@sdelsad/commodity-desk-daily 1.0.67 → 1.0.68

package/covered.md

@@ -23,3 +23,4 @@ Running log. Read before writing a new episode: avoid repeating material, and on
 - **Ep 19** (Fri) — *Basis Deep Dive and Origination*: Ep 19 - Basis Deep Dive and Origination: the four ingredients of basis - freight, farmer selling, end demand and space - and none of them a view on price; the three-bucket decomposition of a hedged merchant's P&L with flat price structurally zero; worked example 1,000,000 bu of central Illinois corn bought at Dec minus 35 with Dec at 533.75 for 498.75, hedged 200 lots, rolled Dec into Mar at 14c of carry, sold at Mar plus 5, so basis 40 plus calendar 14 equals 54c gross or 540,000 dollars, less 16c storage and 8.31c interest for 29.69c net or 296,900 dollars; the carry covered 14 of an 18.23c three-month cost, about 77 percent of full carry, so the basis must earn the rest (ep 16 callback at 46 percent); the farmer contract menu as a table of risk transfers - cash, forward cash, basis contract, hedge-to-arrive, deferred price, minimum price - and what each leaves on the elevator's book; the 1996 HTA inversion and why an open leg is a position; FARMER/ORIGINATOR basis-contract dialogue with a February pricing deadline; why farm selling clusters on round numbers, cash-flow dates and a full bin, and why that clustering lands entirely on the posted bid rather than the board; basis push as paying for delivery speed; relationships as infrastructure - the unprinted quality spread and credit spread that make two neighbours' bids four cents apart and both correct. Pulse: Thu 10 Sep settles Dec corn 533.75 +6, Nov beans 1332.25 +22.75, Dec Chi wheat 741.25 +12.5, Dec KC 818.75 +12.5, Dec spring 762.50 +14.5, Oct meal 350.60 +5.50, Oct oil 71.41 +133 pts, Matif Dec 245.25 +0.50; the bid came from energy with Oct WTI +6.00 to 102.06 on Persian Gulf fighting, transmission named as the oil share, freight and bunkers, and war-risk premium quoted per voyage; China took 272,000 t beans plus 206,500 t unknown; Gulf CIF basis unchanged at 60-66 over Dec corn and 100-102 over Nov beans while flat price rallied, used as the bridge into the lesson; USDA barge freight index 221.70 to 250.44 in one week with truck, rail and ocean all higher; WASDE Friday 11 Sep with the trade looking for 178.1 corn yield against 180.7, production 15,768 m bu and ending stocks 1,533 m bu, beans 52.5 and 289 m bu; Black Sea read - Russian wheat eased to about 210 USD/t with September loadings about 1 Mt behind the 4.6 Mt of a year ago despite strikes on Novorossiysk, Nika-Tera and Makhachkala inside 24 hours, damaged capacity already in the price and no buyer yet short of tonnes.
 - **Ep 20** (Mon) — *Destination Markets, Tenders and the Winner's Curse*: Ep 20 - Destination Markets, Tenders and the Winner's Curse: how importers buy through tenders and how an export desk prices a bid backwards from the destination as a netback; worked example 60,000 t milling wheat CFR North Africa = 2,204,640 bu = 441 lots, FOB Gulf replacement Dec 725.25 plus 92 = 817.25c = 300.29 USD/t, plus freight 31.50, financing 25 days at 6 percent 1.37, outturn 0.15 percent 0.50, bonds and agent 0.35 for a delivered cost of 334.01, awarded at 334.50 for a margin of 0.49 USD/t or 29,400 dollars or 1.33 c/bu; the winner's curse quantified - ten bidders with 2.00 USD/t estimate dispersion means the winning bid sits 1.54 standard deviations low, 3.08 USD/t or 184,656 dollars below true cost, six times the margin, so the two answers are bid shading and bidding only from facts rather than forecasts; tender validity as a free option handed to the buyer for six hours after bids close; TRADER/AGENT tender dialogue where the trader quotes a bid he expects to lose and prices the optional origin; the destination store-or-sell - November CFR 336.00 against January 342.00 pays 6.00 USD/t to wait while silo at 2.20/t/month for two months is 4.40 and financing at 7.5 percent is 4.20 for a total 8.60, so storing loses 2.60 USD/t or 156,000 dollars and the break-even borrowing rate is about 2.9 percent; the depth that a state importer manages days of cover on a subsidy and FX allocation calendar rather than a P&L, so the 2.60 is an insurance premium, and the two consequences for the seller - clustered tender demand moving basis and freight together in the week the bid already fixed them, and importing markets showing less carry than exporting markets because storage sits where capital is cheapest. Pulse: Friday 11 Sep settles after the September WASDE - Dec corn 530.25 -3.5, Nov beans 1296.50 -35.75, Dec Chi wheat 725.25 -16, Dec KC 798.50 -20.25, Dec MIAX spring 745.00 -17.5, Oct meal 346.80 -3.80, Oct oil 69.19 -222 pts; the WASDE print itself as the escalation of the thread built in eps 18 and 19 - corn yield cut to 178.5 from 180.7 but 0.4 above the trade's 178.1, production 15.800 bn bu, carryout 1.567 bn against 1.533 expected, stocks-to-use 9.7 percent, beans yield 52.8 production 4.535 bn carryout 310 m against 290 expected, US wheat carryout 717 m in line, world wheat stocks 276.29 Mmt against 273.0 expected, so all six headline numbers above the trade guess and a cut smaller than the one you are positioned for is a bearish cut; corn export sales 1.929 Mmt to 3 Sep and Mexico a further 264,000 t, wheat commitments 322 m bu -31 percent y/y; GEO escalation of the Black Sea thread to flow substitution - Russian September loadings 1.6-2.0 Mt against 4.9 Mt a year ago while Asian buyers took at least 500,000 t of Australian and Argentine wheat instead, transmission named as differential repricing rather than flat price, evidenced by world wheat stocks revised 3.3 Mmt higher in the same week, supply not missing but misplaced and moving it costing freight
 - **Ep 21** (Wed) — *The Book and P&L Attribution*: Ep 21 - The book and P&L attribution: the position sheet as a statement of exposure rather than inventory, signed long-positive short-negative with physical and paper in the same row, rows as futures months not shipment months because a futures month is the only thing you can trade to change the number; mark to market on physical too, so P&L moves nightly on grain nobody has agreed to buy; worked position sheet of a soft red book netting to zero on the total while Dec is square, Mar is long 400,000 bu unhedged and May is short 80 lots, an 80-lot Mar/May spread nobody decided to own; cost of that error bounded at full carry - Mar/May full carry 22.25c (16c storage at 8c/bu/month plus 6.25c interest at 5 percent on 7.50) against a 14c spread = 63 percent of carry, 8.25c of room to widen on 400,000 bu = 33,000 dollars, and the mirror-image short-the-near position unbounded because an inversion has no ceiling (ep16 and ep18 callback); the location split, long Toledo against short Gulf both hedged Chicago Dec is square by month and long the Toledo-Gulf basis spread, evidenced by Gulf corn basis unchanged at 60-66 over Dec through a flat-price rally (ep19 callback); RISK/TRADER dialogue where neither party names a price; reconciliation of trader sheet, back office and risk system every morning and the discipline of explaining differences rather than agreeing; attribution as six lines - flat price, basis, calendar spread, freight, currency, financing - of which only one is a market view; worked attribution of 45,000 t corn central Illinois to FOB Gulf = 1,771,560 bu = 354 lots, planned 28.07 c/bu = 497,277 dollars (Dec -40 buy at 508.50, Dec +62 sale, 58c freight, 12c elevation, 2c shrink, 1.93c financing at 6 percent for 25 days), realised 195,587 with the bridge basis -124,009 (sold +55 not +62), freight -177,156 (68c not 58c), unhedged bushels +425, margin financing -950, closing exactly to the 301,690 gap; quantity risk as structural because 354 lots is 1,770,000 bu against a 1,771,560 bu cargo so 1,560 bushels carry the entire flat-price P&L of a 9.5m dollar cargo; margin financing as the line that scales with number of cargoes rather than quality and shows up as a treasury problem before a P&L problem; the depth point that the same 195,587 would have printed had Dec corn fallen 27.25c instead of rising it, so a hedged merchant's P&L carries no information about direction and only information about whether costs and differentials were where you said. Pulse: Tue 15 Sep settles Dec corn 535.75 +2.5, Nov beans 1318.75 +14.5, Dec Chi wheat 728.50 +6.5, Oct meal 360.10 +9.90, Oct oil 69.88 +23 pts, after Dec SRW printed 712.50 -9.5 at 8:30 CDT for a roughly 16c intraday round trip; Monday's US presidential post that Russia and Ukraine had agreed to stop attacks on each other's energy infrastructure knocked wheat about 18c before fresh strikes near Odesa took it back, both sides having attached conditions; crop progress corn 57 percent G/E, 86 dented, 42 mature, 8 harvested, beans 58 percent G/E and 6 harvested, spring wheat harvest 93 percent, winter wheat planting 8 against 12 normal; crude strength spilling into beans and corn Monday; GEO escalation of the Black Sea thread from ep20's flow substitution to SCOPE - an energy-infrastructure truce is not a grain corridor, refineries and oil berths are one target set and grain terminals at Novorossiysk and Odesa another, so the board was repricing the scope of the announcement rather than the probability of disruption, transmission running energy-truce to crude to bunkers to freight and only then to grain, against loading data that did not move at all - Russian seaborne grain exports 2.0 Mt in August -62 percent y/y, Ukraine grain exports since 1 July 4.34 Mt -24 percent with wheat 2.1 Mt -48 percent, and Algeria, Pakistan and Saudi Arabia covering elsewhere
+- **Ep 22** (Fri) — *Risk Management and Why Hedges Are Never Perfect*: Ep 22 - Risk management and why hedges are never perfect: the six channels a hedge leaks through once flat price is removed - basis, timing, quality, currency, cross-hedge and quantity - each quantified on one cargo; worked example 27,000 t of 46 percent protein soybean meal sold CFR Rotterdam at 402 EUR/t for November arrival at 1.1750 = 472.35 USD/t, freight 38.00 and finance-insurance-outturn 6.00, planned purchase FOB New Orleans at the October board plus 8.00 per short ton with the board at 368.70, 27,000 t x 1.102311 = 29,762.397 short tons = 297.62 lots rounded up to 298, planned margin 13.11 USD/t = 353,954 dollars; the six leaks - basis -178,574 (paid plus 14 not plus 8), timing -89,287 (hedged October while the physical priced against December at 3.00 over), quality -129,600 (outturn 45.2 against 46.0 on a 6.00 per point protein allowance = 4.80/t), currency -162,810 (10,854,000 euros left unsold as EURUSD went 1.1750 to 1.1600, of which 19,537 was forward points and 143,273 avoidable), cross-hedge -94,500 (FFA index route returned 2.00/t against actual freight up 5.50/t), quantity -628 (37.603 short tons of excess futures on a 16.70 board fall) - total leakage 655,399 for a realised -301,445 or -11.16 USD/t against a planned +13.11, with the board contributing nothing; TRADER/TREASURY dialogue on forward points where minus eighteen points is an adjustment to spot and not a price; three limits over a physical desk - position in tonnes and lots by month, loss as VaR and a stress number, liquidity as days to liquidate - plus the usually unwritten concentration limit on counterparty, port and origin; value at risk worked on 2,000,000 bu of corn basis at 1.2 c/bu daily volatility, 1.645 sd = 1.97 c/bu = 39,480 dollars one-day 95 percent, scaled by root 21 = 180,920, against a 25 c/bu three-week Gulf basis stress = 500,000 dollars, 2.8 times the monthly and 12.7 times the daily figure; three reasons VaR flatters a physical book - marks are assessments rather than trades, no screen exists so the liquidation horizon is fiction (ep 15 days-to-liquidate callback), and the covariance matrix is estimated on quiet days so correlations break on the resolving event (ep 16 callback); risk reports to the CFO because the owner of the P and L cannot also mark and size it, and a limit is a statement about fundable mistake size rather than a forecast; depth point that a chosen risk has a nameable price while an inherited risk does not, that inherited risk is deleted or converted rather than hedged harder, and that the irreducible residue - lot rounding and cross-hedge basis - belongs in the quoted margin rather than the risk report. Pulse: Thu 17 Sep settles Dec corn 530.50 -3.75, Nov beans 1319.75 -0.75, Oct meal 368.70 +7.80, Oct bean oil 68.68 -51 pts, Dec Chi wheat 727.00 -3.75, with Dec KC 799.50 +3.25 from Wednesday; meal up about 7 percent over six sessions while oil lost 2 percent, a product-spread transfer inside a crush that barely changed, used as the bridge into the lesson; export sales w/e 10 Sep beans 1,702.0 kt mostly China and unknown, corn 1,026.7 kt a three-week low read as South American competition, wheat 325.9 kt to the Philippines and Mexico; corn harvest 6 percent, winter wheat planting 12, rice harvest 50; FranceAgriMer cut French soft wheat exports outside the EU to 6.3 Mt from 7.0 and intra-EU to 7.1 from 7.4, soft wheat ending stocks 3.01 Mt from 3.65, maize ending stocks 1.46 Mt -26 percent y/y and the smallest crop since the 1970s; GEO escalation of the Black Sea thread from the scope-of-the-truce reading in ep 21 to the NEGOTIATION CHANNEL - Ukraine floated talks with Russia on restarting grain exports from both countries and Russia called the proposal impractical while shipping stays very limited at both origins, transmission named as insurance rather than announcement because a corridor reopens when underwriters reprice war risk on hulls and cargo and not on a communique, evidenced by Chicago wheat falling on the day the headline printed; the policy channel the board did pay for was Washington-Beijing with about 1 Mt of US beans bought in the week, roughly half of a 25 Mt per year commitment running to 2028, a 10 percent Chinese tariff on US farm goods in play and a 24 September meeting.

