@sdelsad/commodity-desk-daily 1.0.51 → 1.0.52

package/ep16.md

@@ -0,0 +1,256 @@
+# Episode 16 — Spreads: Calendar, Inter-Commodity, Inter-Exchange
+
+## Market pulse
+
+**Wheat lost fifty cents on the week in Chicago, and almost none of it was about wheat.**
+
+Monday was Labor Day, so the CBOT day session was shut and Friday's settlements are the last prints available. USDA's Crop Progress report moves to Tuesday, and the September WASDE lands on Friday 11 September.
+
+| Contract | Settle | Change |
+|---|---|---|
+| Dec corn, CBOT | 536¾ ¢/bu | −4 |
+| Nov soybeans, CBOT | 1309¾ ¢/bu | −6½ |
+| Dec Chicago SRW wheat | 734.00 ¢/bu | −20¼ |
+| Dec Kansas City HRW wheat | 802¼ ¢/bu | −13¼ |
+| Dec Matif milling wheat | €246.25 /t | −2.50 |
+| Dec soymeal, CBOT | $355.10 /t | −0.40 |
+| Dec soybean oil, CBOT | 69.27 ¢/lb | −77 pts |
+
+Wheat did the work, and it did it downward: Chicago fell 50 cents on the week, Kansas City 42, Minneapolis 24¼. The rest of the board was quiet by comparison, with corn effectively unchanged on the week and beans supported by a sixth consecutive business day of flash sales — 250,600 t on Friday alone, taking the run to 1,347,600 t of soybeans booked to China and to unknown destinations.
+
+The pressure on wheat came from two directions at once, and neither was a supply number. On 1 September Moscow cut its export duty on wheat, barley and corn to zero through the end of the year; the wheat duty had been RUB 787.5 a tonne. Then American envoys travelled to Moscow and Kyiv over the weekend of 5–6 September.
+
+**The geopolitical read.** Set both against what has not changed. More than ninety percent of Russia's Azov–Black Sea loading capacity is still offline: all three Novorossiysk terminals suspended since mid-August, Taman since late July, Azov navigation suspended, Tuapse the only terminal working in a basin that shipped 46.3 Mt last season. Russia's August export programme was cut to 2.7–3.1 Mt against 4.5 Mt a year earlier. None of that was repaired last week.
+
+So the transmission is not through supply. It is through expectation. A war-risk premium is a price paid for disruption a buyer thinks is coming, and diplomacy changes what he thinks is coming without mending a single loading arm. The duty cut works the same way — it does not create export capacity, it lowers the tax on whatever capacity survives, and analysts read it as three to four dollars a tonne off Russian FOB offers.
+
+The evidence that this was an expectation trade rather than a supply trade is in the spread. Chicago soft red, the class that competes directly with Black Sea wheat for the same export business, fell twice as far as Minneapolis spring, which largely does not.
+
+```chart
+{"type":"bar","unit":"¢/bu","title":"Wheat's week, by class","caption":"The class that competes head-on with Black Sea wheat lost twice what spring wheat lost. The collapse was a spread, not a market.","source":"CBOT, KCBT and MIAX settlements, week ending Friday 4 September 2026","x":["Chicago SRW","Kansas City HRW","Minneapolis spring"],"series":[{"name":"Change on the week","values":[-50,-42,-24.25]}]}
+```
+
+## Key takeaways
+
+- A flat price tells you a level. A **spread tells you a condition** — and the condition is usually the tradeable part.
+- Read a calendar spread as a **percentage of full carry**, never in cents. Fifteen cents means nothing until you know that carrying the grain costs thirty-three.
+- A carry spread has a **ceiling and no floor**. Full carry caps it, because anyone with a bin can arbitrage past that point. Nothing caps an inversion. Long the carry and short the carry are not the same trade run backwards.
+- An inter-commodity spread is a **distance to substitution**. Wheat 27.6 percent over corn per tonne means the feed bid is nowhere near, so nothing is waiting underneath the market to catch it.
+- An inter-exchange spread is an **opinion, not an arbitrage**. No delivery mechanism forces Paris and Chicago together, in either direction, ever.
