Citi Had No Time for Former Goldman Partners' FX Losses
Here's a well-known chart:
Up until January 15, the Swiss franc was capped against the euro and was pretty quiet against the U.S. dollar, trading at about 1.02 francs to the dollar as of that morning. Then the Swiss National Bank removed the cap and the franc spiked to 74 centimes to the dollar, before pretty quickly bouncing back part of the way. It's now back around parity. If you were short the franc against the dollar at 4:30 a.m. New York time that day, you had lost more than a third of your money less than half an hour later. 1 But half an hour after that, your losses were back to a more manageable 15 percent. And two months later they had pretty much disappeared.
Unless, of course, you closed out your position on January 15, in which case your losses were permanent. And obviously someone was buying francs (selling dollars) all the way up (down). Someone bought francs at the worst possible time.
Who? Well, for one thing, probably a lot of people with no choice in the matter. People like Tormar Associates. "Tormar is the joint family office of former Goldman Sachs partners Ron Marks and John Tormondsen" (get it?), and it did a lot of foreign exchange trading with Citi as its prime broker. Much of that FX trading involved betting against the Swiss franc, using leverage from Citi. When the bad thing happened, Tormar's equity vanished, and an awkward conversation became necessary:
On January 15, 2015, an employee on Citibank’s Foreign Exchange Prime Brokerage desk informed Tormar of a margin (or collateral) deficit of approximately $29 million. Under the parties’ Master Agreement, and the Credit Support Annex, that deficit required either that Tormar post additional collateral in the amount of $29 million or close out its positions in its foreign exchange transactions with Citibank.
Mr. Tormondsen of Tormar explained to Citibank that Tormar could not cover the margin call, nor would it be able to do so in the future. The parties agreed that Tormar would unwind its positions to avoid potential further losses and additional liability to Citibank.
That's Citi's versions of events, which ends with it suing Tormar for $25 million because, as promised, Tormar really didn't ever come up with the money. (Tormar ended up owing Citi about $35 million, according to Citi, but had already posted $10 million of collateral, leaving a net claim of about $25 million.) Apparently though that awkward conversation was also full of misunderstandings; here's Tormar's version:
“The actions Citibank took on January 15, 2015 following the Swiss National Bank’s move to suddenly abandon its efforts to hold down the value of the franc were a panicked response to a temporary market dislocation,” Tormar said in a statement. “Had Citibank taken an appropriate approach, as required by our agreements, and worked with us, neither Citibank nor Tormar would have suffered any losses, as the positions quickly and inevitably rebounded in value.”
"Inevitably" is sort of a crazy word there -- presumably the rebound didn't feel inevitable on the way down -- though "quickly" is true enough. But what about "neither Citibank nor Tormar would have suffered any losses"? If you were short francs against the dollar, you're almost back to breaking even, and if you were short francs against the euro you're still way behind.
On the other hand if you sold out-of-the-money options on the franc that very briefly turned into very in-the-money options, you lost a lot of money and then quickly got it back. And that seems to have been Tormar's situation. This is a fun little case just because Citi has filed most of the documents it had with Tormar, so if you ever wanted to see what a prime brokerage agreement between a big bank and a "joint family office" of rich FX traders looks like, now's your chance. Here are Citi's prime brokerage agreement, ISDA Master Agreement, credit support annex, termination notice and calculation of the amounts that Tormar owed. And that last statement has a pretty complete list of Tormar's trades at the time Citi blew it out. I cannot entirely recommend that you read the list -- it's 23 pages long, in tiny type, and fuzzily scanned -- but you can skim it for a general sense of what was going on.
My general sense, from skimming it, is: Tormar was down $3.7 million on FX spot and forward transactions, but the bigger losses -- responsible for $31.5 million of Citi's claim -- were for option trades. The biggest clusters of trades -- assuming I'm reading the statement right! -- were put options that Tormar sold on the dollar against the Swiss franc, betting that the dollar wouldn't decline. 2 For instance:
- It had sold February/March puts on $100 million against the Swiss franc at 0.9125 francs to the dollar. These were out-of-the-money trades when it put them on in November and December (when the dollar was trading above 0.95 francs), very in-the-money as of the morning of January 15, and back out-of-the-money by the end of January. Citi closed Tormar out of this trade at a cost of some $6 million. 3
- It had sold May puts on $60 million at 0.9475, 0.9525 and 0.96 francs to the dollar, which it got out of at a total cost of some $8.3 million. 4
- It had sold January and early February puts on $47 million at 0.90 francs to the dollar. Tormar had realized a premium of less than $200,000 on selling those options in September and October of last year, when the dollar was above 0.94 francs. The options were set to expire on January 27 and February 4. If Citi had just let them expire, they'd have expired worthless -- the dollar last closed below 0.90 francs on January 23 -- and Tormar could have kept its money. Instead, Citi closed it out at a cost of $1.4 million. 5
That is sad for Tor and Mar, if I may: They really did pick the worst possible day to get blown out of these trades, and they seem to be mostly right that they would have made up all their losses if Citi had just let them ride for a few more weeks. But obviously that is not how prime brokerage works. Citi's job was not to help Tormar double down and recoup its losses. Citi's job was to protect itself. It is a thankless job, a job where Citi collected fee income for taking small-deep-hole gap risk. (In other words, it's sort of like selling puts against the Swiss franc.) I sympathize with prime brokers who behave with utter ruthlessness when a customer gets in trouble. 6
And Citi doesn't even seem to have been all that ruthless. As far as I can tell, it closed out Tormar on the worst possible day, but not at the worst possible time: It seems to have waited until after that horrible spike in the franc to close out Tormar, at relatively normal rates for the admittedly abnormal day of January 15. 7 That is probably not a matter of benevolence but of reaction time: By the time Citi figured out Tormar's deficit, called to ask for margin, and got Tormondsen's response of turning out his empty pockets and shrugging cartoonishly, 8 the worst was already over for Tormar's bet against the franc. If Tormondsen could have somehow drawn out that shrug for another week, he might have broken even. 9
That might have been better for Citi, too; who knows how much it'll end up getting back from Tormar. (Tormar is fighting its claim, calling Citi's losses "self-inflicted.") And of course there's a systemic argument against over-hasty liquidation: Presumably a lot of that short intense spike in the franc was caused by behavior much like this, retail and institutional brokers blowing leveraged clients out of money-losing trades -- which caused the clients to lose more money -- which caused more liquidations -- etc. Again, Citi's liquidation of Tormar seems not to have contributed directly to the spike, but Citi was reported to have lost $150 million on the Swiss move, and it's certainly possible that forced sales at its prime brokerage exacerbated losses for its trading desks. Also notice how leveraged Tormar was: While it had a trading limit of $200 million with Citi, 10 that seems to have been based on the market value of positions, not notional amount. Just the three trades in the bullet points above came to $207 million worth of francs. So Tormar was trading hundreds of millions of dollars worth of francs with $10 million of collateral. Of course FXCM was offering 50:1 leverage to retail FX clients, so I guess this is all pretty standard, but still. As it did for many of those retail clients, that leverage quickly went very wrong for Tormar.