package/ep22.md

@@ -0,0 +1,285 @@
+# Market pulse
+
+**Soybean meal settled $7.80 higher while bean oil lost 51 points. The crush itself barely moved. This was money changing seats inside the complex rather than a new view on beans.**
+
+| Contract | Last | Change |
+|---|---|---|
+| Dec corn (CBOT) | 530.50 c/bu | −3¾ |
+| Nov soybeans (CBOT) | 1,319.75 c/bu | −¾ |
+| Dec Chicago SRW (CBOT) | 727.00 c/bu | −3¾ |
+| Oct soybean meal (CBOT) | $368.70/short ton | +7.80 |
+| Oct soybean oil (CBOT) | 68.68 c/lb | −51 pts |
+
+Corn and wheat both gave a little back on profit-taking and technical selling. Meal's bid came from crush margins and product-spread trade, and oil followed crude lower. Over six sessions meal has added about seven percent while oil has lost two, which is a product spread doing the work rather than a bean story.
+
+Export sales for the week to 10 September: soybeans 1,702.0 kt, mostly China and unknown destinations; corn 1,026.7 kt, a three-week low that the trade read as South American competition; wheat 325.9 kt, largely to the Philippines and Mexico. Corn harvest is 6 percent complete, winter wheat planting 12 percent, rice harvest 50 percent. December Kansas City hard red last settled at 799.50 on Wednesday, up 3¼.
+
+FranceAgriMer cut its French soft wheat export forecast outside the EU to 6.3 Mt from 7.0 Mt in July, and the intra-EU figure to 7.1 Mt from 7.4 Mt. It put maize ending stocks at 1.46 Mt, down 26 percent year on year and the smallest in three decades after a hot, dry summer.
+
+```chart
+{"type":"line","mode":"index","unit":"index, 9 Sep = 100","title":"Meal up, oil down, crush flat",
+ "caption":"Meal has gained about seven percent in six sessions while oil has lost two. A crush that barely changed hides a large transfer between its two products, and a hedge placed on one of them was never a hedge on the other.",
+ "source":"CBOT settlements for October soybean meal and October soybean oil, 9 to 17 September 2026.",
+ "x":["9 Sep","10 Sep","11 Sep","15 Sep","17 Sep"],
+ "series":[{"name":"Oct soybean meal","values":[345.10,350.60,346.80,360.10,368.70]},
+           {"name":"Oct soybean oil","values":[70.08,71.41,69.19,69.88,68.68]}]}
+```
+
+## The geopolitical read
+
+Two policy channels ran at once, and only one of them was paid for.
+
+Ukraine floated sitting down with Russia to restart grain exports from both countries. Russia called the proposal impractical. Shipping stays very limited at both origins. Chicago wheat fell on the day that headline printed, which looks wrong until you name the transmission. A corridor does not reopen on a communiqué. It reopens when underwriters reprice war risk on hulls and cargo, because a charterer cannot fix a vessel against an intention. Until the insurance moves, an agreement moves no tonnes.
+
+The channel the board did pay for was Washington–Beijing. China bought roughly 1 Mt of US soybeans in the week, about halfway towards a commitment of 25 Mt a year running to 2028, with a 10 percent Chinese tariff on US farm goods in play at a meeting on 24 September. That is the policy transmission in its simplest form: a tariff rate is a term in the delivered cost of every US cargo, and no futures contract prices it.
+
+Hold on to the meal number, because today's cargo is a meal cargo.
+
+# Key takeaways
+
+- A hedge removes flat price. It removes nothing else. Six channels stay open after the futures are on: basis, timing, quality, currency, cross-hedge and quantity.
+- The leaks are not small change. On one 27,000 t meal cargo they came to about $655,400 against a planned margin of $353,954, and the flat price contributed nothing at all.
+- The largest leak is almost always basis, and the smallest is almost always quantity. Quantity is still worth naming, because it is the only one that can never be reduced to zero.
+- Value at risk is estimated from recent volatility and recent correlation. It describes the market that has just happened, which is rarely the one approaching.
+- A stress test is not a tail of the VaR distribution. It is a different distribution, drawn from something that actually occurred, which is why it produces numbers several times larger.
+- A limit is not a forecast. It is a statement about how much of a mistake the firm can fund before it has to stop trading in order to pay margin.
+- Risk sits outside trading for one reason: marking a position and sizing a position cannot belong to the person who is paid on the answer.
+- A chosen risk has a price you can name. An inherited risk does not. Inherited risk is deleted or converted, never hedged harder — and the residue that will not delete belongs in the margin you quote rather than in the risk report.
+
+# Vocabulary
+
+| Term | What it means |
+|---|---|
+| **Hedge leakage** | The exposure a hedge leaves behind once flat price is removed, arriving through basis, timing, quality, currency, cross-hedge and quantity |
+| **Value at risk (VaR)** | The loss a book should not exceed on a stated fraction of days, computed from recent volatility and correlation |
+| **Stress test** | Revaluing a book under one specific named scenario drawn from something that actually happened, rather than from an estimated distribution |
+| **Position limit** | A cap on exposure in tonnes or lots, set by commodity and by delivery month |
+| **Loss limit** | A cap expressed as a VaR number or as a stress-scenario loss, breached when the book is too large for the balance sheet rather than when it is wrong |
+| **Concentration limit** | A cap on how much of a book may sit with one counterparty, one port or one origin — the limit most often left unwritten |
+| **Timing leak** | The loss caused by hedging one delivery month while the physical prices against another, equal to the spread between the two |
+| **Protein deficiency allowance** | The clause that discounts a meal or wheat invoice by a stated amount for each percentage point of protein below the contract minimum |
+| **Forward points** | The adjustment applied to a spot exchange rate to price a forward date, quoted in ten-thousandths and equal to the interest differential between the two currencies |
+| **Chosen risk** | An exposure the desk was paid to take, with a size, a price and an exit written into the plan |
+| **Inherited risk** | An exposure that arrived attached to a trade rather than being chosen, recognisable because nobody can name the price they were paid to take it |
+| **Independent risk function** | The reporting line that puts marking and position sizing outside the trading book, usually under the chief financial officer |
+
+# Quiz
+
+**Q1.** A merchant sells 25,000 t of 46 percent protein soybean meal CFR Rotterdam at €396.00/t for December arrival. EUR/USD is 1.1680 on the day of the sale and he does not sell the euros forward. He budgets freight at $36.00/t and finance, insurance and outturn at $5.50/t, and plans to buy FOB New Orleans at the December board plus $9.00 per short ton. The December board is $364.00. He hedges by buying December soybean meal futures, 100 short tons a lot, rounded to the nearest whole lot. Six weeks later he buys the physical: the December board has fallen to $341.50 and he pays the board plus $16.00. The cargo outturns at 45.5 percent protein, and his sales contract discounts $6.00/t for each full percentage point of deficiency, pro rata. EUR/USD is 1.1610 when he is paid. Freight and the other costs land on budget. What is his realised dollar margin on the cargo?
+
+**Q2.** A risk system reports a one-day 95 percent VaR of $39,480 on a book that is long 2,000,000 bu of corn basis. Scaling by the square root of time, what does that imply over 21 trading days?
+
+**Q3.** A wheat position sheet nets to zero across the whole book. March shows long 400,000 bu of physical with no paper against it, and May shows short 80 lots. What position does the desk actually own?
+
+**Q4.** A farmer delivers 50,000 bu of corn on a basis contract at December minus 35 and leaves the futures price open until February. December corn then rallies $1.00. Who pays cash that week, and how much?
+
+**Q5.** Conversion drill. Matif December milling wheat is offered at €244.50/t FOB Rouen with EUR/USD at 1.1680. Freight Rouen to Casablanca is $17.50/t. A buyer bids $305.00/t CFR for 35,000 t. What is the margin on the cargo in dollars?
+
+---
+---
+---
+
+# SOLUTIONS (spoilers)
+
+**A1.** Start with the quantity, because the hedge size falls out of it and so does the trap.
+
+A metric tonne is 1.102311 short tons, so 25,000 t × 1.102311 = 27,557.775 short tons. At 100 short tons a lot that is 275.58 lots, which rounds to **276 lots**, or 27,600 short tons. He is therefore long 42.225 short tons of futures with no physical behind them.
+
+The planned margin:
+
+| Line | Working | Amount |
+|---|---|---|
+| Sale | €396.00 × 1.1680 = $462.528/t × 25,000 | $11,563,200 |
+| Purchase | $373.00 × 27,557.775 short tons | −$10,279,050 |
+| Freight and costs | $41.50/t × 25,000 | −$1,037,500 |
+| **Planned margin** | | **$246,650** |
+
+What actually happened:
+
+| Line | Working | Amount |
+|---|---|---|
+| Sale | €396.00 × 1.1610 = $459.756/t × 25,000 | $11,493,900 |
+| Protein discount | 0.5 point × $6.00 = $3.00/t × 25,000 | −$75,000 |
+| Purchase | $357.50 × 27,557.775 short tons | −$9,851,905 |
+| Freight and costs | $41.50/t × 25,000 | −$1,037,500 |
+| Futures | long 27,600 short tons, 364.00 to 341.50 | −$621,000 |
+| **Realised margin** | | **−$91,505** |
+
+So a fully hedged cargo planned at +$246,650 lands at **−$91,505**, a swing of $338,155.
+
+The bridge names where it went, and this is the part worth memorising:
+
+| Leak | Working | Cost |
+|---|---|---|
+| Basis | $7.00 × 27,557.775 short tons | $192,904 |
+| Quality | $3.00/t × 25,000 | $75,000 |
+| Currency | 0.0070 × €9,900,000 | $69,300 |
+| Quantity | 42.225 short tons × $22.50 | $950 |
+| **Total** | | **$338,154** |
+
+$246,650 less $338,154 is −$91,504, which is the realised figure to the dollar once rounding is allowed for.
+
+The trap is the $22.50 fall in the board. It is the biggest number in the question and it is worth almost nothing: the short physical and the long futures offset each other exactly, apart from 42 short tons of over-hedge. Anyone who tried to compute a flat-price P&L was answering a different question. The line that did the damage is basis, at more than half the total, and basis was the one thing the merchant was actually being paid to judge.
+
+**A2.** $39,480 × √21 = $39,480 × 4.5826 = **$180,920**.
+
+That is the mechanical answer, and the reason to know it is so that you can see what it assumes. Square-root-of-time scaling requires the daily moves to be independent of one another. Basis is the opposite of independent: it trends, because the thing driving it — a vessel queue, a farmer who has stopped selling, a plant that is short — takes weeks to resolve and moves the price the same way every day while it does. Twenty-one correlated days do not scale like twenty-one coin flips.
+
+Nor can the position be exited inside the horizon. There is no screen for central Illinois corn basis. The honest measure of the liquidation risk is days to liquidate, not a one-day number multiplied up.
+
+So $180,920 is the number the model produces, and the reason to treat it as a floor is that both of its assumptions fail in the direction that understates the loss.
+
+**A3.** 400,000 bu is 80 lots at 5,000 bushels a lot. The desk is long 80 lots' worth of physical March against short 80 lots of May.
+
+That is an **80-lot March/May bull spread** — long the nearby, short the deferred — and nobody decided to own it. It is invisible on the total, which nets to zero, and invisible on any line of the sheet that is not split by futures month.
+
