel sold in the country must be biodiesel, made from palm oil. ||| 0.4
+To pay for it, Jakarta raised the export levy on crude palm oil from ten percent to twelve and a half. ||| 0.4
+Hold that thought, because it is the whole subject of the day. ||| 0.7
+Part one. What a vegetable oil is, and why four of them behave like one market. ||| 0.6
+Start from something you already know from episode eight. When you crush a soybean, you get two products. Oil, and meal. The oil is the liquid fat. The meal is the protein left behind, which is fed to animals. ||| 0.5
+Every oilseed does the same thing. Rapeseed gives rapeseed oil and rapeseed meal. Sunflower gives sunflower oil and sunflower meal. ||| 0.4
+And then there is palm, which is different. Palm oil is not pressed from a seed. It is pressed from the flesh of a fruit that grows on a tree, in a narrow band around the equator. Indonesia and Malaysia produce most of it. ||| 0.5
+So there are four big vegetable oils. Palm, soybean oil, rapeseed oil, sunflower oil. Different plants, different continents, different harvest calendars, different exchanges. ||| 0.5
+Now think about who actually buys the oil. ||| 0.4
+A refiner in Rotterdam turning oil into margarine. A bottler in Mumbai selling cooking oil. A snack maker in Jakarta frying crisps. ||| 0.4
+None of them wants a crop. They want a liquid with a certain melting point, a certain shelf life, and a certain price. ||| 0.5
+And within limits, the recipe is a choice. A margarine can be made with more palm and less soy oil, or the other way round. ||| 0.5
+So when one oil gets expensive relative to another, the buyer changes the recipe and moves to the cheaper one. ||| 0.5
+That is why four crops behave like one market. The buyers can switch. ||| 0.5
+And here is the vocabulary. The difference between the price of two oils is called the spread. ||| 0.4
+When the spread is small, nobody bothers to switch. When the spread gets wide enough, buyers reformulate, they move to the cheaper oil, and that extra demand pulls the cheaper oil up and lets the dearer one fall. ||| 0.5
+So the spread between two oils tells you more than either price on its own. The spread is the switch. ||| 0.7
+Now let us put real numbers on that, and to do it we need palm oil's price and soybean oil's price in the same unit. ||| 0.5
+Palm first. Palm is the biggest of the four by a long way, and the reason is yield. ||| 0.4
+A hectare of oil palm gives around three and a half to four tonnes of oil a year. A hectare of soybeans gives about half a tonne of oil. Seven or eight times less. ||| 0.5
+That is why a crop grown in one narrow strip of the tropics supplies more vegetable oil than any other single source. ||| 0.5
+Palm's benchmark price comes from a futures contract in Kuala Lumpur, on the Bursa Malaysia exchange. The ticker is F C P O, for futures, crude palm oil. ||| 0.5
+Here is the unit moment. ||| 0.4
+One F C P O contract is twenty-five tonnes of crude palm oil. It is priced in Malaysian ringgit per tonne, not in dollars. The minimum price move is one ringgit, so one tick is worth twenty-five ringgit on one contract. ||| 0.5
+On Wednesday, September palm closed at four thousand six hundred and forty-eight ringgit a tonne. ||| 0.4
+To turn that into dollars, divide by the exchange rate. The ringgit was at about four point zero eight to the dollar. Four thousand six hundred and forty-eight divided by four point zero eight is about eleven hundred and thirty-nine dollars a tonne. ||| 0.6
+Now soybean oil, which trades in Chicago. ||| 0.4
+Soybean oil is quoted in cents per pound, and one contract is sixty thousand pounds. ||| 0.4
+To compare it with palm we need dollars per tonne, and the bridge is one number. One cent per pound equals twenty-two dollars and five cents per tonne. ||| 0.5
+Here is why. A tonne is two thousand two hundred and four point six pounds. If every one of those pounds is worth one cent more, the tonne is worth two thousand two hundred and four point six cents more. That is twenty-two dollars and five cents. ||| 0.6
+So the rule is simple. Take the soybean oil price in cents per pound, multiply by twenty-two point zero five, and you have dollars per tonne. ||| 0.5
+In the middle of August soybean oil was near sixty-nine cents a pound. Sixty-nine times twenty-two point zero five is about fifteen hundred and twenty dollars a tonne. ||| 0.5
