Conversation

2) What went wrong with Nickel?

Quote Tweet

Bloomberg

@business

Mar 14

Official

The LME will reopen its nickel market on Wednesday, more than a week after the exchange suspended trading and canceled about $3.9 billion of transactions as it grappled with an unprecedented short squeeze trib.al/DvBvLjj

7:05 AM · Mar 15, 2022Twitter Web App

3) Let's say that some entity (X) has $2b of collateral and puts on a $3b short in a contract trading at $1 (so 3b contracts). Markets start to go up in price--eventually they're at $1.33. Now the trader has: position size: 3b contracts = $4b collateral: $1b

Replying to

4) X now has a huge position on, and is fairly leveraged--$4b notional, 4x leverage. Another 25% move and the trader is underwater; each 25% move beyond that costs _someone_ -- the exchange, or insurance, or whatever is backstopping it -- about $1b!

5) The traditional margin model says: reach out to X and warn then to top up, _or else_. But X has at least a business day to top up. If it's a Friday evening, then they have at least 72 hours. Except there's a war going on in Ukraine, and commodity prices are skyrocketing.

6) By the time the next business day ends, there have been 3 days for volatility. A 25% move bankrupts the trader. A 50% move means that someone's out $1b of insurance. That could happen in 3 days! And, in fact, it did, in Nickel. (Note the weekend gap!)

7) There are really two problems here. First, the exchange margin calls based on _time_, not _price_. What does that mean? Well, when X reaches 6x leverage on a $4b position, it's time to issue a margin call.

8) The goal of the margin call: before X runs out of collateral, either (a) X tops up their collateral (b) X reduces their position Once the contract reaches a price of $1.66, X will fully run out of collateral; any move beyond that and X goes _negative_ account value.

9) Which means--the goal is for X to de-leverage before the contract reaches $1.66. Time is arbitrary! The contract could move 1% in 3 days, or it could move 150%. What matters is the price. You have to close down before contracts hit $1.66.

10) Say that closing down X's position--buying back 3b contracts--would have 10% impact. The means that, if X isn't going to top up, you have to start reducing their position before the contract hits around $1.50--adding 10% impact to that would impact markets to around $1.66.

11) So in this hypothetical, you have to begin liquidating X's position when the contract hits $1.50. completely independently of _when_ that happens. If a week later the contract is at $1.35, it's fine. But if a minute later it's at $1.50, it's time to start closing down X.

12) This is how liquidations work on FTX--we begin de-leveraging a position as soon as it's running too low on collateral, independent of how long it's been. We *have* to--that's the only way we can stop things like this! Sometimes 1 day is too long.

Quote Tweet

Bloomberg

@business

Mar 14

Official

117

13) Anyway, that's one of the core innovations of digital asset exchanges--real-time margining. The other problem that LME Nickel ran into was the weekend. The exchange wasn't open for a few days, and so the _fastest_ they could margin call was 72 hours.

147

14) This exacerbated the original problem--instead of eating 1 day's risk, they were eating 3 days'. Another difference with crypto markets, which operate 24/7--allowing 24/7 margin calculations and de-leveraging.

103

15) These are some of the issues at the heart of our

@CFTC

application: comments.cftc.gov/PublicComments

141