+The cost is bounded. A long March against a short May loses as the carry widens, and the carry cannot widen past full carry, because beyond that anybody with a bin can buy March, store, and deliver against May for free. With March/May full carry at 22.25c and the spread at 14c, there is 8.25c of room, which on 400,000 bu is $33,000.
+
+The mirror image is the one to be frightened of. Short the near and long the deferred has no such ceiling, because an inversion can go as far as the market needs it to go.
+
+**A4.** The elevator pays, and it pays $50,000 that week.
+
+50,000 bu is 10 lots. On a basis contract the differential is fixed and the futures price is left open, so the elevator has bought grain it must hedge: it sells 10 December lots. When December rallies $1.00 the short hedge loses $1.00 × 50,000 bu = $50,000, settled as variation margin in cash, the same day.
+
+The farmer pays nothing. He has not fixed, so his side of the contract has no mark.
+
+The elevator gets the $50,000 back the moment the farmer fixes at the higher price — assuming the farmer fixes. That assumption is the entire exposure. The elevator is not carrying price risk here, it is carrying a funding cost and a credit risk, and the two are worth naming separately because they fail differently. The funding cost is certain and budgetable. The credit risk is binary: if the farmer cannot or will not perform, the $50,000 is simply gone, and it was never written down anywhere as a position.
+
+**A5.** €244.50 × 1.1680 = $285.576/t. Add freight of $17.50 and the delivered cost is $303.076/t. Against a bid of $305.00 the margin is $1.924/t, and on 35,000 t that is **$67,340**.
+
+The fast check: at 1.1680, going euros to dollars means adding about 17 percent. €244.50 plus a sixth is roughly $285, which is close enough to tell you in two seconds whether the bid is anywhere near workable.
+
+# The written edition
+
+## The hedge that leaks
+
+Every risk course teaches the hedge that works. Desks learn from the hedge that leaks.
+
+Here is the structure of what follows. A merchant sells a meal cargo, hedges the flat price to the nearest lot, and loses $301,445. Not one cent of that loss comes from the board. All of it comes from six channels that stay open after the futures are on, and each of those channels has a name, a size and, in most cases, a fix.
+
+### The cargo
+
+A merchant sells 27,000 t of 46 percent protein soybean meal CFR Rotterdam for November arrival at €402.00/t. EUR/USD is 1.1750, so the sale is worth $472.35/t. He budgets freight at $38.00/t and finance, insurance and outturn at $6.00/t.
+
+He does not own the meal yet. He is short 27,000 t of physical, and he covers the flat price by buying futures.
+
+The quantity arithmetic is the first place a leak appears, and it appears before anything has gone wrong. A metric tonne is 1.102311 short tons, so the cargo is 29,762.397 short tons. The CBOT meal contract is 100 short tons, which makes the cargo 297.62 lots. He buys **298**.
+
+He plans to buy the physical at the October board plus $8.00 per short ton, with the board at $368.70.
+
+| Line | Working | $/t |
+|---|---|---|
+| Sale CFR Rotterdam | €402.00 × 1.1750 | 472.35 |
+| Freight | | −38.00 |
+| Finance, insurance, outturn | | −6.00 |
+| Planned FOB purchase | $376.70 × 1.102311 | −415.24 |
+| **Planned margin** | | **13.11** |
+
+On 27,000 t that is **$353,954**. It is a thin, ordinary, entirely respectable merchant margin, and the flat price is fully covered.
+
+### Six leaks, one cargo
+
+**Basis.** Gulf meal basis does not sit still. Crush margins are strong, domestic feeders want October meal, and he ends up paying the board plus $14.00 instead of plus $8.00. Six dollars a short ton on 29,762.397 short tons is **$178,574**. This is the leak the merchant is paid to judge, and it is the one that hurts most.
+
+**Timing.** He hedged October. The physical he buys prices against December, and December sits $3.00 above October. Three dollars on 29,762.397 short tons is **$89,287**. Nobody made a bad call. The hedge was in the wrong month, which is a bookkeeping decision with a price tag.
+
+**Quality.** He sold 46 percent protein and the cargo outturns at 45.2 percent. His sales contract discounts $6.00/t for each full percentage point of deficiency, so eight tenths of a point is $4.80/t, or **$129,600**. Quality risk is not an abstraction. It is a table in a contract, and somebody has to read it before the cargo is loaded rather than after it is discharged.
+
+**Currency.** The sale is in euros: 27,000 × €402.00 = €10,854,000. He did not sell them forward. EUR/USD goes from 1.1750 to 1.1600, which costs **$162,810**.
+
+**Cross-hedge.** He covered his freight with an FFA on a Baltic route that is not his route. His actual freight rose $5.50/t; the index he owned rose $2.00. The difference is $3.50/t, or **$94,500**. The hedge worked perfectly. It worked on somebody else's voyage.
+
+**Quantity.** 298 lots is 29,800 short tons against a cargo of 29,762.397, which leaves 37.603 short tons of futures with nothing behind them. The board fell $16.70, so that costs **$628** — the smallest leak on the list, and the only one that can never be zero.
+
+```chart
+{"type":"waterfall","unit":"$ on the cargo","title":"How $353,954 became a loss",
+ "caption":"Every bar after the first is a risk the merchant did not choose and was not paid for. The board fell $16.70 over the life of the trade and contributed nothing, because that is the one risk he did hedge.",
+ "source":"Worked example, episode 22: 27,000 t of 46 percent protein soybean meal, FOB New Orleans to CFR Rotterdam.",
+ "steps":[{"label":"Planned","value":353954,"kind":"base"},
+          {"label":"Basis","value":-178574},
+          {"label":"Timing","value":-89287},
+          {"label":"Quality","value":-129600},
+          {"label":"Currency","value":-162810},
+          {"label":"Freight FFA","value":-94500},
+          {"label":"Quantity","value":-628},
+          {"label":"Realised","kind":"total"}]}
+```
+
+Total leakage is $655,399 against a planned margin of $353,954, so the cargo lands at **−$301,445**, or −$11.16/t where the plan said +$13.11/t. The board moved $16.70 and delivered exactly what it was supposed to deliver: nothing.
+
+### The conversation that was not had
+
+The currency leak is the easiest of the six to remove, and removing it takes about twenty seconds.
+
+> **TRADER:** Rotterdam's done. Twenty-seven thousand at four oh two.
+> **TREASURY:** Euros for when?
+> **TRADER:** Payment's forty days after outturn. Call it end November.
+> **TREASURY:** Spot's one seventeen fifty. End November I make you minus eighteen points.
+> **TRADER:** So one seventeen thirty-two.
+> **TREASURY:** Ten point eight five million euros at one seventeen thirty-two. Or you stay long euros until Christmas.
+
+"Minus eighteen points" is not a price. **Forward points** are an adjustment to spot, quoted four decimal places out, and they are the interest differential between the two currencies rather than anybody's view on the currency pair.
+
+That distinction matters for how the leak is scored. Selling forward at 1.1732 would have cost €10,854,000 × 0.0018 = $19,537 of forward points. That is a financing cost and it belongs in the margin. The remaining $143,273 was avoidable, and it is the part that should appear in a post-mortem.
+
+## The machinery meant to catch it
+
+### Three limits
+
+Three limits sit over a physical desk, and they constrain different things.
+
+| Limit | Expressed as | What it caps |
+|---|---|---|
+| Position | tonnes and lots, by commodity and by month | how wrong one view can be |
+| Loss | VaR and a stress-scenario loss | how much the balance sheet can fund |
+| Liquidity | days to liquidate | how long it takes to stop |
+
+There is a fourth that is often unwritten and matters more than any of them in a bad week: the **concentration limit**, on how much of the book may sit with one counterparty, one port or one origin.
+
+### Why value at risk flatters a physical book
+
+Take a book long 2,000,000 bu of corn basis, with daily basis volatility of 1.2 c/bu. At 95 percent over one day, that is 1.645 standard deviations, or 1.97 c/bu, which on 2,000,000 bu is $39,480.
+
+The sentence the system produces is: on nineteen days out of twenty you lose less than forty thousand dollars. Three things are wrong with it.
+
+First, it is computed on marks, and a basis position is marked to an assessment rather than to a trade. The input is an informed opinion, and the output inherits its error.
+
+Second, it assumes the position can be exited inside the horizon. There is no screen for central Illinois corn basis.
+
+Third, the covariance matrix behind it is estimated on quiet days. Correlations are highest when nothing is happening and they break on precisely the event that resolves the thesis — which is to say, they are reliable everywhere except where they are needed.
+
+### The stress number
+
+Scale the daily figure to a month and the arithmetic is $39,480 × √21 = $180,920. Now stress it instead. Gulf basis has moved 25 c/bu in three weeks, and not as a tail event. On 2,000,000 bu that is $500,000.
+
+```chart
+{"type":"bar","unit":"$ on 2 m bu of corn basis","title":"The model and the event",
+ "caption":"The stress number is not the tail of the VaR distribution. It is a different distribution, drawn from something that actually happened, and it is 12.7 times the daily figure the system prints.",
+ "source":"Worked example, episode 22: 2,000,000 bu of corn basis at 1.2 c/bu daily volatility, against a 25 c/bu three-week basis move.",
+ "x":["1-day 95% VaR","Scaled to 21 days","25c basis stress"],
+ "series":[{"name":"Loss","values":[39480,180920,500000]}]}
+```
+
+Five hundred thousand dollars is 2.8 times the monthly VaR number and 12.7 times the daily one. Value at risk describes the market that has already happened. A stress test describes the market that is coming.
+
+## Who says no
+
+The person who says no does not sit on the desk. Risk reports to the chief financial officer, not to the head of trading, and the reason is structural rather than cultural: the person who owns the P&L cannot also be the person who marks it and sizes it.
+
+It follows that a limit is not a forecast. Nobody sets a 60,000 t position limit because they believe 60,001 t is where the market turns. A limit is a statement about how much of a mistake the firm can fund before it has to stop trading in order to pay margin — which is why limits are set against the balance sheet and the credit lines, and why a trader who is right can still be told to reduce.
+
+## The risk you chose and the risk you inherited
+
+This is the distinction that makes the rest of it usable.
+
+A **chosen risk** is one the desk was paid to take. It has a size, an exit, and a price written into the plan. The basis was chosen, and $13.11/t was the price of taking it.
+
+An **inherited risk** arrived attached to something else. Nobody decided to own €10,854,000. Nobody decided to be short protein. Nobody chose to hedge in a month the physical would not price against.
+
+The test is a single question: *can you name the price you were paid to take it?* If not, it is inherited.
+
+And an inherited risk is not managed by hedging it harder. It is either converted into a chosen one or deleted outright. Sell the euros the day you sell the cargo. Buy the protein you sold. Hedge the month you price in. Each of those is one instruction, given once, at the start.
+
+Some of it will not delete. Lot rounding never reaches zero, and a freight index is never your voyage. Those are irreducible, and the correct place for an irreducible risk is the margin you quote — not the risk report, where it will be measured every night and fixed never.
+
+That is the whole difference between a desk that loses money and learns something, and a desk that loses money twice.