+Put the two side by side. Palm, eleven hundred and thirty-nine. Soybean oil, fifteen hundred and twenty-one. ||| 0.4
+The spread is about three hundred and eighty dollars a tonne. Soybean oil is nearly four hundred dollars a tonne dearer than palm. ||| 0.6
+What does that mean for a real buyer? ||| 0.4
+Take a European refiner who uses forty thousand tonnes of soybean oil a year, and whose recipes allow him to replace up to thirty percent of it with palm. ||| 0.4
+Thirty percent of forty thousand is twelve thousand tonnes. Twelve thousand tonnes, times three hundred and eighty-two dollars, is about four point six million dollars a year. ||| 0.5
+That is why food companies employ people whose only job is reformulation. ||| 0.5
+Here is how the switch sounds on the phone. ||| 0.5
+REFINER: What are you showing me on September palm, C I F Rotterdam? ||| 0.25
+BROKER: Call it forty over the board. Soft. Nobody wants September. ||| 0.25
+REFINER: And the soy? ||| 0.25
+BROKER: Soy is not competing. It is bid by the fuel guys, not by you. ||| 0.25
+REFINER: Then I take palm and I reformulate. ||| 0.6
+A few things to unpack in that call. ||| 0.4
+Forty over the board means the physical oil, delivered to Rotterdam, costs forty dollars a tonne more than the futures price in Kuala Lumpur. That extra forty is the basis, the same idea as in the grain episodes. ||| 0.5
+Soft means the seller is willing to negotiate down, because there is more September oil around than buyers for it. ||| 0.4
+And the refiner did not argue about the level. He asked for the spread, and when the spread was wide enough, he changed his recipe. ||| 0.5
+One more thing. The switch is not free. The refiner who moves to palm picks up two risks he did not have before. ||| 0.5
+The first is currency. Palm is priced in ringgit. If he hedges with F C P O futures, his hedge moves with the ringgit while his costs are in euros or dollars. He now has a currency position that nobody in the physical trade asked for. ||| 0.5
+The second is policy. Palm's export price includes an Indonesian export levy, and a ministry can change that levy overnight, as it just did. Soybean oil has policy risk too, but not a tax that moves on a fortnight's notice. ||| 0.6
+So far, so much like any other commodity. Buyers compare prices, and they switch. Now the part that makes vegetable oils unusual. ||| 0.7
+Part two. What a mandate does to demand. ||| 0.6
+First, the chemistry, in one sentence. If you react a vegetable oil with methanol, you get a fuel that runs in a normal diesel engine. That fuel is called biodiesel, or by its chemical name, FAME. ||| 0.5
+Second, the law. A blending mandate is a rule that says a certain share of all the diesel sold in a country must be biodiesel. Indonesia's B fifty means fifty percent. The fuel companies are the ones who have to comply. ||| 0.5
+Now think about how an ordinary buyer behaves. Every demand line on a balance sheet assumes that when the price goes up, somebody stops buying. The food company reformulates. The importer waits. The farmer feeds less. ||| 0.5
+A mandate breaks that. A fuel company that is legally required to put fifty percent biodiesel in the pool does not walk away when palm gets expensive. It pays, or it cannot sell fuel. ||| 0.6
+In market language, a bid is an offer to buy. A standing bid is an offer to buy that is always there, at whatever the price is. ||| 0.5
+That is what a mandate is. A political decision, converted into a standing bid for a crop. And it is the single most important structural fact about vegetable oils today. ||| 0.7
+Let us size the Indonesian one, slowly, because the arithmetic is the lesson. ||| 0.5
+Jakarta set this year's biodiesel allocation at sixteen point seven five million kilolitres. ||| 0.4
+A kilolitre is a thousand litres, which is one cubic metre. Asian governments state fuel mandates in volume, but we trade oil in tonnes, so the first step is volume to weight. ||| 0.5
+Biodiesel is a little lighter than water. A cubic metre of it weighs about zero point eight eight tonnes. ||| 0.4
+Sixteen point seven five million, times zero point eight eight, is about fourteen point seven million tonnes of finished biodiesel. ||| 0.5
+Then fuel to feedstock. It takes a little more than one tonne of palm oil to make one tonne of biodiesel, because a bit of the oil's weight is lost in the reaction. Call it one point zero three. ||| 0.5