+- A spread is not a smaller position. The **currency arrives free**, the **margin credit buys size**, and the correlation holding the two legs together is an assumption rather than a contract — one that tends to fail exactly when the story that created it resolves.
+
+## Vocabulary
+
+| Term | Meaning |
+|---|---|
+| **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 — how the trade actually quotes a curve |
+| **bull spread** | A calendar position long the nearer month and short the deferred, which profits when the carry narrows or the curve inverts |
+| **bear spread** | A calendar position short the nearer month and long the deferred, which profits when the carry widens toward full carry |
+| **leg** | One of the individual contracts making up a spread, each executed and margined in its own right |
+| **legging in** | Executing a spread one leg at a time rather than as a single spread order, accepting outright exposure in between in exchange for a better fill |
+| **spread margin credit** | The reduction in initial margin an exchange grants a recognised spread, which lowers the cost of a position without lowering its risk per tonne |
+| **wheat–corn spread** | The price difference between wheat and corn futures, read as the distance wheat must still fall before feeders substitute it into a ration |
+| **FX leg** | The currency exposure that arrives unbidden in an inter-exchange spread whose two legs settle in different currencies |
+| **relative value** | A position expressing a view on the difference between two prices rather than on the direction of either |
+| **convergence** | The pull of a futures price toward the cash value of its deliverable as delivery approaches, which disciplines a calendar spread and has no counterpart across two exchanges |
+
+## Quiz
+
+**Q1.** On Friday 4 September, Chicago December wheat settled at 734.00 ¢/bu and Matif December milling wheat at €246.25/t, with the euro at $1.1629. A relative-value desk thinks the European market is too dear against Chicago and sells the premium in 30,000 t: short Matif December, long CBOT December, equal tonnage. Use 36.744 bu to the tonne.
+
+- What is the Matif premium over Chicago, in euros per tonne, at the moment the trade goes on — and how many contracts is each leg?
+- Two weeks later Chicago December is 772.00 and Matif December is €243.00, with the euro at $1.1900. What is the P&L on the spread, in euros?
+- Of that P&L, how much came from wheat and how much from the currency?
+
+**Q2.** Chicago December wheat settled at 734.00 ¢/bu and March 2027 at 749.25 ¢/bu. Money costs 5 percent and commercial storage runs 8 ¢/bu per month. What percentage of full carry is the December–March spread paying?
+
+**Q3.** The cotton on-call report of 21 August 2026 showed March 2027 carrying 12,519 lots more unfixed sales than unfixed purchases. Does that balance represent latent buying or latent selling in March futures?
+
+**Q4.** A coffee exporter has sold on a buyer's-call price-to-be-fixed contract and is fully hedged with a short futures position. The market rallies thirty cents a pound before the buyer fixes, and he remains flat on price throughout. Which exposure has grown?
+
+**Q5.** *Conversion drill.* Kansas City December hard red winter wheat settled at 802.25 ¢/bu. What is that in dollars per tonne?
+
+## SOLUTIONS (spoilers)
+
+**A1.** The trade is three positions wearing the costume of two. Work each leg in its own currency and convert once, at the end — that discipline is what makes the third part of the question answerable at all.
+
+*The premium on day one.* Chicago has to be dragged into Paris's units before the two numbers can be compared.
+
+| Step | Value |
+|---|---|
+| CBOT Dec | 734.00 ¢/bu |
+| × 36.744 bu/t | $269.70 /t |
+| ÷ 1.1629 $/€ | €231.92 /t |
+| Matif Dec | €246.25 /t |
+| **Matif premium** | **€14.33 /t** |
+
+*The legs.* 30,000 t × 36.744 = 1,102,320 bu, which at 5,000 bu a lot is **220 lots** of CBOT wheat. The Matif contract is 50 t, so the other leg is **600 contracts**. Note that 220 lots is 1,100,000 bu, or 29,937 t — the hedge does not fit the tonnage exactly, and on a spread that residual is an outright position in Chicago, small but real.
+
+*The P&L.* Recompute the premium on the new prices and the new rate.