package/ep22.script.txt

@@ -0,0 +1,86 @@
+A soybean meal cargo, hedged to the last lot. Not one cent of flat price left open. ||| 0.4
+It lost three hundred thousand dollars. ||| 0.6
+This is Soft Commodity Trading, episode twenty-two. Risk management, and every reason a hedge leaks. ||| 0.7
+First, Thursday's tape. ||| 0.4
+December corn settled five thirty and a half, down three and three quarter cents. November beans thirteen nineteen and three quarters, down three quarters. December Chicago wheat seven twenty-seven, down three and three quarters. ||| 0.5
+The interesting move was inside the soybean complex. October meal settled three sixty-eight seventy, up seven dollars eighty. October bean oil sixty-eight sixty-eight, down fifty-one points. ||| 0.5
+Meal has added about seven percent in six sessions. Oil has lost two. The crush itself barely moved. The money changed seats inside it. ||| 0.6
+Export sales for the week to the tenth. Beans one point seven million tonnes, mostly China. Corn one point zero million, a three-week low. ||| 0.5
+Then the policy, and two channels were running at once. ||| 0.4
+Ukraine has floated sitting down with Russia to restart grain exports from both countries. Russia called the proposal impractical. Shipping stays very limited at both origins. ||| 0.4
+Chicago wheat fell on the day that headline printed. ||| 0.5
+Here is why. A corridor does not reopen on a communique. It reopens when underwriters reprice war risk on hulls and cargo. Until the insurance moves, an agreement moves no tonnes. ||| 0.6
+The channel the board did pay for was the other one. China bought about a million tonnes of American beans last week, roughly halfway to a twenty-five million tonne a year commitment, with a ten percent tariff on American farm goods on the table at a meeting on the twenty-fourth. ||| 0.6
+And France Agri Mer cut the French soft wheat export forecast outside the E U to six point three million tonnes from seven. ||| 0.6
+Hold on to that meal number, because today's cargo is a meal cargo. ||| 0.7
+Every risk course teaches the hedge that works. Desks learn from the hedge that leaks. ||| 0.6
+So take one cargo. Twenty-seven thousand tonnes of forty-six percent protein soybean meal, sold C F R Rotterdam for November arrival at four hundred and two euros a tonne. ||| 0.5
+At an exchange rate of one seventeen fifty, that is four seventy-two thirty-five a tonne in dollars. Freight thirty-eight dollars. Finance, insurance and outturn, six. ||| 0.5
+He does not own the meal yet. He is short twenty-seven thousand tonnes. ||| 0.4
+So he buys futures. Twenty-seven thousand tonnes is twenty-nine thousand seven hundred sixty-two short tons. The meal contract is a hundred short tons. That is two hundred ninety-seven point six lots, so he buys two hundred ninety-eight. ||| 0.5
+He plans to buy the physical at the October board plus eight dollars a short ton, with the board at three sixty-eight seventy. ||| 0.4
+Planned margin, thirteen dollars eleven a tonne. Three hundred fifty-four thousand dollars on the cargo. ||| 0.6
+Flat price is covered. Now watch six things go wrong that have nothing to do with it. ||| 0.7
+One. Basis. ||| 0.3
+Gulf meal basis does not sit still. Crush margins are strong, feeders want October meal, and he pays the board plus fourteen instead of plus eight. ||| 0.4
+Six dollars a short ton, on twenty-nine thousand seven hundred sixty-two short tons. A hundred seventy-eight thousand five hundred seventy-four dollars. ||| 0.7
+Two. Timing. ||| 0.3
+He hedged October. The physical he buys prices off December, and December sits three dollars above October. ||| 0.4
+Three dollars a short ton. Eighty-nine thousand two hundred eighty-seven dollars. Nobody made a bad call. The hedge was in the wrong month. ||| 0.7
+Three. Quality. ||| 0.3
+He sold forty-six percent protein. The cargo outturns at forty-five point two. ||| 0.35
+His sales contract discounts six dollars a tonne for each full point of protein deficiency. Eight tenths of a point is four dollars eighty a tonne. A hundred twenty-nine thousand six hundred dollars. ||| 0.7
+Four. Currency. ||| 0.3
+The sale is in euros. Ten point eight five million of them. He did not sell them forward. ||| 0.35
+Euro dollar goes from one seventeen fifty to one sixteen. A hundred sixty-two thousand eight hundred ten dollars, and the conversation that would have prevented all of it takes about twenty seconds. ||| 0.6
+TRADER: Rotterdam's done. Twenty-seven thousand at four oh two. ||| 0.25
+TREASURY: Euros for when? ||| 0.25
+TRADER: Payment's forty days after outturn. Call it end November. ||| 0.25
+TREASURY: Spot's one seventeen fifty. End November I make you minus eighteen points. ||| 0.25
+TRADER: So one seventeen thirty-two. ||| 0.25
+TREASURY: Ten point eight five million euros at one seventeen thirty-two. Or you stay long euros until Christmas. ||| 0.6
+Minus eighteen points is not a price. It is an adjustment to spot, four decimal places out, the cost of carry between two currencies. ||| 0.5
+Selling forward would have cost nineteen thousand five hundred thirty-seven dollars of those points. The other hundred forty-three thousand two hundred seventy-three was avoidable. ||| 0.7
+Five. Cross-hedge. ||| 0.3
+He covered freight with an F F A on a Baltic route that is not his route. His actual freight rose five dollars fifty a tonne. The index he owned rose two. ||| 0.4
+Three dollars fifty a tonne against him. Ninety-four thousand five hundred dollars. The hedge worked perfectly. It worked on somebody else's voyage. ||| 0.7
+Six. Quantity. ||| 0.3
+Two hundred ninety-eight lots is twenty-nine thousand eight hundred short tons, against a cargo of twenty-nine thousand seven hundred sixty-two. Thirty-seven short tons of futures with nothing behind them. ||| 0.4
+The board fell sixteen dollars seventy. Six hundred twenty-eight dollars. ||| 0.5
+The smallest leak on the list, and the only one that is never zero. ||| 0.7
+Add them up. Six hundred fifty-five thousand four hundred dollars of leakage, against a planned margin of three hundred fifty-four thousand. ||| 0.5
+The trade lands at minus three hundred one thousand four hundred forty-five dollars. Minus eleven dollars sixteen a tonne, where the plan said plus thirteen eleven. ||| 0.5
+And the board? Fully offset. Sixteen dollars seventy of flat price did nothing at all. ||| 0.8
+Now the machinery that is meant to catch all that. ||| 0.5
+Three limits sit over a physical desk. Position limits in tonnes and lots, by commodity and by month. Loss limits, as value at risk and as a stress number. Liquidity limits, as days to liquidate. ||| 0.6
+Value at risk is the one people quote, and the one to distrust. ||| 0.5
+Take a book long two million bushels of corn basis. Daily basis volatility, one point two cents. ||| 0.4
+Ninety-five percent over one day is one point six four five standard deviations. One point nine seven cents. Thirty-nine thousand four hundred eighty dollars. ||| 0.5
+So the system says: on nineteen days out of twenty, you lose less than forty thousand dollars. ||| 0.5
+Three things are wrong with that sentence. ||| 0.6
+First, it is computed on marks, and a basis position is marked to an assessment, not a trade. The input is an opinion. ||| 0.5
+Second, it assumes you can get out inside the horizon. There is no screen for central Illinois corn basis. ||| 0.5
+Third, its covariance matrix is estimated on quiet days, and correlations break on precisely the event that resolves your thesis. ||| 0.7
+Scale it to a month and the honest arithmetic is square root of twenty-one. Four point five eight. A hundred eighty thousand nine hundred twenty dollars. ||| 0.5
+Now stress it instead. Gulf basis has moved twenty-five cents in three weeks before, and not as a tail event. On two million bushels, five hundred thousand dollars. ||| 0.5
+Two point eight times the monthly value at risk number. Twelve and a half times the daily one. ||| 0.6
+Value at risk describes the market you have already had. A stress test describes the market that is coming for you. ||| 0.8
+Which is why the person who says no does not sit on the desk. ||| 0.5
+Risk reports to the chief financial officer, not to the head of trading. ||| 0.4
+The person who owns the P and L cannot also be the person who marks it and sizes it. ||| 0.5
+And a limit is not a forecast. A limit is a statement about how much of a mistake the firm can fund before it has to stop trading in order to pay margin. ||| 0.8
+There is a risk you chose, and there is a risk you inherited. ||| 0.5
+A chosen risk is one you were paid to take. It has a size and it has an exit. The basis was chosen, and thirteen dollars eleven a tonne was the price of taking it. ||| 0.5
+An inherited risk arrived attached to something else. Nobody decided to own ten point eight five million euros. Nobody decided to be short protein. Nobody chose the wrong contract month. ||| 0.5
+The test is one question. Can you name the price you were paid to take it? ||| 0.5
+If you cannot, it is inherited. ||| 0.7
+You do not manage an inherited risk by hedging it harder. You convert it into a chosen one, or you delete it. Sell the euros the day you sell the cargo. Buy the protein you sold. Hedge the month you price in. ||| 0.6
+Some of it will not delete. The lot rounding never reaches zero. The freight index is never your voyage. ||| 0.4
+Those belong in the margin you quote, not in the risk report. ||| 0.8
+So, what to remember. ||| 0.4
+A hedge removes flat price and nothing else. Six channels stay open: basis, timing, quality, currency, cross-hedge, quantity. ||| 0.5
+Value at risk measures the past. The stress number is the one worth arguing about. ||| 0.5
+Risk sits outside trading because marking and sizing cannot belong to the person who is paid on the answer. ||| 0.5
+And every loss is either a price you agreed to pay, or a risk you never noticed accepting. ||| 0.8
+Next episode: trade finance, contracts and counterparty risk. Letters of credit, G A F T A terms, washouts, and what happens when the other side does not perform. ||| 0.5
+Four questions in the notes today. The first one is a whole cargo, fully hedged, and it loses money. Work out how much. ||| 0.7