+Fourteen point seven million, times one point zero three, is about fifteen million tonnes of crude palm oil. ||| 0.5
+Fifteen million tonnes. Indonesia produces somewhere around forty-seven million tonnes a year. So that is close to a third of the crop, consumed at home. ||| 0.5
+And for scale, fifteen million tonnes is roughly what India, the world's largest vegetable oil importer, buys from the whole world in a year. ||| 0.5
+A single cabinet decision, sized like the world's largest importer. ||| 0.7
+Now, who pays for it. Because biodiesel costs more to make than ordinary diesel, the government pays the fuel companies the difference. That payment is the blending subsidy. ||| 0.5
+The money for the subsidy comes from the export levy. A levy is a tax charged on every tonne of palm oil that leaves the country. ||| 0.4
+And here is the loop. The mandate's whole purpose is to keep palm oil at home, so it reduces exports. Fewer exports means less levy collected. Less levy means a smaller pot to pay the subsidy from. ||| 0.5
+The more successful the programme, the smaller the pot that pays for it. ||| 0.5
+That is exactly why the levy just went from ten percent to twelve and a half. The pot was running short, so the rate went up. ||| 0.5
+So if you want to know whether the mandate is still affordable, do not watch the mandate headline. Watch the levy rate. Every time it rises, the programme is telling you it is under strain. ||| 0.7
+Part three. Why a fuel policy is always a protein policy. ||| 0.6
+Cross the Pacific. The United States has its own mandate, called the Renewable Fuel Standard. It sets the minimum volume of renewable fuel that must be blended into American transport fuel each year. ||| 0.5
+I am going to skip the exact American volumes today, because there is a trap in them. Some of the numbers are stated in physical gallons and some in credit gallons, where one gallon of biodiesel counts as one and a half. Mix them up and your demand estimate is fifty percent too big. ||| 0.6
+So do it the safe way. Work with the change, not the total. ||| 0.5
+Suppose a rule change adds one billion pounds of soybean oil demand a year. ||| 0.4
+Go back to episode eight. A sixty-pound bushel of soybeans gives about eleven pounds of oil and forty-four pounds of meal. Those proportions are fixed by the bean. You cannot ask a soybean for more oil and less meal. ||| 0.5
+So, one billion pounds of oil, divided by eleven pounds per bushel, is about ninety-one million bushels of extra crush. ||| 0.5
+And here is where it stops being an oil story. ||| 0.4
+Those same ninety-one million bushels also give forty-four pounds of meal each. Ninety-one million times forty-four is about four billion pounds of meal. ||| 0.5
+Four billion pounds is two million short tons of soybean meal. Nobody in the fuel industry asked for it, and nobody in the fuel industry can use it. ||| 0.5
+It has to be fed to an animal, somewhere, at some price. And the only thing that can make an animal feeder take an extra two million tons is a lower price. ||| 0.5
+So the chain runs like this. A fuel rule bids up oil. Crushers run harder to make more oil. The extra meal floods the feed market. The meal price falls. ||| 0.6
+Now connect that to the crush margin from episode eight. The board crush is meal times zero point zero two two, plus oil times zero point one one, minus the bean price. ||| 0.5
+An oil-driven demand shock lifts the oil term. But it depresses the meal term, because of the flood. And it raises the bean term, because crushers are competing for beans. ||| 0.5
+Two of the three legs move against you. So the crush margin captures far less of the oil rally than the oil chart suggests. ||| 0.5
+That is why a trader who believes in the oil story does not buy the whole crush. He buys oil and sells meal. That isolates the one thing the policy actually changes, the value of oil relative to meal. ||| 0.6
+The general rule is worth saying cleanly. When demand arrives through one joint product, the shock is released through the other. The by-product is the shock absorber, and the by-product market is where the price damage shows up. ||| 0.6
+A fuel policy is always, whether it intends to be or not, a protein policy. ||| 0.7
+Remember Wednesday's tape? Meal up two percent, oil up a quarter, beans up twenty cents. Work the formula and the crush margin actually fell on a day when every price was green. That is in the quiz. ||| 0.6
