Case study: negative oil, April 2020

On Monday 20 April 2020, the May 2020 WTI crude oil futures contract settled at −$37.63 per barrel. A long who held one contract from the morning's price of around $18 to that settlement lost roughly $55,000 on a position that had, a few hours earlier, been worth $18,000. Several retail traders ended the day owing their brokers more than their accounts had ever held. Every mechanism in this course was involved, which is why it is worth walking through.

The setup

By April 2020, pandemic lockdowns had cut global oil demand by a large fraction while production had barely slowed. Crude was being pumped and put in storage, and the delivery point for the WTI contract, the tank farm at Cushing, Oklahoma, was filling up. Anyone taking delivery in May needed somewhere to put 1,000 barrels per contract, and the places to put it were being booked out.

The May contract (CLK20) was due to stop trading on Tuesday 21 April. Physically delivered contract, last trading day approaching, delivery capacity effectively gone. Every long still holding on Monday afternoon had to either sell to someone or take delivery of oil they could not store.

What the exchange had done

CME had announced in the preceding two weeks that its systems would accept negative prices for energy contracts and had switched its options pricing model to one that permitted them. This was public. Some brokers relayed it; some trading platforms could not display a negative price; a number of retail traders had never considered the possibility because it had never happened.

The day

Time (ET, approximate)	May WTI price	What was happening
Friday 17 April settle	$18.27	Already the lowest in decades; June contract around $25
Monday morning	$15 to $18	Steady selling; open interest still tens of thousands of contracts
Early afternoon	$10 to $5	Bids thinning as storage-holding buyers step back
Around 2:00 pm	$0	Zero crossed for the first time in the contract's history
2:08 to 2:30 pm	−$10 to −$40	Longs paying anyone to take the contract; low print around −$40.32
2:30 pm settlement	−$37.63	VWAP of the final two minutes
Tuesday 21 April	recovered to about +$10 at the final close	The last of the longs cleared; contract expired

The June contract fell that day too, to around $20, but stayed positive. The spread between May and June, normally a dollar or two, ended the day at nearly $60. The entire collapse was in the one contract whose holders had run out of time.

Who was hurt, and how

Retail longs in the May contract. Some had bought "cheap oil" at $10 or $5 on the theory that it could not go lower. The arithmetic for one CL contract from $5 to −$37.63 is 42.63 × $1,000 = $42,630, on a position whose entire notional at entry was $5,000. Account leverage was not the problem; the problem was that the loss exceeded the notional, which is impossible in a stock and had never occurred in this contract.

One large retail broker later disclosed a loss of roughly $100 million from customer accounts that went negative and could not be collected in full, and several platforms admitted their software had been unable to process prices below zero, so customers could not see or exit their positions.

A retail structured product sold by a large Asian bank, which tracked the front-month contract and rolled late, lost more than its customers had invested; the bank absorbed part of the loss after public pressure.

Traders in the June contract, and in micros. MCL had launched the previous year; a micro long from $5 to −$37.63 lost $4,263, painful but survivable. June longs lost money but never faced a negative price.

What the case teaches

Each of the following is a lesson from an earlier module, made concrete:

Physically delivered contracts behave differently near expiry (Module 1). The participants left in the last days are the ones dealing with delivery, and price reflects storage, not oil.
Notional is not the maximum loss. In a physical contract the loss is bounded by the cost of disposal, which can exceed the price.
Brokers' cut-offs exist for a reason (Module 1, Lesson 4). Traders who followed the standard "be out before the last few days" rule were in June, not May.
Platform and broker infrastructure is a risk (Module 5). If the software cannot display the price, you cannot manage the position.
Micros bound the damage (Module 2). Same lesson, one-tenth the tuition.
The public warning was there. The exchange's announcement was available two weeks earlier. Knowing what the exchange has said about a contract you hold is part of holding it.

Key idea: A physically delivered contract near expiry is a contract about storage and logistics, not about the commodity's price. Be out of the front month well before the last trading day, and never assume zero is a floor.

What a rule looks like

A simple one: never hold a physically delivered contract into the final five trading days before its last trading day, and never hold any front-month energy contract through a period when the exchange has issued special notices about it. Add the last trading day and first notice day to your calendar the day you enter, as Module 1 said. The traders who did that in April 2020 watched the event on the news.

Try it: Look up the current CL front month's last trading day and the day on which most volume has already moved to the next month. Compute the difference in calendar days. Then write the rule you will follow for exiting physically delivered contracts, in one sentence, and put it in your risk plan.

Recap

On 20 April 2020 the May WTI contract settled at −$37.63 the day before expiry because Cushing storage was full.
Longs lost more than the contract's notional; some retail accounts went negative and some platforms could not display the price.
The June contract stayed positive; the whole collapse was in the one contract whose holders had run out of time.
Physically delivered contracts near expiry price storage and logistics, not the commodity.
Exit the front month well before the last trading day, read exchange notices, and use micros to bound the damage.