package/feed.xml

@@ -18,6 +18,18 @@
       <title>Soft Commodity Trading</title>
       <link>https://storage.googleapis.com/podcast-audio-2647223968/index.html</link>
     </image>
+    <item>
+      <title>Ep 22 — Risk Management and Why Hedges Are Never Perfect</title>
+      <link>https://storage.googleapis.com/podcast-audio-2647223968/commodity-desk-daily/ep22.html</link>
+      <description><![CDATA[<p>A soybean meal cargo hedged to the last lot still loses three hundred thousand dollars, and the six channels a hedge leaks through are named and priced one by one. Then value at risk against a stress test, who says no on a trading floor, and the difference between a risk you chose and a risk you inherited.</p><p><a href="https://storage.googleapis.com/podcast-audio-2647223968/commodity-desk-daily/ep22.html">Read this episode, with the charts, the glossary and the quiz &rarr;</a></p>]]></description>
+      <itunes:summary>A soybean meal cargo hedged to the last lot still loses three hundred thousand dollars, and the six channels a hedge leaks through are named and priced one by one. Then value at risk against a stress test, who says no on a trading floor, and the difference between a risk you chose and a risk you inherited.
+
+Read this episode: https://storage.googleapis.com/podcast-audio-2647223968/commodity-desk-daily/ep22.html</itunes:summary>
+      <enclosure url="https://storage.googleapis.com/podcast-audio-2647223968/commodity-desk-daily/ep22.mp3" length="0" type="audio/mpeg"/>
+      <guid isPermaLink="false">https://storage.googleapis.com/podcast-audio-2647223968/commodity-desk-daily/ep22.mp3</guid>
+      <pubDate>Fri, 18 Sep 2026 05:00:00 GMT</pubDate>
+      <itunes:duration>0</itunes:duration>
+    </item>
     <item>
       <title>Ep 21 — The Book and P&amp;L Attribution</title>
       <link>https://storage.googleapis.com/podcast-audio-2647223968/commodity-desk-daily/ep21.html</link>

package/glossary.md

@@ -51,10 +51,12 @@ Units, conventions and desk expressions, accumulated as the show introduces them
 - **certified stock** — coffee sampled, graded and stamped as deliverable against the futures contract and held in an exchange-licensed warehouse, the deliverable float rather than world inventory  _(ep 12)_
 - **CFR** — cost and freight, the seller pays the voyage to a named destination but risk still passes at loading  _(ep 4)_
 - **charter party** — the contract hiring the vessel, between charterer and shipowner  _(ep 4)_
+- **chosen risk** — an exposure a desk was paid to take, with a size, a price and an exit written into the plan  _(ep 22)_
 - **CIF** — cost insurance and freight, CFR plus the seller buys the marine insurance the buyer would claim on  _(ep 4)_
 - **citrus greening** — huanglongbing, the bacterial disease that permanently reduces an infected orange tree's yield and cannot be cured  _(ep 15)_
 - **Coffee C (KC)** — the ICE arabica futures contract, 37,500 lb quoted in US cents per pound with a 0.05 cent tick worth 18.75 dollars  _(ep 12)_
 - **collar (fence)** — buying a put and selling a call against the same position so the price is bounded on both sides, the standard hedging structure around unpriced physical  _(ep 17)_
+- **concentration limit** — a cap on how much of a book may sit with one counterparty, one port or one origin, the limit most often left unwritten  _(ep 22)_
 - **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  _(ep 16)_
 - **conversion cost** — the variable cost of turning beans into products, gas, power, hexane, labour and maintenance, typically 35 to 50 cents a bushel at a modern plant  _(ep 8)_
 - **conversion factors** — 36.7 bushels per tonne for wheat and beans and 39.4 for corn, so cents per bushel times 0.367 or 0.394 gives dollars per tonne  _(ep 1)_
@@ -122,6 +124,7 @@ Units, conventions and desk expressions, accumulated as the show introduces them
 - **flat price exposure** — outright price risk, removed deliberately by hedging so only the basis remains  _(ep 2)_
 - **FOB** — free on board, the cargo is priced at the load port with the buyer taking it from the ship's rail  _(ep 2)_
 - **forward freight agreement (FFA)** — a cash-settled swap on a Baltic index route or basket over a calendar month, the only liquid way to hedge freight  _(ep 10)_
+- **forward points** — the adjustment applied to a spot exchange rate to price a forward date, quoted in ten-thousandths and equal to the interest differential between the two currencies  _(ep 22)_
 - **front month** — the nearest actively traded contract month, where liquidity is deepest  _(ep 3)_
 - **full carry** — storage plus interest per month of holding grain, the practical ceiling on a carry spread  _(ep 3)_
 - **FX leg** — the currency exposure that arrives unbidden in an inter-exchange spread whose two legs settle in different currencies  _(ep 16)_
@@ -137,6 +140,7 @@ Units, conventions and desk expressions, accumulated as the show introduces them
 - **harvest basis** — the seasonal low in the cash-minus-futures spread, set when a year of crop arrives in six weeks into a pipe sized to move it over twelve months  _(ep 11)_
 - **harvest run** — the six to eight weeks in which a full year of crop arrives at facilities sized to ship it over twelve months  _(ep 19)_
 - **harvested acres** — area actually cut for grain, roughly 8 million acres below planted for US corn, and the denominator that yield is quoted against  _(ep 6)_
+- **hedge leakage** — the exposure a hedge leaves behind once flat price is removed, arriving through basis, timing, quality, currency, cross-hedge and quantity  _(ep 22)_
 - **hedge-to-arrive** — the mirror of a basis contract, fixing the futures price now and leaving the differential to be agreed later  _(ep 19)_
 - **hexane** — the solvent used to extract the last of the oil from the flaked bean, and a real line in the conversion cost  _(ep 8)_
 - **hit** — your bid was taken by a seller  _(ep 1)_
@@ -147,7 +151,9 @@ Units, conventions and desk expressions, accumulated as the show introduces them
 - **import premium** — the amount a destination market pays above the exporting market's replacement value, which is what draws cargoes towards that destination rather than another  _(ep 20)_
 - **inclusion rate** — the share of a single ingredient in a feed ration, capped by nutrition and by anti-nutritional factors  _(ep 6)_
 - **Incoterms** — the standard three-letter trade terms that allocate cost and risk between buyer and seller  _(ep 4)_
+- **independent risk function** — the reporting line that puts marking and position sizing outside the trading book, usually under the chief financial officer, because the owner of a profit cannot also be its scorekeeper  _(ep 22)_
 - **indication** — a guide price that is not firm  _(ep 1)_
+- **inherited risk** — an exposure that arrived attached to a trade rather than being chosen, recognisable because nobody can name the price they were paid to take it  _(ep 22)_
 - **initial margin** — the deposit the clearing house takes per lot when a position is opened  _(ep 3)_
 - **inter-exchange spread** — the price gap between two exchanges pricing related but different goods, such as Kansas City over Chicago  _(ep 5)_
 - **inverse (backwardation)** — a curve with nearer months above later ones, the market pays a premium for immediate delivery  _(ep 3)_
@@ -168,6 +174,7 @@ Units, conventions and desk expressions, accumulated as the show introduces them
 - **long the basis** — owning physical hedged with futures, so the position gains when the differential strengthens and is indifferent to the board  _(ep 19)_
 - **long the spread** — holding the nearby month against a short in the deferred, which loses as the carry widens and whose loss is bounded by full carry  _(ep 21)_
 - **long ton** — 2,240 lb, the imperial weight unit the sugar No. 11 contract is still sized in at 50 long tons a lot  _(ep 14)_
+- **loss limit** — a cap expressed as a value-at-risk number or as a stress-scenario loss, breached when a book is too large for the balance sheet rather than when it is wrong  _(ep 22)_
 - **lot** — one futures contract, 5,000 bushels for Chicago grains, the unit desks count positions in  _(ep 1)_
 - **managed money** — speculative funds reported as non-commercial in exchange positioning data, which trade direction rather than physical  _(ep 13)_
 - **margin financing** — the cost of funding variation margin paid out on a losing futures leg while the offsetting gain on the physical is still unrealised  _(ep 21)_
@@ -218,12 +225,14 @@ Units, conventions and desk expressions, accumulated as the show introduces them
 - **point (softs)** — one hundredth of a cent per pound, so up 300 points means up 3 cents  _(ep 1)_
 - **polarisation (pol)** — the sucrose purity of a sugar measured by the rotation of polarised light and expressed in degrees, the basis on which raw sugar is priced and settled  _(ep 14)_
 - **pollination** — the roughly one-week corn window in mid-July in the northern hemisphere after which the ear count is fixed and no forecast can change it  _(ep 6)_
+- **position limit** — a cap on exposure in tonnes or lots, set by commodity and by delivery month  _(ep 22)_
 - **position sheet** — a desk's record of net exposure by futures month and by location, kept in lots with long positive and short negative, in which physical and paper appear as the same row  _(ep 21)_
 - **posted bid** — the price an elevator displays to growers for immediate delivery, quoted as a differential to a named futures month and used to manage the delivery queue as much as to set a price  _(ep 19)_
 - **price assessment** — a published daily price built by surveying brokers and exporters, used where no futures contract exists  _(ep 5)_
 - **price-later deadline** — the date by which an unpriced farmer contract must be fixed, after which the buyer prices it at the market  _(ep 19)_
 - **price-to-be-fixed (PTBF)** — a physical contract where quantity, quality, shipment and differential are agreed now and the futures price is set later  _(ep 13)_
 - **prompt** — the nearby month or shipment window, ready to move now  _(ep 1)_
+- **protein deficiency allowance** — the contract clause that discounts a meal or wheat invoice by a stated amount for each percentage point of protein below the contract minimum  _(ep 22)_
 - **protein spec** — the contractual protein percentage that turns the word wheat into a price  _(ep 5)_
 - **putting on the crush** — buying bean futures and selling meal and oil futures against them in a 10-11-9 lot ratio, which fixes the processing margin  _(ep 8)_
 - **quality basis** — the spread between the grade you own and the grade the futures contract delivers  _(ep 5)_
@@ -271,6 +280,7 @@ Units, conventions and desk expressions, accumulated as the show introduces them
 - **statement of facts** — the port log of events both sides use to fight laytime claims  _(ep 4)_
 - **stocks-to-use** — ending stocks divided by total use, the market's tension gauge  _(ep 2)_
 - **storage tariff** — the published charge for commercial storage, quoted in cents per bushel per month or per day, or in dollars per tonne per month  _(ep 11)_
+- **stress test** — revaluing a book under one specific named scenario drawn from something that actually occurred, rather than from an estimated distribution  _(ep 22)_
 - **substitution spread** — the price gap between two competing vegetable oils, which sets the point at which a refiner reformulates from one to the other  _(ep 9)_
 - **sugar mix** — the share of a mill's recoverable sugars turned into sugar rather than ethanol, bounded above by the plant's crystallisation capacity  _(ep 14)_
 - **Supramax** — a dry bulk vessel of roughly 50,000 to 60,000 dwt, normally carrying its own cranes, working minor bulks and shorter legs  _(ep 10)_
@@ -285,12 +295,14 @@ Units, conventions and desk expressions, accumulated as the show introduces them
 - **ticker** — the short screen code a contract is spoken by, ZW wheat, ZC corn, ZS soybeans, ZM meal, ZL oil, KC coffee, SB sugar, CT cotton  _(ep 3)_
 - **time charter** — hiring the vessel itself for a period at a price in dollars per day, with the charterer taking speed, weather, port delay and usually fuel  _(ep 10)_
 - **time charter equivalent (TCE)** — a voyage's economics restated as dollars per day, which is how a shipowner compares one employment against another  _(ep 10)_
+- **timing leak** — the loss caused by hedging one delivery month while the physical prices against another, equal to the spread between the two  _(ep 22)_
 - **toll refining** — refining someone else's raws for a fee per tonne, which converts the white premium from a trading position into a fixed margin  _(ep 14)_
 - **total supply** — carry-in plus production plus imports, the top block of a balance sheet  _(ep 7)_
 - **total use** — domestic use plus exports, the bottom block of a balance sheet  _(ep 7)_
 - **trade average** — the published mean of analysts' pre-report estimates, and therefore the expectation already contained in the price  _(ep 7)_
 - **trend yield** — the yield a crop would produce on normal weather, the baseline against which a weather premium is measured  _(ep 6)_
 - **unfixed** — the state of a price-to-be-fixed contract whose futures leg has not yet been set, so the exposure is still outright  _(ep 15)_
+- **value at risk** — the loss a book should not exceed on a stated fraction of days, computed from recent volatility and correlation, which describes the market that has just happened rather than the one approaching  _(ep 22)_
 - **variable storage rate** — the CBOT rule that resets the daily storage charge on a wheat certificate according to where a nearby calendar spread sits as a percentage of full carry  _(ep 18)_
 - **variation margin** — the daily cash settlement of a position mark to market, paid the same day  _(ep 3)_
 - **VHP** — very high polarisation raw sugar of around 99 degrees, the grade Brazil exports and which trades at a premium to the No. 11 screen  _(ep 14)_