+One last piece, and it is short. Why balance sheets now contain an energy term. ||| 0.5
+The mandate is one bid under vegetable oils. There is a second one, and it is quieter. ||| 0.4
+Gasoil is the wholesale price of diesel. If gasoil trades above the price of palm oil, a refiner can blend palm into diesel and make money on the blend itself. No subsidy, no law, pure arithmetic. That is called discretionary blending. ||| 0.5
+Earlier this year gasoil rallied about thirty percent in two weeks and went above both palm and soybean oil. For a while, fuel buyers were competing with food buyers for the same tonnes. ||| 0.5
+The trouble for a crop analyst is that this bid switches on and off with the crude oil price. When crude falls, it disappears within a week. And nothing in a crop model, a stocks report or a weather forecast tells you where crude will trade. ||| 0.5
+So the discipline is this. Treat the energy demand as a regime, not a level. Write down the condition under which it is live, gasoil above vegetable oil, size it, and set it to zero the moment the condition fails. ||| 0.6
+Add it at the top of the cycle and carry it forward, and you will overstate demand by exactly the amount you used to miss. ||| 0.7
+Three things to take away. ||| 0.5
+One. The four vegetable oils are one market, because buyers can switch recipes. The spread between two oils is the switch, and it tells you more than either price. ||| 0.5
+Two. A mandate converts a political decision into a standing bid, demand that does not respond to price. That is why the biggest risk in this complex is a ministry, not the weather. ||| 0.5
+Three. Oil and meal are joined at the bushel. Extra oil demand means extra meal supply, so a fuel policy is always a protein policy, and the crush margin captures less of the story than the oil chart. ||| 0.6
+And a habit worth building. Whenever you see a mandate number, convert it. Kilolitres or gallons into tonnes of oil, tonnes of oil into bushels, bushels into tonnes of meal. ||| 0.5
+The headline is in gallons. The trade is in the by-product. ||| 0.7
+Next, freight. Vessel classes, chartering, and why the arb dies when the market rallies twenty dollars before you fix the boat. ||| 0.5
+The written edition and today's quiz are in the notes, with the full solutions. Work the conversions before you look. ||| 0.5

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@sdelsad/commodity-desk-daily",
-  "version": "1.0.50",
+  "version": "1.0.51",
   "description": "Soft Commodity Trading - Ep 16: Spreads: Calendar, Inter-Commodity, Inter-Exchange",
   "license": "CC-BY-4.0",
   "keywords": [
@@ -8,5 +8,13 @@
     "commodities",
     "trading",
     "education"
+  ],
+  "files": [
+    "ep09.html",
+    "ep09.md",
+    "ep09.script.txt",
+    "ep09_chart1.png",
+    "ep09_chart2.png",
+    "ep09_chart3.png"
   ]
 }

package/covered.md

@@ -1,20 +0,0 @@
-# Soft Commodity Trading — episodes aired
-
-Running log. Read before writing a new episode: avoid repeating material, and only make callbacks to episodes listed here.
-
-- **Ep 1** (Mon) — *The Units and the Language of the Desk*: Units and quoting grammar; three desk dialogues; see glossary. Pulse: Dec corn 4.65, Nov beans 11.82, Sep wheat 6.51; Black Sea lifting wheat; Midwest rain weighing on corn/beans; WASDE Wednesday named with trade expectations 182.4 corn / 52.9 beans.
-- **Ep 2** (Tue) — *What a Merchant Does, and Why Basis Is the Whole Game*: Space/time/form; physical vs paper; hedging kills flat price; basis as residual; asset-light vs heavy mapped to the transformations; ABCD; why a bull market does not enrich a hedged merchant; three surviving risks. Worked example Santos -20 to Shandong +80 = 30c = 660k on 60kt; 1 dollar board move nets zero, 10c basis = 220k. Broker dialogue on line-ups and river levels. Pulse: levels Dec corn 4.65 Nov beans 11.82 Sep wheat 6.51; WASDE tomorrow with 182.4 vs 183 corn yield and the residual-nature-of-ending-stocks explanation; GEOPOLITICS: Black Sea squeeze, 67 strikes on Ukrainian port facilities in July, Novorossiysk and Taman terminals restricted (20+ mt/yr), Port Kavkaz closed, Azov ~25% of Russian exports constrained, yet Platts milling wheat fell to 225.50 on 4 Aug (13-month low), Russian FOB ~224, Ukrainian domestic -30%, war-risk premium 2-3% of hull, freight premiums +40-80%, up to 10 dollars a tonne — used as the bridge into the basis lesson.