+
+| | Day one | Two weeks later |
+|---|---|---|
+| CBOT Dec | 734.00 ¢/bu | 772.00 ¢/bu |
+| CBOT in $/t | $269.70 | $283.66 |
+| EUR/USD | 1.1629 | 1.1900 |
+| CBOT in €/t | €231.92 | €238.37 |
+| Matif Dec | €246.25 | €243.00 |
+| **Premium** | **€14.33** | **€4.63** |
+
+The desk was short the premium, so it profits as the premium narrows: €14.33 − €4.63 = **€9.70/t**, and on 30,000 t that is **€291,000**.
+
+*Wheat versus currency.* Hold the exchange rate at 1.1629 and run it again. Chicago at $283.66 would have been €243.93, so the premium would have gone to €243.00 − €243.93 = **−€0.93** — Chicago above Matif, a €15.26 narrowing, worth **€457,800**.
+
+The euro took the difference: €457,800 − €291,000 = **€166,800**, more than a third of the wheat P&L.
+
+**The trap the question is testing:** the desk put on a wheat trade and was paid in wheat and in euros, in roughly two parts to one. Long CBOT is long a dollar-denominated asset — about $8.1 million of it on 30,000 t — and the euro strengthened. Nobody sized that position, nobody approved it, and it does not appear on a wheat risk report. It arrived attached to the spread. The fix is a separate FX hedge on the euro value of the dollar leg, rolled as the leg's value moves; the mistake is believing that a spread whose two legs are equal in tonnes is a position that is flat in anything.
+
+**A2.** Full carry is what it costs to own the grain for the three months between the contracts.
+
+| | ¢/bu |
+|---|---|
+| Interest: $7.34 at 5% for three months | 9.18 |
+| Storage: 8 ¢/bu × 3 months | 24.00 |
+| **Full carry, Dec to Mar** | **33.18** |
+
+The market is paying 749.25 − 734.00 = 15.25 ¢. So 15.25 ÷ 33.18 = **46 percent of full carry**.
+
+Read it: the market is covering slightly under half the cost of storing wheat until March. Near full carry — above roughly 80 percent — the market is paying almost anyone to take grain off its hands, which is what a glut looks like on a curve. Under half, storing is a losing business and the market would rather the grain moved now. Forty-six percent is an ordinary, adequately supplied market with no urgency in either direction.
+
+The second half of the reading is the asymmetry. That 46 percent can rise to about 100 and then stops, because past full carry anyone with an empty bin buys December, stores the wheat, sells March and collects the difference risk-free. There is no equivalent force on the way down. The spread can go to zero and invert without limit. A bear spread — short the front, long the deferred — is therefore a bounded trade; a bull spread is not.
+
+**A3.** **Latent buying**, and the direction is the part that catches people.
+
+An unfixed *sale* is cotton a merchant has sold to a mill at a differential, with the mill holding the right to fix. The mill has the cotton and has not priced it, so its cost rises with the board. To stop that, it must eventually buy futures. Net 12,519 lots of unfixed sales in March 2027 is therefore 12,519 lots of buying that has to arrive in the March contract before first notice day, whatever the mills would prefer.
+
+The trap is symmetry: an unfixed *purchase* — a merchant who has bought from a grower with the grower holding the right to fix — is the mirror image, and resolves as latent selling. Reading the total instead of the net, or reading the net with the sign backwards, turns a forced-buying signal into a forced-selling one.
+
+**A4.** **Credit** — and, alongside it, cash.
+
+He is flat on price: the physical sale and the short futures move against each other cent for cent, which is exactly what the hedge is for. But a thirty-cent rally on a 37,500 lb Coffee C contract is 30 × 375 = **$11,250 a lot**, and his short hedge pays that out in variation margin, in cash, every day the market goes up. The buyer, who holds the winning side of the unfixed leg, has posted nothing at all — his gain sits as an unrealised claim against a contract, not as money in an account.
+
+That is the structure worth remembering: fixing risk is sold as market risk and settled as credit risk. The exporter's exposure is no longer to the coffee price but to whether the buyer is still solvent and still willing to fix when the time comes — and that exposure grows by $11,250 a lot for every thirty cents the market rallies. The desk that funds the margin call is carrying the counterparty, not the market.