While we're on that topic, here's how Citi calculated how much margin Tormar owed it: 11
"Required Margin" is determined by Party A in its sole discretion by multiplying the Net Open Position for each currency by its applicable Spot Margin Percentage and then aggregating all such amounts to a single USD amount
"Spot Margin Percentage" shall mean a percentage number assigned to a currency based upon the volatility of that currency, as determined by Party A in its sole discretion.
Citi was Party A. There was more math than this, but all the math was built on: Citi makes up a number, based on how nervous it feels, and Tormar posts it. Before January 15, when Swiss franc volatility was very low, that number was presumably also very low. Afterwards, when volatility was rather higher, you'd sort of expect that number to get higher. There's an obviously pro-cyclical element to that too: Clients lose money, prime broker gets nervous, prime broker raises margin requirements (in its sole nervousness-based discretion), more clients have margin calls that they can't meet, more client assets get sold, more clients lose money, prime broker gets more nervous, etc. Again, there probably wasn't even time for that to happen here, but if you want to worry about market stability it's one more thing to worry about: Nervousness itself can create problems, and the more that nervousness is built into contracts, the more problems it can create.
To see that, it helps to flip the chart around. The franc was at about $0.98 at 4:30 a.m. New York time; it peaked at about $1.32 at 4:56 a.m. (+35%), and by 5:30 it was back about $1.127 (+15%). So by "you had lost more than a third of your money" I mean, like, you had dollar losses equal to more than a third of the dollar notional of your position.
Citi's complaint (paragraph 29) says:
Tormar’s foreign exchange portfolio was heavily exposed to the Swiss Franc, effectively amounting to a net short position with respect to the currency. In other words, Tormar was betting that the Swiss Franc would fall in relation to other currencies.
And that's true enough as a matter of, like, delta. I'm just being overly literal when I say that it was a bet that the dollar wouldn't decline against the franc.
I should say, there were a ton of other trades, in various currencies and expirations and so forth, and I don't pretend to be doing any sort of justice to Tormar's overall strategy. I'm just saying, the trades that Citi is coming after it for look pretty much like: Tormar sold puts betting that the franc wouldn't rise against the dollar, and that bet went wrong.
See the bottom of page 8 and the top of page 9 of the calculation statement. I read this cluster to mean that Tormar had sold ("S" in the "B/S" column) options on the "Nominal Amount" of USD for the positive premiums in the "Premium Amount" column at various dates in the "Trade Date" column, and had bought some of them back at later trade dates, but that on January 15 Citi essentially bought in the remaining $100 million of options in four transactions -- the top 4 trades listed in this cluster, beginning with "Trade Key" 42071602 -- at a premium of $916,756.64, $317,096.24, $2,040,613.48, and ... call the last one $2.8 million, it's pretty smudged.
That's further down page 9 of the statement, and I'm combining a few clusters.
Back on page 8 of the statement.
Related: I enjoy the utter ruthlessness that JPMorgan and Citi seem to have applied to their clearing relationships with Lehman Brothers. These are very similar: If you have a counterparty that has defaulted, or is about to default, and you have the ability to sell off that counterparty's assets, you should do that with gleeful abandon and try to make some money for yourself while you're at it.
Again, if I'm reading the statement right. If you look at the USD/CHF spot/forward transactions on pages 5 and 6 of the statement, there are what seem to be more or less offsetting forward trades for round numbers of dollars, with one odd trade for $3,661,739.49 at 0.85732 francs per dollar, far away from all the other trades which were at at least 0.87. There's only one day in the last two years on which the franc got that low, and it's January 15. So my guess is that that $3.7 million was the plug in the spot/FX trades that Citi was owed, and it was calculated at a rate that was well away from the horrible spike on that day.
Or the telephonic equivalent thereof. Though I guess Tormar disputes Citi's description of the conversation.
Ask me sometime about the client who forgot to post margin on a trade for a week because all of its executives were on vacation! Good times.
See page 5 of the prime brokerage agreement.
That's from page 14 of the CSA.
To contact the author on this story:
Matt Levine at email@example.com
To contact the editor on this story:
Zara Kessler at firstname.lastname@example.org