-- **Ep 3** (Wed) — *Futures Plumbing and the Shape of the Curve*: Futures plumbing: tickers (ZW ZC ZS ZM ZL, KC SB CT), liquid months, front month, rolling executed as a calendar spread (Sep-Dec fifteen Dec over dialogue), initial vs variation margin, hedge converts price risk into liquidity risk with 2022 European wheat margin-call case; curve as information: carry/contango vs inverse/backwardation, full carry ceiling and cash-and-carry arb, percent of full carry as message, store-or-sell worked example (wheat 6.30, storage 5c, interest 3c, 24c full carry vs 15c spread, elevator vs own-bin answers), inverse as scream punishing storage twice. Vocab: ticker, front month, roll, calendar spread, Dec over, carry market, inverse, full carry, initial margin, variation margin, limit move. Pulse: WASDE print day — trade avg 182.5 corn yield vs USDA 183, range 180-185, first survey-based state-by-state report, trade-the-surprise framing; Tue closes Dec corn 4.6050 (-1.25), Nov beans 11.6875 (-10.75, 5-wk low), Chi Sep wheat 6.3025 (-10.25), KC 6.99, Matif Sep -5.25 EUR; wheat fell on rumours of Russia-Ukraine safe-passage talks in Turkey — priced the un-trapping mechanism (capacity reopens, world price down, origin basis up, freight/war-risk premiums compress); Ukraine 26/27 export forecast cut to 38-40 Mt
-- **Ep 4** (Thu) — *The Physical Chain, End to End*: Incoterms as risk allocation (FOB/CFR/CIF, risk passes at loading, cost vs risk separate, who charters/insures); execution clock laycan-nomination-NOR-laytime-demurrage/despatch; worked example 60kt FOB Santos beans at ~434 USD/t = 26M cargo, 3 days over at 24k/day = 72k vs 660k margin (11%), interest 4.3k/day; statement of facts and cascading demurrage claims; laycan miss = cancellation into a 40c rally; documents: draft survey, certificate final at load, bill of lading as title, backdating = fraud; execution desk as profit centre; OPS/TRADER dialogue on NOR and turn time. Vocab: Incoterms, CFR, CIF, charter party, nomination, NOR, laytime, weather working day, despatch, statement of facts, draft survey, bill of lading, cancelling date. Pulse: WASDE aftermath - corn yield cut to 180.7 (trade 182.5, prior 183), new-crop ending stocks 1.653bn vs 1.79 July, Dec corn +20.25c to 4.8075 two-week high; beans production +44M above July yet Nov +14.5c to 11.8325 on crush +30M (trade whole sheet, not one row); Chi wheat +22.5c to 6.5275, KC +21.5c to 7.2075; GEO escalation: Tue talks rumour died overnight, Ukraine struck Novorossiysk idling Demetra (8.5Mt) + NKHP (7.1Mt) grain terminals ~15.5Mt/yr, Russian Aug exports est 3.0-3.4Mt, Turkey two-corridor proposal, vessels-on-demurrage-clock bridge into lesson
-- **Ep 5** (Fri) — *Wheat: The Map and the Screens*: Wheat classes and specs (SRW ~10 Chicago, HRW 11-12.5 KC, HRS 13.5+ Minneapolis, durum, Black Sea milling 11.5-12.5); protein, test weight and falling number as the real price, low falling number demotes milling to feed at ~40 USD/t. Four exchanges for one grain: Chicago and KC 5000 bu in c/bu, Minneapolis HRS, Matif EU milling 50 t lots in EUR/t delivered Rouen-Dunkirk; tick symmetry 12.50 dollars vs 12.50 euros; 60kt = 440 Chicago lots vs 1200 Matif lots. KC over Chicago 68c/bu = 26 USD/t as the protein spread and an export-bid signal. Black Sea has no futures - daily price assessments, why an assessment cannot be bought sold or hedged. Cross-hedge worked example: 60kt Russian 12.5 FOB at 224 hedged with 1200 Matif lots, Europe +10 EUR/t = -692k against physical +4 USD/t = +240k, net -452k = 7.5 USD/t slippage; cross-hedge protects against the world moving not your own market; correlation highest on quiet days; EUR/USD exposure created by the hedge itself (~13-14m EUR). MILL/SELLER dialogue on protein, falling number, test weight and the 9-dollar spec spread. Pulse: Thu 13 Aug give-back - Dec corn 4.7775 -0.6 percent, Nov beans 11.8175 flat, Chi Sep wheat 6.5125 -0.2 percent, KC Sep 7.2075 Wed settle; China bought new-crop US beans three days running totalling 505,000 t; GEO escalation - Russia struck Izmail on the Danube, Ukraine's fallback after deepwater loadings ~zero since 22 July, Ukrainian early-Aug shipments -76 percent y/y, wheat export forecast 8.3 Mt, USDA cut Russia+Ukraine exports 2.5 Mt, yet Chicago finished the week unchanged because US sales were only 255,900 t (-14 percent w/w) and the US share of world trade was cut to 9.9 from 10.9 percent - flow substitution needs a buyer who actually switches origin, and they call France, Argentina and Australia.