+
+**A5.** Wheat converts at 36.744 bu to the tonne, so cents per bushel become dollars per tonne by multiplying by 0.36744.
+
+$8.0225/bu × 36.744 = **$294.78 /t**.
+
+Mentally: take a third of 802 and add a tenth of that third — 267 + 27 ≈ 294. Close enough to quote across a desk, and worth carrying because Kansas City trades in cents while the buyer in Algeria or Nigeria is thinking in dollars a tonne.
+
+## The episode, in writing
+
+### One market, two months
+
+The simplest spread there is: one contract, two delivery months.
+
+Chicago December wheat settled at 734.00 ¢/bu on Friday. March 2027 settled at 749.25. March is 15¼ cents over December, which is another way of saying the market will pay you fifteen cents to hold the wheat for three months instead of selling it now.
+
+Is fifteen cents a lot? On its own the question has no answer. It needs a yardstick, and the yardstick is what holding the wheat actually costs: money and space.
+
+| | ¢/bu |
+|---|---|
+| Interest on $7.34 at 5%, three months | 9.18 |
+| Commercial storage, 8 ¢/bu/month | 24.00 |
+| **Full carry** | **33.18** |
+
+Fifteen and a quarter against thirty-three and a fifth is **46 percent of full carry**, and that is the number a desk actually says out loud. Nobody quotes the December–March at fifteen and a quarter. They say it is at forty-six percent of carry, because the percentage travels between commodities and across years while the cents do not.
+
+The reading is direct. Near full carry, the market is desperate for someone to store grain — supply has arrived faster than demand can absorb it, and the curve is bidding for bin space. Below about half, storage is a losing proposition and the market is asking for the grain now. Forty-six percent describes an unexceptional market: enough wheat, no emergency, no glut.
+
+### The asymmetry that makes a carry trade dangerous backwards
+
+Here is what the percentage hides. The spread has a ceiling and no floor.
+
+It cannot travel far past full carry, because if it did the trade would be free: buy December, put the wheat in a bin, sell March, deliver, and collect the excess over your costs. That arbitrage is available to every commercial with storage, so it caps the carry in practice.
+
+Nothing whatsoever caps the other direction. A carry can narrow to zero and then invert, and it can keep inverting for as long as somebody needs the grain in front of them more than they need it later. There is no counter-trade, because you cannot borrow wheat out of the future.
+
+So the two sides of the same instrument are not mirror images:
+
+- **Bear spread** — short the front, long the deferred. Bounded. The most you can lose is the distance to full carry.
+- **Bull spread** — long the front, short the deferred. Unbounded. An inversion has no theoretical limit.
+
+Desks that blow up on calendar spreads almost always blow up on the second one, having sized it as though it behaved like the first.
+
+### Paris, leaning the other way
+
+Now the same instrument in Europe. Matif December milling wheat settled at €246.25/t, March at €244.50. December is €1.75 *over* March.
+
+There is no percent of carry to compute, because the carry is negative. The market is not paying anyone to store wheat. It is charging them. In plain terms, Europe wants wheat now rather than in March — which is what you would expect of the origin that has to serve the buyers the Black Sea currently cannot.
+
+Two curves, the same grain, the same Friday, leaning in opposite directions.
+
+```chart
+{"type":"line","mode":"index","unit":"index, Dec 26 = 100","title":"Two wheat curves, opposite shapes","caption":"Chicago pays you to wait and Paris charges you for it. Rebased to December, the American curve rises across the year and the European one falls away.","source":"CBOT settlements (USDA AMS) and Euronext milling wheat settlements, Friday 4 September 2026","x":["Dec 26","Mar 27","May 27","Sep 27"],"series":[{"name":"CBOT wheat","values":[734.00,749.25,756.75,756.50]},{"name":"Matif milling wheat","values":[246.25,244.50,244.25,233.25]}]}
+```
+
+### Two crops, one month
+
+The second axis. December wheat at 734.00 against December corn at 536¾ is a spread of 197¼ ¢/bu — wheat is nearly two dollars a bushel over corn.