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@sdelsad/commodity-desk-daily",
-  "version": "1.0.67",
-  "description": "Soft Commodity Trading - Ep 21: The Book and P&L Attribution",
+  "version": "1.0.68",
+  "description": "Soft Commodity Trading - Ep 22: Risk Management and Why Hedges Are Never Perfect",
   "license": "CC-BY-4.0",
   "keywords": [
     "podcast",

package/ep21.md

@@ -1,307 +0,0 @@
-# Market pulse
-
-**Chicago wheat fell nine and a half cents by mid-morning and settled six and a half cents higher. The round trip was not the market changing its mind about peace — it was the market reading the small print on what had actually been announced.**
-
-| Contract | Last | Change |
-|---|---|---|
-| Dec corn (CBOT) | 535.75 c/bu | +2½ |
-| Nov soybeans (CBOT) | 1,318.75 c/bu | +14½ |
-| Dec Chicago SRW (CBOT) | 728.50 c/bu | +6½ |
-| Oct soybean meal (CBOT) | $360.10/short ton | +9.90 |
-| Oct soybean oil (CBOT) | 69.88 c/lb | +23 pts |
-
-On Monday the US president said Russia and Ukraine had agreed to stop attacks on each other's energy infrastructure. Both sides attached conditions. Wheat dropped roughly eighteen cents on the headline, then recovered most of it when fresh Russian strikes were reported around Odesa. By Tuesday's 8:30 CDT prints December soft red was at 712½, down 9½; it closed at 728½, up 6½ on the day.
-
-Beans and corn had their own bid. Crude oil strength spilled into both on Monday, and the meal market added $9.90 on Tuesday. The crop is arriving on schedule rather than early: corn 57 percent good to excellent, 86 percent dented, 42 percent mature and 8 percent harvested; soybeans 58 percent good to excellent with 6 percent cut. Spring wheat harvest is 93 percent done. Winter wheat planting is the one number behind, at 8 percent against a normal 12.
-
-```chart
-{"type":"line","unit":"c/bu","title":"Chicago wheat's truce round trip",
- "caption":"December soft red traded 712½ intraday on Tuesday, about eighteen cents below Monday's settle, and closed at 728½. The export loading data behind the move did not change at any point during it.",
- "source":"CBOT settlements, 4 to 15 September 2026. The 14 September level is Friday's settle less the reported 3¼-cent Monday change.",
- "x":["4 Sep","9 Sep","10 Sep","11 Sep","14 Sep","15 Sep"],
- "series":[{"name":"Dec Chicago SRW","values":[734.00,728.75,741.25,725.25,722.00,728.50]}]}
-```
-
-## The geopolitical read
-
-An energy truce is not a grain corridor. Refineries, power stations and oil berths are one target set. Grain terminals at Novorossiysk and the loading infrastructure at Odesa are another, and nothing announced on Monday covered the second one.
-
-That is why the eighteen cents came back. The board was not repricing the probability of disruption; it was discovering the **scope** of the sentence it had just read. The transmission chain runs the other way round from the one the headline implies: an energy-infrastructure truce lowers the risk to Russian refining and to oil loadings, which shows up in crude and in bunker costs, and only reaches grain through freight.
-
-The numbers that actually set the wheat balance did not move. Russia shipped 2.0 Mt of grain by sea in August, down 62 percent year on year. Ukraine's grain exports since 1 July are 4.34 Mt, down 24 percent, with wheat at 2.1 Mt, down 48 percent. Algeria, Pakistan and Saudi Arabia are covering their needs elsewhere. That is loading data, and loading data does not respond to a post.
-
-Sixteen cents, out and back, inside one session. A hedged book should not have noticed. Whether it noticed is a question you answer with a position sheet, not with a profit number — which is today's subject.
-
-# Key takeaways
-
-- A position sheet measures exposure, not inventory. A tonne in a bin and a tonne sold forward are the same row with opposite signs, and the row is a futures month rather than a shipment month.
-- A net position of zero is a claim, not a fact. It only becomes a fact once the book has been split by futures month **and** by location, because either split can hide a spread nobody decided to own.
-- The mistakes that survive are the ones that look like bookkeeping. A hedge added to the wrong month is one keystroke, and it is invisible on every line of the sheet except the one nobody prints.
-- A finished trade decomposes into six lines: flat price, basis, calendar spread, freight, currency and financing. Only one of them is a market view.
-- The bridge from planned margin to realised margin must close to the dollar. A gap that does not reconcile is not a rounding difference — it is a line you have not found.
-- In a hedged book, flat price is supposed to contribute nothing. When it contributes something, that is an unhedged quantity, not skill.
-- A good total can hide a broken process. A freight overrun is a purchasing failure and a basis miss is a trading call; they belong to different people and a single P&L number cannot tell you which one to fix.
-
-# Vocabulary
-
-| Term | What it means |
-|---|---|
-| **Position sheet** | A desk's record of net exposure by futures month and by location, kept in lots, with long positive and short negative, in which physical and paper appear as the same row |
-| **Net position** | Physical long, less physical short, plus futures, for one month and one location — the only number that actually describes exposure |
-| **Square** | Desk shorthand for a net position of zero in a given month or location |
-| **Open position** | A month or location where physical and futures do not offset, whether or not anybody decided to have one |
-| **Mark to market** | Repricing every open line at the day's settlement, physical included, so a book's value moves daily on grain nobody has yet agreed to buy |
-| **Reconciliation** | The morning discipline of making the trader's sheet, the back office's and the risk system's agree — or of explaining why they do not, before the market opens |
-| **Long the spread** | Holding the nearby month against a short in the deferred, which loses as the carry widens and whose loss is bounded by full carry |
-| **Quantity risk** | The exposure left over when a hedge cannot exactly match a cargo, because futures trade in whole lots and a cargo rarely divides by five thousand bushels |
-| **Attribution** | Decomposing a finished trade into flat price, basis, calendar spread, freight, currency and financing, so the result can be explained rather than merely counted |
-| **Attribution bridge** | The line-by-line reconciliation from planned margin to realised margin, which has to close to the dollar |
-| **Margin financing** | The cost of funding variation margin paid out on a losing futures leg while the offsetting gain on the physical is still unrealised |
-
-# Quiz
-
-**Q1.** A desk finishes a soybean trade and wants to know what it was paid for. The cargo is 30,000 t, hedged with 220 November lots. It was bought in Iowa at November minus 75 when November settled at 1,239.50, and sold FOB Gulf at November plus 96 when November settled at 1,318.75. The plan, written when the trade was put on, had the sale going at November plus 105, freight and elevation at 132 c/bu, and inventory financing at 4.79 c/bu. Freight and elevation actually came to 141 c/bu; financing landed on plan. Funding the variation margin on the futures leg cost $1,720, which was not in the plan. Build the attribution bridge from planned margin to realised margin, and state what flat price contributed.
-
-**Q2.** A wheat book is net long 600,000 bu in March and net short 600,000 bu in May, with every other month square. The March/May carry widens from 11 cents to 19 cents. What is the profit or loss on the book?
-
-**Q3.** A destination buyer was offered November shipment at $336.00/t CFR against January at $342.00/t CFR, with silo at $2.20/t per month and money at 7.5 percent, so buying early and storing loses $2.60/t. The importer buys the November anyway. What is the $2.60 buying?
-
-**Q4.** In an exchange for physical, 184 December wheat lots change hands and the two sides agree to strike the futures leg 7 cents lower, raising the differential by 7 cents so that the miller's invoice for the wheat is unchanged. The merchant was short those 184 futures. What does her futures line show?
-
-**Q5.** Conversion drill. Brazil's safrinha corn yield is put at 5.9 t/ha. What is that in bushels per acre?
-
----
----
----
-
-# SOLUTIONS (spoilers)
-
-**A1.** Start with the quantity, because everything scales off it.
-
-30,000 t × 36.744 = **1,102,320 bu**. At 5,000 bu a lot that is 220.46 lots, so the hedge is **220 lots = 1,100,000 bu**, leaving **2,320 bu unhedged**. Hold that number.
-
-*The plan.* Basis captured is the sale differential plus the purchase discount: 105 + 75 = 180 c/bu. Against that, 132 of freight and elevation and 4.79 of financing, for 136.79. Planned margin is **43.21 c/bu**, and 1,102,320 × 0.4321 = **$476,312**.
-
-*What actually happened.* Basis captured was 96 + 75 = 171. Costs were 141 + 4.79 = 145.79. That leaves 25.21 c/bu, or 1,102,320 × 0.2521 = $277,895. Add the unhedged bushels and subtract the margin funding.
-
-| Line | Amount |
-|---|---|
-| Planned margin | $476,312 |
-| Basis: sold at +96 against +105, 9c on 1,102,320 bu | −$99,209 |
-| Freight and elevation: 141c against 132c, 9c | −$99,209 |
-| Quantity: 2,320 bu unhedged into a 79¼c rally | +$1,839 |
-| Margin financing | −$1,720 |
-| **Realised margin** | **$278,013** |
-
-Check it against the direct calculation: $277,895 + $1,839 − $1,720 = $278,013. The bridge closes, which is the only evidence you have that no line is missing.
-
-*What flat price contributed.* November went from 1,239.50 to 1,318.75, a rally of 79¼ cents. The physical earned that on 1,102,320 bu, which is $873,589. The short futures lost it on 1,100,000 bu, which is $871,750. The difference is **$1,839** — the 2,320 unhedged bushels, and nothing else.
-
-The trap is the $871,750. It looks like a disastrous hedge and it is nothing of the kind: it is the exact mirror of the physical gain, and a desk that books it as a trading loss has misunderstood what a hedge is for. The real story is two nine-cent misses, one on the sale differential and one on freight, which between them removed $198,418 — 42 percent of the planned margin — from a trade that still finished comfortably profitable.
-
-**A2.** Long the near month against a short in the deferred is **long the spread**. The spread widening means the deferred month is gaining on the nearby, which is a loss for that position.
-
-11 cents to 19 cents is 8 cents of widening, on 600,000 bu: 600,000 × $0.08 = **a loss of $48,000**.
-
-Two things worth noticing. First, nothing in this requires a view on wheat: the book is square in flat price and the loss comes entirely from a relationship between two contract months. Second, the loss is bounded. A carry spread cannot widen past full carry, because at that point anyone with a bin can buy the nearby, store it and sell the deferred for a risk-free return, and that arbitrage caps the spread. The mirror-image position — short the near, long the deferred — has no such ceiling, because an inversion can go as far as the shortage requires.
-
-**A3.** Days of cover.
-
-The $2.60/t is not a trading loss the importer has failed to notice; it is the price of holding stock in the country rather than on a seller's promise. A state buyer is managing consumption cover against an FX allocation calendar, a subsidy budget and a political risk of running short, not a P&L. Paying $2.60/t to move the grain two months earlier converts a supply risk into a known, small, budgeted cost — which is the definition of an insurance premium.
-
-The consequence for the seller is more useful than the arithmetic. It means an importer's buying is timed by a calendar you cannot see on a price screen, and it explains why importing markets typically show less carry than exporting markets: the destination is not paying for storage, it is paying to have arrived.
-
-**A4.** A gain of **$64,400**.
-
-184 lots is 920,000 bu, so one cent is $9,200 and seven cents is $64,400. The merchant is short those futures; closing a short at a price seven cents lower is a gain of seven cents on every bushel.
-
-The trap is the phrase "the invoice is unchanged". It is — the differential was raised by exactly the amount the futures leg was lowered, so the miller pays the same for the wheat. But the miller is long 184 futures and closes them 7 cents lower, which is a $64,400 loss on his futures line. His *all-in* cost is $64,400 higher, and the merchant's all-in revenue is $64,400 higher. Where the futures leg is struck is real money, moving in opposite directions across two books, which is exactly why exchanges require both parties to an EFP to hold a genuine related physical position and why the futures level has to be commercially defensible rather than merely agreed.
-
-**A5.** About **94 bu/ac**.
-
-The fast method for corn is to multiply tonnes per hectare by 16: 5.9 × 16 = 94.4. The exact factor is 0.0628 t/ha per bu/ac, so 5.9 ÷ 0.0628 = 93.9 bu/ac. Either way, call it 94.
-
-Worth carrying: US corn runs around 178 bu/ac this year, which is about 11.2 t/ha. A safrinha crop at 94 bu/ac is roughly half the US yield — and it is still three quarters of Brazil's corn, because it is planted on an enormous area behind the soybeans.
-
-# The written edition
-