-- **Ep 6** (Mon) — *Corn, Crop Calendars and Weather Risk*: Corn as a demand story (feed ~2/5, ethanol grind and its margin switch, exports 3.275bn bu, stepped demand curve); corn-wheat feed substitution priced both ways - Dec corn 477.5 = 188 USD/t vs Dec SRW 679 = 249.5 USD/t, 4 percent feeding credit gives a 195 USD/t switch level, 54.50 USD/t gap = 148 c/bu, wheat would need 531; 654k a month on a 20kt mill at 60 percent inclusion; reverse ceiling corn at 240 USD/t = 609 c/bu; BROKER/FEEDER dialogue quoting flat-to-corn rather than a wheat price. Crop calendar table US/Ukraine/Brazil full-season/safrinha/Argentina, US and Ukraine share a hemisphere so not diversified, safrinha is 3/4 of Brazilian corn and its risk is the soybean harvest date in front of it (wet October to May pollination in the dry season). Anatomy of a weather premium: price of a distribution vs trend yield, builds 10-14 days before the window, decays on the calendar not the forecast; Aug WASDE case - yield cut 183 to 180.7 removed 204m bu on 88.6m harvested acres but 2.8m acres found lifted production to 16.013bn, second largest ever; planted vs harvested acres as two denominators, ~8m acres never cut for grain. Pulse: Fri 14 Aug closes higher across the board with wheat up 4 percent on the week (Sep corn 459 +11, Sep beans 1177.75 +11.75, Sep meal 310.20, Sep oil 69.44, Sep Chi wheat 674.75 +22, Dec SRW 679, Dec KC 747.25, Dec corn 477.5, Matif spot 228.25 EUR); GEO escalation - all three Novorossiysk grain terminals suspended by Ukrainian drone strikes, Russian August loadings ~2.5 Mt = under half the five-year pace and weakest August since 2016/17, Ukraine MTD 201.7 kt -76 percent y/y, deepwater corridor shut since 22 July, Russia rejected partial ceasefire for civilian shipping - flow substitution moved from threat to actual buying, which is why this week the price moved and last week it did not
-- **Ep 7** (Tue) — *WASDE and Building a Balance Sheet*: How a grain balance sheet is built line by line, and why ending stocks — the line nobody measures — moves about ten times faster than the crop itself. Plus feed and residual, the line that hides the sins, and why two competent analysts agree on supply and fight about demand.
-- **Ep 8** (Wed) — *The Soybean Complex and the Crush*: One seed, three markets: beans, meal and oil, and the processing margin that runs the industry. Board crush arithmetic step by step, why the plant never earns the screen number, and where a crusher's real optionality sits.
-- **Ep 9** (Thu) — *Vegetable oils and biofuels*: Palm, soy, rape and sun trade as one system, and the spread between them is the switch that rations demand. Then biofuels: how a mandate turns a political decision into a standing bid for a crop, and why a fuel policy is always a protein policy.
-- **Ep 10** (Fri) — *Freight: Dry Bulk and Chartering*: Freight and chartering (see ep10 notes). Pulse: Thu 20 Aug CBOT closes, corn led with Dec above five dollars, Pro Farmer Illinois corn 184.2 vs 199.6 year-ago, BDI 2791; Pulse: Sea of Azov closed to Russian grain, read as a vessel-class constraint rather than a tonnage constraint.
-- **Ep 11** (Mon) — *Storage, Elevation and Trade Flows*: Ep 11 — Storage, Elevation and Trade Flows: the elevator as a seller of space rather than a speculator; elevation margin versus basis-and-carry as two separate businesses; storage tariff in cents per bushel per month and shrink as a percentage; the posted bid as a queue-management tool rather than a price; worked example buying corn at 45 under Dec and selling at 15 under Mar with Mar 18 over Dec, restated against one month as a 48c basis gain less 7.5c interest and 3c shrink for 37.5c net on 3m bu; the carry belongs only to whoever has a bin (ep 3 callback); US storage capacity flat at 25.3 bn bu since 2019 against a 27.5 bn trend, on-farm 13.6 and off-farm 11.9, 80% on-farm utilisation at 1 Dec 2025 and ~5% system surplus, tightest since 1988; temporary storage as the cost that floors the basis; blending as the cheapest form change, worked example 40kt at 12.4% and 20kt at 11.2% blending to exactly 12.0% at 244 against a 250 sale for 6 USD/t gross and 3 net = 180,000 on the cargo; why the blender sets the discount; protein moisture and test weight average while aflatoxin, infestation, unapproved events and falling number do not; replacement value and the bottleneck asset as the answer to why merchants rent ships but own elevators. Pulse: Fri 21 Aug closes Dec corn 508.5 +5 (2.5-year high, +25.25 on week), Nov beans 1239.5 +3 (+47 on week), Sep meal 317.70, Sep oil 69.35, Chi Sep wheat 681.5, KC 756.25, MGE 698.25; Pro Farmer final tour corn 173.2 bu/ac and 15.344 bn bu against USDA 180.7, beans 53.3 against 52.7; GEO escalation on the Black Sea — the storage transmission: 90%+ of Russian Azov-Black Sea export capacity offline, three Novorossiysk terminals suspended, Taman since late July, Azov navigation suspended since July, one working deepwater terminal in a basin that moved 46.3 mt last season, ~140 mt harvested, exporters stopped buying, grain backing up inland and 4th-class Russian wheat at ~12,000 roubles/t against 15,000 a year ago — world price up and farmgate price down in the same crop.