+
+Cents per bushel is the wrong unit for that comparison, because a bushel of wheat and a bushel of corn are not the same weight. Corn converts at 39.368 bu to the tonne, wheat at 36.744. On a tonne:
+
+| | $/t |
+|---|---|
+| Dec wheat | 269.70 |
+| Dec corn | 211.31 |
+| **Wheat over corn** | **58.39, or 27.6%** |
+
+That number has a use. Wheat has a second life as animal feed, and when it gets cheap enough relative to corn, feeders substitute it into the ration. That substitution is the demand that switches on underneath a falling wheat price — the closest thing wheat has to a floor.
+
+It switches on near parity per tonne, a little above if anything, since wheat carries more protein. Twenty-eight percent over corn is not near parity. So the spread is saying something specific this morning: wheat is still trading as food, and there is no feed bid waiting below it. On a week when wheat fell fifty cents, that is worth knowing.
+
+### Two exchanges, and a currency nobody ordered
+
+The third axis is the hard one, because the two markets are not quoted in the same anything. Chicago is cents per bushel. Paris is euros per tonne. Getting them into one number takes two steps and introduces a third position.
+
+| Step | Value |
+|---|---|
+| CBOT Dec wheat | 734.00 ¢/bu |
+| × 36.744 bu/t | $269.70 /t |
+| ÷ €1 = $1.1629 | €231.92 /t |
+| Matif Dec | €246.25 /t |
+| **Matif over Chicago** | **€14.33 /t** |
+
+Here is how it gets quoted on a desk:
+
+> **TRADER:** Where's Matif–Chicago December?
+> **BROKER:** Fourteen and a third. Paris over.
+> **TRADER:** It was under eight a fortnight ago.
+> **BROKER:** It was. Chicago's done the moving, not us.
+> **TRADER:** Show me thirty in Dec. Sell the premium.
+> **BROKER:** Thirty, Paris over Chicago, working.
+
+Neither of them said whether wheat was going up or down. They quoted one number — the difference — and the trader sold it. He has no view on the wheat price. He has a view on whether Paris and Chicago move apart or together. That is relative value, and it is where physical desks live, because a physical desk very rarely has a flat-price opinion worth acting on.
+
+### Why €14.33 is not an arbitrage
+
+The instinct is to treat a gap that size as free money: buy the cheap market, sell the dear one, wait for convergence. Run it both directions and the instinct dies.
+
+**Buy Chicago, sell Paris.** To collect the €14.33 you would have to deliver wheat against the Matif contract. Matif delivers French milling wheat into French silos, against a specification — around 11 percent protein, a specific weight, a falling number. American soft red winter does not meet it, and it is on the wrong side of an ocean.
+
+**Sell Chicago, buy Paris.** Now you need French wheat sitting in a registered warehouse in the Toledo delivery territory. Same ocean, opposite direction, against a spread worth about $16.67 a tonne. Transatlantic freight alone is several times that before anyone has paid for elevation.
+
+So no delivery mechanism forces these two prices together, in either direction. That is the structural difference between the three spreads in this episode:
+
+- A **calendar spread** inside one contract is disciplined by delivery. Convergence is enforced.
+- An **inter-commodity spread** is disciplined by substitution. Feeders enforce it, eventually, with real demand.
+- An **inter-exchange spread** is disciplined by nothing but the habits of the people trading it. It can widen for six months for no nameable reason, and there is no date on which anyone is obliged to make it stop.
+
+The shape of the premium tells you what it is really pricing.
+
+```chart
+{"type":"bar","unit":"€/t","title":"What Paris pays over Chicago","caption":"Fourteen euros in December, five by May. The premium is dated: it is a price for how long the market expects the Black Sea to stay broken, not a gap waiting to be arbitraged.","source":"Derived from CBOT and Euronext settlements of 4 September 2026, at 36.744 bu/t and EUR/USD 1.1629","x":["Dec 26","Mar 27","May 27"],"series":[{"name":"Matif over CBOT","values":[14.33,7.76,5.14]}]}
+```
+
+### Three ways a spread is bigger than the outright it replaced
+
+A spread sounds safer. Two legs, they offset, the market risk is out. On a desk it is how people lose more money than they ever lost on outrights, for three reasons that compound.