-## The sheet is the only thing that knows
-
-A trading book is not a pile of grain and it is not a list of contracts. It is a statement of **exposure**, and the document that holds it is the position sheet.
-
-The first thing to understand about it is the sign convention, because it is the thing that makes physical and paper commensurable. A tonne sitting in a bin and a tonne sold forward to a miller are the same row on the sheet, with opposite signs. Ownership is positive. An obligation to deliver is negative. Futures sit in the same column as the physical they offset, and the net of the two is the only number that describes what the desk is exposed to.
-
-The second thing is the row. **Rows are futures months, not shipment months.** This trips up almost everybody on their first sheet, because operationally the world runs on shipment dates. But the sheet is not an operations document. It exists so that somebody can look at a number and know what to trade to change it, and the only instrument available is a futures month. A sheet organised by shipment date tells you what you owe. It does not tell you what you are exposed to.
-
-The third is **mark to market**. Every open line is repriced at the day's settlement, physical included, whether or not anything was bought or sold. Unsold corn in a Toledo bin gets a price every night. The consequence is that a merchant's P&L moves daily on grain nobody has yet agreed to buy, which is uncomfortable, and is also the only mechanism by which a problem becomes visible before the trade is over.
-
-## A flat book that is not flat
-
-Here is a small soft red wheat book, as it would print this morning. Quantities in bushels; a Chicago lot is 5,000.
-
-| Futures month | Physical long | Physical short | Net physical | Futures | Net |
-|---|---|---|---|---|---|
-| December | +1,500,000 | −900,000 | +600,000 | −120 lots (−600,000) | 0 |
-| March | +400,000 | 0 | +400,000 | 0 | **+400,000** |
-| May | 0 | 0 | 0 | −80 lots (−400,000) | **−400,000** |
-| **Total** | +1,900,000 | −900,000 | +1,000,000 | −200 lots (−1,000,000) | **0** |
-
-The total line reads zero. On any summary a manager is likely to see, this book is flat.
-
-It is not flat. It is long 400,000 bu of March against short 400,000 bu of May — an eighty-lot March/May spread that nobody decided to own.
-
-```chart
-{"type":"bar","unit":"lots, net position","title":"A flat book, month by month",
- "caption":"The total line reads zero, so nothing is wrong with it. The month lines are an eighty-lot March/May spread that no one put on deliberately, and it will not appear on any report that nets the book to a single number.",
- "source":"Worked example, episode 21.",
- "x":["December","March","May"],
- "series":[{"name":"Net position","values":[0,80,-80]}]}
-```
-
-How it happened is boring, which is exactly why it happened. Somebody bought 400,000 bu of farmer wheat for March shipment, and the hedge was added to the standing May line — because May is where the book's liquidity already sat, or because a roll had already taken the whole position out to May and the new purchase joined it. One entry. Invisible on the total.
-
-It surfaces in a thirty-second conversation with the risk desk:
-
-> **RISK:** Book's flat on the total. March/May is showing eighty.
->
-> **TRADER:** That's the Ohio wheat. Ships in March.
->
-> **RISK:** Hedge is in May.
->
-> **TRADER:** Roll took the whole line.
->
-> **RISK:** Then you're long the spread, and nobody bought it.
-
-Five lines, and neither party mentions a price. The conversation is entirely about *where* a position sits, which is what a risk conversation on a physical desk almost always is.
-
-### What the error costs
-
-Put a number on it. Take March/May carry at 14 cents and March wheat at $7.50 for the interest line.
-
-| Component | Calculation | Value |
-|---|---|---|
-| Storage, two months | 8c/bu/month × 2 | 16.00c |
-| Interest, two months | $7.50 × 5% × 2/12 | 6.25c |
-| **Full carry, March/May** | | **22.25c** |
-| Spread as traded | 14.00 ÷ 22.25 | 63% of full carry |
-| Room to widen | 22.25 − 14.00 | 8.25c |
-| Exposure | 8.25c × 400,000 bu | **$33,000** |
-
-Being long the near month, the worst case is the spread going all the way to full carry, and that is $33,000. It is bounded, because a spread wider than full carry is a free trade for anyone with a bin, and the bin owners arbitrage it away. This is the **good** version of the mistake.
-
-Reverse the signs and it is a different animal. A book short the near month and long the deferred loses as the market inverts, and an inversion has no ceiling at all: it goes as far as the people who need grain now are willing to pay. The same clerical slip, made in the other direction, has a bounded loss or an unbounded one depending purely on which month the grain happened to be in.
-
-### The split most people forget
-
-The same book is long wheat at Toledo and short wheat at the Gulf, both hedged in Chicago December. Split by month, it is square. Split by location, it is long the Toledo–Gulf basis spread.
-
-That is a real position. Toledo basis is set by farmer selling, local space and rail; Gulf basis is set by export demand, barge freight and vessel line-ups. The two move for different reasons and frequently in opposite directions. Last week made the point cleanly: the Gulf corn basis sat unchanged at 60 to 66 over December through a rally in flat price, because the flat price move came from energy and the export bid did not follow it.
-
-A position sheet split only by month hides this. Split it both ways, always.
-
-### How errors actually surface
-
-They rarely announce themselves. A position error looks like a number that is slightly different from the number in the other system.
-
-There are normally three versions of the same book: the trader's own sheet, the back office's from the contract records, and the risk system's from the trade capture feed. They are reconciled every morning. The discipline is not that they agree — they routinely do not, for reasons as dull as an unbooked washout or a contract entered with the wrong month. The discipline is that somebody has to *explain* each difference before the market opens. A difference that gets carried forward "to look at later" is how a real position ends up living inside a rounding argument for three weeks.
-
-## Attribution: what were you actually paid for?
-
-The trade is finished, the money is in, and the P&L says a number. That number, on its own, tells you almost nothing.
-
-A finished trade decomposes into six lines, and only one of them is a market view:
-
-| Line | What it is |
-|---|---|
-| Flat price | The board. In a hedged book this should be approximately zero |
-| Basis | The differentials you bought and sold at, against futures |
-| Calendar spread | What the roll gave you or cost you between months |
-| Freight | The physical cost of moving it, budgeted against actual |
-| Currency | Any FX leg, including one the hedge created rather than the trade |
-| Financing | Interest on the inventory, and on margin |
-
-Populating those six honestly is the whole exercise. Here is one done in full.
-
-### The trade
-
-45,000 t of corn, bought in central Illinois, moved to the Gulf, sold FOB. 45,000 t × 39.368 = **1,771,560 bu**, which at 5,000 bu a lot is 354.3 lots, hedged with **354 lots**.
-
-The plan, written down on 21 August with December corn at 508.50:
-
-| Line | Planned |
-|---|---|
-| Purchase basis, central Illinois | Dec −40 |
-| Sale basis, FOB Gulf | Dec +62 |
-| Basis capture | 102.00c |
-| Barge freight | −58.00c |
-| Elevation and handling | −12.00c |
-| Shrink and outturn | −2.00c |
-| Financing, 25 days at 6% on 468.50 | −1.93c |
-| **Planned margin** | **28.07 c/bu** |
-
-28.07c on 1,771,560 bu is **$497,277**.
-
-### What happened
-
-December corn settled at 535.75 on 15 September, up 27¼ cents from where the trade was put on. The cargo sold at December plus 55, not plus 62 — the Gulf bid did not follow the board up. Barge freight came in at 68 cents against 58 budgeted, because the USDA barge index had moved sharply in the intervening weeks. Financing landed on plan. Realised margin: **$195,587**.
-
-That is $301,690 short. The point of attribution is to say where every dollar of it went.
-
-```chart
-{"type":"waterfall","unit":"USD","title":"From planned margin to realised",
- "caption":"Flat price rose 27¼ cents on a 45,000 t cargo and contributed $425. Two estimates — a sale differential and a freight budget, neither of them a market view — took 61 percent of the planned margin.",
- "source":"Worked example, episode 21, using CBOT December corn settlements of 21 August and 15 September 2026.",
- "steps":[{"label":"Planned margin","value":497277,"kind":"base"},
-          {"label":"Basis miss","value":-124009},
-          {"label":"Freight overrun","value":-177156},
-          {"label":"Unhedged bushels","value":425},
-          {"label":"Margin financing","value":-950},
-          {"label":"Realised","kind":"total"}]}
-```
-
-| Line | Calculation | Amount |
-|---|---|---|
-| Planned margin | 28.07c × 1,771,560 bu | $497,277 |
-| Basis | sold Dec +55 against Dec +62, 7c | −$124,009 |
-| Freight | 68c against 58c budgeted, 10c | −$177,156 |
-| Flat price | 1,560 bu unhedged × 27.25c | +$425 |
-| Margin financing | 12 days at 6% on $482,325 | −$950 |
-| **Realised margin** | | **$195,587** |
-
-The four variance lines sum to $301,690, which is exactly the gap. That is the test. **A bridge that does not close means there is a line you have not found**, and the correct response is to go and find it rather than to book the difference as "other".
-
-### Read the flat price line again
-
-27¼ cents of rally, on a cargo of 1,771,560 bushels, contributed **$425**.
-
-The reason is arithmetic, not luck. 354 lots is 1,770,000 bushels against a cargo of 1,771,560, so 1,560 bushels were never hedged. The entire flat price result of a $9.5 million cargo through a two-week rally is the price move on those 1,560 bushels. Everything else cancelled: the physical gained $482,750 and the short futures lost $482,325.
-
-This is **quantity risk**, and it is structural rather than careless. Futures trade in whole lots and cargoes do not divide by 5,000 bushels, so a residual always exists. On this trade it was trivially small. On a book where somebody rounds 354.3 down to 350 because it is a rounder number, the residual is 21,560 bushels and a 27-cent move is $5,875 of P&L nobody authorised.
-
-### The margin financing line
-
-The short futures leg lost $482,325 as corn rallied. That money left the account daily as variation margin. The offsetting gain sat unrealised in the physical until the cargo was sold.
-
-Funding that gap for an average of twelve days at 6 percent cost about **$950**. Trivial on one cargo — and the reason it is in the bridge anyway is that it is the only line here that scales with the *number* of cargoes rather than their quality. A desk running forty positions through a trending market discovers this line as a treasury problem long before it appears as a P&L problem, and a merchant who has not modelled it finds out when the credit line stops rather than when the P&L prints.
-
-## Why a good P&L can hide a broken process
-
-The trade made $195,587. It is a winner. A desk that looks only at the total books it and moves on.
-
-Attribution tells a different story. Sixty-one percent of the planned margin was consumed by two estimates — a sale differential and a freight budget — and **neither of them was a market view**. Those are two separate failures owned by two separate people. A freight number that came in ten cents high is a purchasing and execution problem. A sale differential seven cents below plan is a market call, and one worth arguing about: the Gulf bid had been flat for a week while the board rallied, which was visible at the time.
-
-A single profit number cannot tell you which of those to go and change. It cannot even tell you that either of them happened.
-
-Now run the counterfactual, because it is the part that makes the case. Suppose December corn had *fallen* 27¼ cents instead of rising. The short futures would have made $482,325 and the physical would have lost it straight back. Basis and freight would have behaved exactly as they did. The realised margin would have been **the same $195,587**.
-
-That is the whole argument. The P&L of a hedged merchant carries almost no information about whether the market went your way, because by construction it is not supposed to. What it carries is information about whether your costs and your differentials were where you said they were — and you can only read that information if somebody has written the bridge.
-
-## The thing to carry away
-
-Two documents, and they answer different questions.
-
-The position sheet answers *what do I own right now*, and it only answers it honestly when it is split by futures month and by location. A single net number is a summary, and summaries are where positions hide.
-
-The attribution bridge answers *what was I paid for*, and it only answers it when it closes to the dollar. Six lines, one of which is a market view and five of which are estimates you made and can check.
-
-And the number that ties the two together: in a properly hedged book, flat price contributes nothing. If it contributed something, that is not a good trade. That is a position you did not know you had.