-- **Ep 12** (Wed) — *Coffee: The Market*: Arabica and robusta are two different plants on two different exchanges in two different units, and on Monday one settled at 2.2 times the other. Then certified stocks: why 226,242 bags, under half a day of world consumption, can move a global market five percent in a session.
-- **Ep 13** (Fri) — *Coffee: Differentials, PTBF and Volatility*: A coffee contract does not name a price, it names a differential — and an exporter's entire business fits inside eleven cents a pound. Then price-to-be-fixed: how one trade becomes two decisions, and why fixing risk is sold as market risk and settled as credit risk.
-- **Ep 14** (Mon) — *Sugar: Two Contracts, the Switch and the Refiner*: Ep 14 — Sugar: Two Contracts, the Switch and the Refiner: raws vs whites as two screens one refining step apart (No. 11 is 112,000 lb or 50 long tons in c/lb FOB origin, No. 5 is 50 t in USD/t delivered, bridge 22.0462), white premium 133.17 USD/t on Friday; Center-South Brazil as swing supplier pricing a decision rather than a crop; ATR as the unit of that choice with CONSECANA factors 1.0495 kg ATR per kg sugar, 1.6913 per litre hydrous, 1.7651 per litre anhydrous; one tonne of ATR worth 368.87 as sugar against 264.65 as hydrous and 285.71 as anhydrous at Friday prices, sugar ahead by 104.22 or 40 percent; ethanol parity 12.60 c/lb on hydrous and 13.60 on anhydrous against a 17.56 screen, headroom 109 USD/t that must still cover mill-to-port logistics; the switch-is-spent argument, that far above parity a rally pulls no extra Brazilian tonnes and can only ration demand; two demand curves and the fuel floor, moved by the 32 percent anhydrous blend mandate, crude and the real; refiner's margin per tonne of white 520.30 less 1.06 t of raws at 410.36 less 70 refining equals 39.94, and break-even white premium 93.23 at 17.56 raws against 85.87 at 12c because melt loss is a percentage and not a fee; TRADER/ANALYST parity dialogue. Pulse: Fri 28 Aug settles Oct No.11 17.56 minus 0.63 (-3.5%), Oct No.5 520.30 minus 8.50, Sep Chi wheat 767 plus 24.25 at a three-year high, Sep beans 1276.25 plus 19.75, Sep meal 338.20 plus 8.00, Sep corn 512 plus 1.75; sugar still up ~21% on the month after a 14-month high on 18 Aug; supply cuts Brazil CS June sugar -26.3% y/y to 3.903 Mt, Thailand 26/27 9.5 Mt -15.6%, EU+UK 14.98 Mt an eleven-year low, 26/27 flipped from surplus to deficit (ISO -262 kt, Green Pool -3.2 Mt, StoneX -1.7 Mt), screen ~2c above Brazil's ~15.7 c/lb FOB cost of production; GEO/policy read: India opened a 1 Mt duty-free sugar import window to 31 Oct against a standing 100% duty, monsoon 13% below normal through 26 Aug, retail 48 to ~55 rupees/kg, the largest consumer flipping from occasional exporter to buyer, tempered by a permission not being a purchase with one forecaster at no more than 500 kt clearing.