+
+**One: the currency arrives free.** Long Chicago and short Paris on 30,000 t is not two positions, it is three. The Chicago leg is worth about $8.1 million, denominated in dollars, and the book is in euros. Nobody sized that exposure or approved it. It came attached to the spread, and it does not show up on a wheat risk report. In the worked example above it took €166,800 of a €457,800 wheat profit.
+
+**Two: the exchange helps you make it bigger.** A recognised spread earns a margin credit, frequently 70 to 80 percent off the outright requirement. The same margin that carried a hundred lots outright carries four hundred lots of spread. Risk per tonne fell; tonnes rose by more. That is not risk reduction, it is leverage wearing a hedge's clothes — and it is granted automatically, by a clearing system, to a desk that believes it has just become more conservative.
+
+**Three: the correlation is an assumption, not a contract.** Chicago and Paris moved together through August because one story was driving both. Then Moscow zeroed its export duty — and Russian wheat competes with French wheat for North African business far more directly than it competes with American wheat. The story that made the two markets move together is precisely the story whose resolution pulls them apart.
+
+That is the general form, and it is worth stating plainly: a spread is correlated right up until the moment it matters. The event that resolves the thesis is usually the same event that breaks the relationship the position depended on. Which is why the honest way to size a spread is not "these two legs offset" but "what do I lose if they stop offsetting on the day I find out I was right?"

package/ep16.script.txt

@@ -0,0 +1,111 @@
+Same grain. Same Friday. Two exchanges. ||| 0.5
+In Chicago, December wheat was cheaper than March, and March was cheaper than May. ||| 0.4
+In Paris, December was the most expensive month on the board. ||| 0.6
+One of those curves is telling you there is too much wheat. The other is telling you there is not enough. ||| 0.7
+This is Soft Commodity Trading, episode 16. Spreads: calendar, inter-commodity, and inter-exchange. ||| 0.7
+First, the tape. Monday was Labor Day, so the board was shut. The last settlements we have are Friday's. ||| 0.5
+December corn, five thirty six and three quarters, down four cents. November beans, thirteen oh nine and three quarters. ||| 0.4
+December Chicago wheat, seven thirty four, down twenty and a quarter on the day. ||| 0.4
+And that is the number that matters. On the week, Chicago wheat lost fifty cents. ||| 0.5
+Kansas City lost forty two. Minneapolis lost twenty four. ||| 0.5
+Wheat did not fall because of wheat. ||| 0.6
+Two things happened. On the first of September, Moscow cut its grain export duty to zero, through the end of the year. Wheat had been carrying seven hundred and eighty seven and a half roubles a tonne. ||| 0.5
+And over the weekend, American envoys travelled to Moscow and to Kyiv. ||| 0.5
+More than ninety percent of Russia's Azov and Black Sea loading capacity is still offline. Novorossiysk is still down. The berths are still broken. ||| 0.6
+So what actually repriced? ||| 0.4
+Not the supply. The expectation. ||| 0.5
+A war risk premium is a price paid for disruption you think is coming. Diplomacy changes what you think is coming. It does not repair a single loading arm. ||| 0.6
+And notice which market took it hardest. Chicago, fifty cents. Minneapolis, twenty four. ||| 0.4
+Chicago soft red is the class that competes with Black Sea wheat for the same export business. Spring wheat mostly does not. ||| 0.5
+That difference is a spread. And spreads are today's subject. ||| 0.7
+Start with the simplest spread there is. One market, two months. ||| 0.5
+Chicago December wheat, seven thirty four. Chicago March, seven forty nine and a quarter. ||| 0.4
+March is fifteen and a quarter cents over December. The market is paying you fifteen cents to wait. ||| 0.5
+Now. Is fifteen cents a lot? ||| 0.5
+On its own, that question has no answer. You need a yardstick, and the yardstick is what it costs to hold a bushel from December to March. Money, and space. ||| 0.5
+Money first. Seven dollars thirty four, at five percent, for three months. That is about nine cents. ||| 0.5