package/ep21.script.txt

@@ -1,108 +0,0 @@
-A book can be perfectly flat and completely wrong. ||| 0.5
-The total line says zero. The risk is in the rows underneath it. ||| 0.6
-This is Soft Commodity Trading, episode twenty-one. Today, the book and P and L attribution. How a desk knows what it owns, and how it finds out what it actually got paid for. ||| 0.8
-Yesterday Chicago wheat fell nine and a half cents by mid-morning and settled six and a half cents higher. ||| 0.5
-December soft red ended at seven twenty-eight and a half. ||| 0.4
-December corn, five thirty-five and three quarters, up two and a half. ||| 0.35
-November beans, thirteen eighteen and three quarters, up fourteen and a half, with October meal at three hundred and sixty dollars ten, up nine ninety. ||| 0.5
-The wheat round trip is the story. ||| 0.4
-On Monday the American president said Russia and Ukraine had agreed to stop attacking each other's energy infrastructure. ||| 0.45
-Wheat dropped about eighteen cents on the headline. ||| 0.4
-Then fresh strikes were reported around Odesa, and most of it came back. ||| 0.6
-Here is what the market was actually repricing, and it was not the probability of peace. ||| 0.45
-An energy truce is not a grain corridor. ||| 0.45
-Refineries and power stations are one target set. Grain berths at Novorossiysk and Odesa are another. ||| 0.5
-Nothing announced on Monday covered the second one. ||| 0.5
-So the board was not pricing a lower chance of disruption. It was discovering the scope of the sentence it had just read. ||| 0.6
-And here is the number that did not move. Russia shipped two million tonnes of grain by sea in August. Down sixty-two percent on the year. ||| 0.5
-Ukraine's wheat exports since July stand at two point one million tonnes, down forty-eight percent. ||| 0.45
-Algeria, Pakistan and Saudi Arabia are buying somewhere else. ||| 0.5
-That is loading data. It did not change on Monday, and it will not change on a headline. ||| 0.7
-Sixteen cents, out and back, inside one session. ||| 0.4
-A hedged book should not have noticed. ||| 0.4
-Whether it noticed is a question you answer with a position sheet, not with a profit number. ||| 0.7
-So. The position sheet. ||| 0.4
-It measures exposure, not inventory. ||| 0.4
-A tonne sitting in a bin and a tonne sold forward are the same row, with opposite signs. ||| 0.5
-The rows are futures months. Not shipment months. ||| 0.4
-The column is lots. Five thousand bushels a lot in Chicago. Long is positive, short is negative. ||| 0.5
-And the reason the rows are futures months is simple. The only thing you can trade to change the number is a futures month. ||| 0.5
-A sheet organised by shipment date tells you what you owe. It does not tell you what you are exposed to. ||| 0.7
-One more convention before the numbers. ||| 0.35
-Every open line is marked to market daily, at the settlement, whether or not anything was sold. ||| 0.5
-Physical included. Unsold corn in a Toledo bin gets a price every night. ||| 0.45
-Which means a merchant's profit and loss moves every day on grain nobody has agreed to buy yet. ||| 0.5
-If that sounds uncomfortable, it should. It is also the only way to see a problem before the trade is over. ||| 0.7
-Take a small soft red wheat book, this morning. ||| 0.4
-In December, one and a half million bushels bought from farmers, nine hundred thousand sold to a miller. Net long six hundred thousand. Short a hundred and twenty December lots against it. December is square. ||| 0.6
-In March, four hundred thousand bushels bought, nothing sold, and no hedge. Long four hundred thousand. ||| 0.5
-In May, no physical at all, and eighty lots short. Short four hundred thousand. ||| 0.6
-Add it up. Net physical, long one million. Futures, short two hundred lots, one million. ||| 0.45
-Total position, zero. The book is flat. ||| 0.5
-It is not flat. It is long March against short May, eighty lots, and nobody decided to own that. ||| 0.8
-How that happens is boring, which is why it happens. ||| 0.4
-Somebody bought farmer wheat for March shipment and the hedge went onto the standing May line, because May is where the book's liquidity already was. Or the roll took the whole position at once. ||| 0.6
-One entry, invisible on the total, and it surfaces in a thirty-second conversation with the risk desk. ||| 0.6
-RISK: Book's flat on the total. March May is showing eighty. ||| 0.25
-TRADER: That's the Ohio wheat. Ships in March. ||| 0.25
-RISK: Hedge is in May. ||| 0.25
-TRADER: Roll took the whole line. ||| 0.25
-RISK: Then you're long the spread, and nobody bought it. ||| 0.6
-Five lines, and neither of them said a price. ||| 0.5
-So what does it cost? ||| 0.35
-March May carry is running around fourteen cents. Full carry, over two months, is about twenty-two and a quarter. Sixteen cents of storage at eight a month, plus six and a quarter of interest on seven fifty wheat. ||| 0.6
-So the spread sits at about sixty-three percent of full carry. ||| 0.5
-Being long the near month, the worst case is the spread going to full carry. Eight and a quarter cents, on four hundred thousand bushels. Thirty-three thousand dollars. ||| 0.6
-Bounded. And that is the good version of this mistake. ||| 0.5
-Reverse the signs. Short the near, long the deferred. An inversion has no ceiling at all. ||| 0.7
-One more split, and most people forget it. ||| 0.4
-The same book is long wheat at Toledo and short wheat at the Gulf. ||| 0.4
-Both hedged in Chicago December. Flat on the board. ||| 0.4
-But Toledo basis and Gulf basis are two different prices that move for different reasons. ||| 0.5
-Last week proved it. The Gulf corn basis sat at sixty to sixty-six over December while flat price rallied. ||| 0.5
-A sheet split by month and not by location hides a basis spread. Split it both ways. ||| 0.7
-Errors on a position sheet almost never look like errors. ||| 0.45
-They look like a number that is slightly different from the one in the other system. ||| 0.5
-The trader's own sheet, the back office's, the risk system's. Three versions of the same book, reconciled every morning. ||| 0.5
-And the discipline is not that they agree. It is that somebody has to explain why they don't, before the market opens. ||| 0.7
-Now the second half. The trade is finished. What did you actually get paid for? ||| 0.5
-A finished trade splits into six lines, and only one of them is a market view. ||| 0.5
-Flat price. Basis. The calendar spread you rolled through. Freight. Currency. Financing. ||| 0.6
-Say those six once and you have the whole framework. The work is populating them honestly. ||| 0.6
-Here is one, in full. Forty-five thousand tonnes of corn, bought in central Illinois, moved to the Gulf, sold F O B. ||| 0.5
-Forty-five thousand tonnes is one million seven hundred and seventy-one thousand five hundred and sixty bushels. Three hundred and fifty-four lots. ||| 0.6
-The plan, written down on the twenty-first of August. Buy at December minus forty, with December at five oh eight and a half. Sell at December plus sixty-two. ||| 0.55
-Freight fifty-eight cents, elevation twelve, shrink two, financing one point nine three. ||| 0.5
-Planned margin, twenty-eight point oh seven cents a bushel. Four hundred and ninety-seven thousand dollars. ||| 0.7
-What happened. December corn went to five thirty-five and three quarters. Up twenty-seven and a quarter. ||| 0.5
-The cargo sold at December plus fifty-five, not plus sixty-two. ||| 0.45
-Barge freight came in at sixty-eight cents, not fifty-eight. ||| 0.5
-Realised margin, one hundred and ninety-five thousand five hundred and eighty-seven dollars. ||| 0.6
-Three hundred and one thousand short of the plan. Now attribute it. ||| 0.7
-Basis. Seven cents of sale differential missed, on one point seven seven million bushels. Minus a hundred and twenty-four thousand. ||| 0.55
-Freight. Ten cents over budget. Minus a hundred and seventy-seven thousand. ||| 0.55
-Financing the margin calls. The short futures bled four hundred and eighty-two thousand dollars of cash, paid daily, and carrying that cost about nine hundred and fifty dollars. Nobody put that in the plan. ||| 0.6
-And flat price. Twenty-seven and a quarter cents of rally. ||| 0.45
-Contribution, four hundred and twenty-five dollars. ||| 0.6
-Four hundred and twenty-five, because three hundred and fifty-four lots is one million seven hundred and seventy thousand bushels, against a cargo of one million seven hundred and seventy-one thousand five hundred and sixty. ||| 0.55
-Fifteen hundred and sixty bushels unhedged. That is the entire flat price result. ||| 0.7
-Those four lines add to three hundred and one thousand six hundred and ninety. Exactly the gap. ||| 0.6
-And that is the test. A bridge that does not close means you have a line you have not found yet. ||| 0.7
-Now the part that matters. ||| 0.4
-That trade made a hundred and ninety-five thousand dollars. It is a winner. ||| 0.5
-A desk that looks only at the total books a win and moves on. ||| 0.5
-Attribution says something else. Sixty-one percent of the planned margin was taken by two estimates, and neither of them was a market view. ||| 0.6
-A freight number is a purchasing failure. A basis number is a trading call. Different people, different fixes. ||| 0.6
-A single profit number cannot tell you which one to go and change. ||| 0.7
-And run it the other way. If December corn had fallen twenty-seven cents instead, the hedge would have made four hundred and eighty-two thousand and the physical would have lost it straight back. ||| 0.6
-Same margin. Same hundred and ninety-five thousand. ||| 0.5
-Which is the point. The profit of a hedged merchant carries almost no information about whether the market went your way. ||| 0.55
-It carries information about whether your costs and your differentials were where you said they were. ||| 0.7
-That is why attribution is how desks actually learn, and why the total is not. ||| 0.7
-Three things to keep. ||| 0.4
-One. A net position of zero is a claim, not a fact, until you have split it by month and by location. ||| 0.6
-Two. Every finished trade gets a bridge from what you planned to what you got, and the bridge has to close to the dollar. ||| 0.6
-Three. In a hedged book, flat price is supposed to contribute nothing. When it contributes something, that is not skill. That is an unhedged bushel you did not know about. ||| 0.8
-Next time, risk management, and every reason a hedge leaks. Basis, quantity, quality, timing, currency, cross-hedge. A fully hedged trade that still loses money, taken apart line by line. ||| 0.6
-Four questions in the written edition, with the solutions worked in full, and a conversion drill at the end. ||| 0.5
-Thanks for listening. ||| 0.7