-- **Ep 15** (Fri) — *Cotton, Rice and Juice*: Ep 15 — Cotton, Rice and Juice: the ICE Cotton No. 2 contract at 50,000 lb with 500 dollars a cent and 5 dollars a point, the 480 lb US bale and about 104 bales to the lot; on-call as cotton's public version of price-to-be-fixed (ep 13 callback), an unfixed on-call sale read as latent mill buying with a first-notice-day deadline and an unfixed on-call purchase as latent grower selling; the 21 August CFTC report of 79,167 unfixed sales against 67,696 purchases for 11,471 net, decomposed by month to Dec 26 minus 1,845, Mar 27 plus 12,519, May 27 plus 7,185, Jul 27 plus 12,651 and Dec 27 minus 18,583, so the signal is a spread and not a flat price; worked example of 620 lots on call against March at plus 780 points fixed at 93.40 instead of 89.93, giving a 101.20 delivered cost, a 31,372,000 dollar invoice and 1,075,700 of cost for waiting, plus the day-one hedge that would have offset it exactly; MILL/MERCHANT dialogue that is entirely about a calendar; thinness as depth rather than notional with Dec corn 27,038, Nov rice 31,400, Dec wheat 37,713 and Dec cotton 43,225 a lot, and days-to-liquidate replacing notional limits; rice thin because only about a tenth of production is traded and policy is the supply curve, juice thin because greening is a permanent reduction in trees. Pulse: Thu 3 Sep settles Dec corn 540.75 -2.75, Nov beans 1316.25 +6, Oct meal 348.60 +5.70, Oct bean oil 69.63 -101 pts, Dec Chi wheat 754.25 -19.75 (-2.6%), Dec cotton 86.45 -248 pts, Nov rough rice 15.70 -2.5c; cotton's late-August contract high near 89.45 on a 38% good crop against 55% a year ago and world ending stocks the lowest since 2011/12; GEO/policy read on China's state reserve cotton auctions clearing in full for 24 consecutive sessions and about 192,497 t placed by 21 August, read as domestic tightness that must eventually be met by imports rather than as a price cap.
-- **Ep 16** (Mon) — *Spreads: Calendar, Inter-Commodity, Inter-Exchange*: Ep 16 - Spreads: calendar, inter-commodity, inter-exchange: the spread as a condition rather than a level; percent of full carry as the only meaningful way to read a calendar spread, CBOT Dec/Mar wheat 15.25c against a 33.18c full carry (9.18c interest at 5 percent on 7.34 plus 24c storage at 8c/bu/month) = 46 percent of carry; the ceiling-and-no-floor asymmetry, so a bear spread is bounded by the free bin-and-deliver arbitrage and a bull spread is not; Matif Dec 246.25 over Mar 244.50 as negative carry and what an inversion says about who needs grain now; wheat-corn 197.25c/bu restated per tonne as 269.70 against 211.31, wheat 27.6 percent over corn and nowhere near the feed-substitution floor; the inter-exchange conversion 734.00c x 36.744 = 269.70 USD/t at 1.1629 = 231.92 EUR/t against Matif 246.25 for a 14.33 EUR/t premium compressing to 7.76 in Mar and 5.14 in May; why that is relative value and not an arb, run both directions against the Matif French milling spec and a Toledo warehouse receipt; TRADER/BROKER spread-quoting dialogue where neither party names a price; three ways a spread carries more risk than the outright it replaced - the unbidden FX leg (30,000 t worked example where the euro took 166,800 of a 457,800 wheat profit), spread margin credit at 70-80 percent buying four times the size, and correlation as an assumption that breaks on the very event that resolves the thesis. Pulse: Labor Day closure so Friday 4 Sep settles - Dec corn 536.75 -4, Nov beans 1309.75 -6.5, Dec Chi wheat 734.00 -20.25 and -50 on the week, Dec KC 802.25 -13.25 and -42, MIAX spring -24.25 on the week, Dec meal 355.10, Dec oil 69.27, Matif Dec 246.25 -2.50; sixth straight business day of soybean flash sales, 250,600 t Friday for 1,347,600 t cumulative; GEO escalation of the Black Sea thread - Russia zeroed its wheat, barley and corn export duty from 1 Sep to 31 Dec (wheat had been RUB 787.5/t) and US envoys travelled to Moscow and Kyiv over the weekend of 5-6 Sep, so the war-risk premium deflated on expectation while 90 percent-plus of Azov-Black Sea loading capacity stays offline and August exports were cut to 2.7-3.1 Mt against 4.5 Mt - transmission read as expectation repricing rather than supply repairing, evidenced by Chicago SRW falling twice as far as Minneapolis spring