+Space. Commercial storage runs about eight cents a bushel a month. Three months, twenty four cents. ||| 0.5
+Nine plus twenty four. Full carry, December to March, is about thirty three cents. ||| 0.6
+The market is paying fifteen. Fifteen over thirty three. ||| 0.4
+The market is paying about forty six percent of full carry. ||| 0.7
+Percent of full carry. Learn that phrase, because it is how the entire grain trade talks about a curve. ||| 0.5
+Nobody says the December March is fifteen and a quarter. They say it is at forty six percent of carry. ||| 0.6
+Above eighty percent, the market is begging somebody to store grain. Below about half, storing is a losing business. ||| 0.5
+Forty six percent is the middle. Adequate supply, no glut. ||| 0.6
+That spread has a ceiling, and it has no floor. ||| 0.6
+It cannot go far past full carry. If it did, anyone with an empty bin would buy December, store the wheat, sell March, and collect the difference for nothing. ||| 0.6
+But there is nothing stopping it going the other way. A carry can collapse to zero, and then invert, and no arbitrage anywhere makes it stop. ||| 0.6
+So long the carry is a trade with a floor under it. Short the carry, long the front and short the deferred, is not. ||| 0.6
+Same instrument. Completely different animal depending which way round you hold it. ||| 0.7
+Now Paris. ||| 0.5
+Matif milling wheat. December, two hundred and forty six twenty five a tonne. March, two hundred and forty four fifty. ||| 0.5
+December is over March. ||| 0.5
+There is no percent of carry to compute, because the carry is negative. The market is not paying you to store. It is charging you. ||| 0.6
+In plain language. Europe wants wheat now. Not in March. Now. ||| 0.5
+Which makes sense, because Europe is the origin that has to serve the buyers the Black Sea cannot. ||| 0.6
+So, on the same Friday. Chicago paying you forty six percent of carry to wait. Paris charging you for the privilege. ||| 0.6
+Two curves, same grain, opposite shapes. ||| 0.7
+Second axis. Two different crops, same month. ||| 0.5
+December wheat, seven thirty four. December corn, five thirty six and three quarters. Wheat is a dollar ninety seven over corn. ||| 0.5
+But cents per bushel is the wrong unit for that comparison, because a bushel of wheat and a bushel of corn are not the same weight. ||| 0.5
+So put them both on a tonne. Wheat, two hundred and sixty nine dollars seventy. Corn, two hundred and eleven dollars thirty one. ||| 0.6
+Wheat is twenty eight percent more expensive than corn, per tonne. ||| 0.6
+That matters because wheat has a second life as animal feed. When it gets cheap enough, feeders put it in the ration instead of corn. ||| 0.5
+That substitution is the floor under the wheat price, and it switches on near parity with corn per tonne. A little above, maybe, because wheat carries more protein. ||| 0.5
+Twenty eight percent over is not near parity. ||| 0.6
+So the wheat corn spread is telling you something quite specific this morning. Wheat is still trading as food. ||| 0.5
+There is no feed bid underneath it. Nothing is waiting to catch it. ||| 0.7
+Third axis. The hard one. Two exchanges. ||| 0.5
+Chicago quotes cents per bushel. Paris quotes euros per tonne. Before you can compare them at all, you have to get them into the same money. ||| 0.5
+Chicago. Seven dollars thirty four a bushel, times thirty six point seven four four bushels to a tonne of wheat. Two hundred and sixty nine dollars seventy a tonne. ||| 0.6
+Then the euro. Friday's rate, one dollar sixteen twenty nine to the euro. Divide. Two hundred and thirty one euros ninety two. ||| 0.6
+Paris settled at two hundred and forty six twenty five. ||| 0.5
+So Matif is trading fourteen euros thirty three over Chicago. ||| 0.7
+Here is how that actually gets talked about. ||| 0.5
+TRADER: Where's Matif Chicago December? ||| 0.25
+BROKER: Fourteen and a third. Paris over. ||| 0.25
+TRADER: It was under eight a fortnight ago. ||| 0.25
+BROKER: It was. Chicago's done the moving, not us. ||| 0.25
+TRADER: Show me thirty in Dec. Sell the premium. ||| 0.25
+BROKER: Thirty, Paris over Chicago, working. ||| 0.6
+Notice what neither of them said. Neither said whether wheat was going up or down. ||| 0.5
+They quoted one number, the difference. And the trader sold that difference. ||| 0.5
+He does not care where wheat goes. He cares whether Paris and Chicago move apart or move together. ||| 0.6
+That is relative value, and it is where physical desks live. A physical desk rarely has a view on flat price. It has a view on one market against another. ||| 0.7
+So. Fourteen euros thirty three. Is that an arbitrage? ||| 0.6
+Nearly everyone new to this assumes it must be. Buy the cheap one, sell the dear one, wait for them to converge. ||| 0.5
+It is not. And understanding why is the most valuable thing in this episode. ||| 0.6
+Run it both directions. Buy Chicago, sell Paris. To collect that fourteen euros you must deliver wheat against the Matif contract. ||| 0.5
+Matif delivers French milling wheat into French silos. Eleven percent protein, a specific weight, a falling number. ||| 0.5
+American soft red winter does not meet that specification, and it is on the wrong side of an ocean. ||| 0.6
+Other way. Sell Chicago, buy Paris. Now you need French wheat sitting in a registered warehouse in Toledo. ||| 0.5
+Same ocean, running the other way, against a spread worth about sixteen dollars a tonne. Freight alone is several times that. ||| 0.6
+So there is no delivery mechanism anywhere that forces those two prices together. And that is the whole point. ||| 0.6
+A calendar spread inside one contract is disciplined. Delivery is the enforcement. ||| 0.5
+An inter-exchange spread is disciplined by nothing except the habits of the people trading it. ||| 0.6
+It can widen for six months for no reason you can name, and there is no date on which anybody is obliged to make it stop. ||| 0.7
+A spread sounds safer than an outright. Two legs, they offset each other, you have taken the market risk out. ||| 0.5
+On a desk, the spread is how people lose more money than they ever lost on outrights. Three reasons. ||| 0.6
+One. You are carrying a currency you never asked for. ||| 0.5
+Long Chicago and short Paris on thirty thousand tonnes is not two positions. It is three. ||| 0.5
+The Chicago leg is worth about eight million dollars. Those are dollars. Your book is in euros. ||| 0.5
+Nobody decided to take that risk. It arrived free with the spread. ||| 0.6
+Two. The exchange will help you make it bigger. ||| 0.5
+A recognised spread gets a margin credit. Often seventy or eighty percent off the outright requirement. ||| 0.5
+So the same margin that carried a hundred lots outright will carry four hundred lots of spread. ||| 0.5
+Risk per tonne went down. Tonnes went up by more. That is not a hedge. That is leverage wearing a hedge's clothes. ||| 0.6
+Three, and this is the one that does the damage. The correlation you are relying on is an assumption. It is not a contract. ||| 0.6
+Chicago and Paris moved together all through August, because a single story was driving both of them. ||| 0.5
+Then Moscow zeroed its export duty. And Russian wheat competes with French wheat for North African business far more directly than it competes with American wheat. ||| 0.6
+So the story that made those two markets move together is precisely the story whose resolution pulls them apart. ||| 0.6
+Your spread is correlated right up until the moment it matters. ||| 0.7
+So. What to keep. ||| 0.5
+A flat price tells you a level. A spread tells you a condition. ||| 0.6
+Read a calendar spread as a percentage of full carry, never in cents, because the percentage is the only version of it that means anything. ||| 0.5
+Read an inter-commodity spread as a distance to substitution. It tells you whether there is a bid waiting underneath the market. ||| 0.5
+Read an inter-exchange spread as an opinion, not an arbitrage. Nothing in the world makes it converge. ||| 0.6
+And a spread is not a smaller position. It is a different one, and very often a bigger one. ||| 0.7
+Next time, options. Not the pricing theory. The way hedgers actually use them on a desk. ||| 0.5
+The quiz is in the notes and in the e-mail. Four questions and a conversion drill. ||| 0.4
+Question one puts this spread on in thirty thousand tonnes, and asks how much of the profit the euro quietly took. ||| 0.5
+That's Soft Commodity Trading. ||| 0.6