Fabrice Tourre, the former Goldman Sachs vice president recently found liable by a jury for, like, the financial crisis, wants a do-over, which is understandable enough. Yesterday he filed a motion asking the judge to throw out some bits of the SEC's fraud case against him, and grant him a new trial on the other ones. It's sort of boring but the gist of it is mostly "no jury could have decided against me so ignore the jury's decision against me and put it in the hands of another jury." That sort of thing usually sounds better to lawyers than to judges, so I'm short Fab's chances here.

Still there's a fun bit here that I did not know about.* The schematic story of Fab goes something like this:

1. Hedge fund manager John Paulson came to Fab for help in betting against some mortgages.

2. Fab hooked him up with ACA, which was willing to bet on the mortgages.

3. ACA was fairly relaxed about reviewing the quality of these mortgages because for some reason it thought Paulson was betting on the mortgages along with ACA, instead of betting against them. This was a weird thing to think about a bet, but whatever. The main claim of fraud is that ACA thought this weird thing because Fab Tourre told them lies. Fab's defense was that what he told them was "not accurate" but also not "false." How could a jury find against him, etc.

4. So they documented up a deal and named it Abacus.

5. In the deal, Paulson bet against the mortgages with Goldman.

6. Goldman hedged by betting against the mortgages with ABN Amro.

7. ABN Amro hedged by betting against the mortgages with ACA.

8. ACA didn't hedge, it was just betting on the mortgages.

Where did ABN Amro come from, you might well ask, and the answer is: Fab and his crew at Goldman worried that ACA might not be good for the money, if the mortgages ever went wrong, so they asked ABN Amro to intermediate between Goldman and ACA. ABN Amro agreed, and negotiated a fee of a few basis points on the $909 million notional. In exchange for that fee, ABN Amro was on the hook if both (1) the mortgages went south and also (2) ACA went south.

It stood to reason that, since ACA was basically in the business of betting on mortgages, if the mortgages went south then so would ACA. They did, and it did, with perhaps surprising directness; ABN Amro ended up losing some $840 million on the deal.

Anyway, in the course of arguing that none of this was Fab's fault, his lawyers argue that, in essence, neither ACA nor ABN Amro was deceived by any of the which-side-was-Paulson-on stuff, because neither of them cared about who picked the mortgages, or whether they were any good. This was surely true. But awkward.

The argument about ACA is pretty boring; it's that "ACA’s pricing decisions rested not on the price, yield, value or volatility of the reference portfolio, but on anticipated correlation factors," and that correlation "is essentially a measure of systemic risk, meaning that ACA’s premium calculations were very heavily influenced by its projections of how the subprime sector as a whole would perform." Probably, but the whole trial was basically a parade of ACA people being all "we trusted John Paulson to pick the best mortgages," so you could see a judge not being convinced here.

The argument about ABN, which, remember, actually lost the money here, is that ABN didn't care about the mortgages at all. ABN cared only about ACA:

As former ABN employee Dean Atkins testified, when determining whether to enter into an intermediation trade with a monoline insurer like ACA, ABN “[p]rincipally” considered the availability of a credit line for that insurer, and the economics of the proposed trade, meaning the spread ABN would be paid for taking that credit risk. … Mr. Atkins reaffirmed that, in entering into intermediation transactions with monoline insurers like ACA, its “principal concern was the creditworthiness of the counterparty.”

This makes sense: The deal here is that ABN just gets free money as long as ACA doesn't go bankrupt, so its first concern is that ACA not go bankrupt. And ACA probably wasn't going to go bankrupt from losing money on these specific mortgages: ABN would only be in trouble if ACA lost money on lots and lots of mortgages. In which case the relative quality of these mortgages would be somewhat beside the point. So Fab is probably right that no one at ABN cared one way or the other about how good these mortgages were, what side Paulson was on, etc.; they just wanted to make sure ACA would stay around.

So what did they do? Well they looked at their credit limits to ACA, sure, but they also hedged their exposure to ACA. With, of course, Goldman:

ABN’s focus on ACA credit risk is apparent as well from its requirement that Goldman sell it protection on ACA corporate credit in an amount directly proportionate to the size of the potential amount due under the ABN/GSI swap. Tourre Trial Tr. 2348:9-11 (“ABN Amro had requested us to sell them protection on ACA with a notional amount of three percent of the super senior swap.”) ….

So Goldman didn't want to take ACA's credit risk, so it found ABN to stand in the middle. And ABN didn't want to take ACA's credit risk, so it bought protection from Goldman. And so trades are made. Got it? No? This might not help but still, here:

The trick is to have as many arrows going in as many directions as possible; that's how value is added.

In general, the value can be added in a bunch of places: The swap that Goldman writes to ABN on ACA might have different accounting or regulatory capital or credit limits or whatever from the swap that ABN writes to Goldman on Abacus, so Goldman and ABN can benefit by buying and selling economically equivalent-ish things to and from each other.

But the value added here is probably just that Goldman offloaded $909 million of ACA exposure on ABN, and ABN only offloaded three percent of that back onto Goldman. The reasoning for that seems to be roughly that ABN figured the expected payoff of the Abacus protection that ACA was selling was about three percent of notional -- that is, roughly, that on average ACA would wind up owing ABN $30 million -- and so they wanted protection against that expected amount.** As the expected amount got bigger -- as the mortgages started looking worse and worse -- then theoretically ABN would buy more protection on ACA, though of course that protection would then be more expensive. It appears that they did not follow the theoretically appropriate hedging strategy, which is psychologically understandable,*** so they lost a ton of money.

That's not much of a value add? I mean you might ask questions like: Who, between Goldman and ABN, would be better at appropriately adjusting that hedge over time to reflect the actual probability of ACA owing money on Abacus that it can't pay? I would have said Goldman, who were, y'know, a bit more in the mortgage-trading business than ABN Amro. And I probably would have been right; they couldn't have done a much worse job than ABN anyway.

But I guess at the time it looked like it added value, insofar as ABN went to bed that night feeling like it wasn't, in expectation, exposed to any ACA credit risk. This little curlicue on the Abacus deal meant that, just as no one spent too much time worrying about the underlying mortgages, no one spent too much time worrying about ACA's credit risk either. Paulson was actually betting against ACA, but had Goldman and ABN in between, so he didn't need to care if ACA survived or not. Goldman and ABN could both tell themselves that their risk to ACA was hedged: Goldman by intermediating the trade through ABN, ABN by hedging with Goldman.

As a first cut, it turned out Goldman was mostly right and ABN was mostly wrong. Good for Goldman! You're supposed to be better at evaluating risk than your counterparties, that's why they're your counterparties. As a second cut, though, not so much: Between the $550 million fine, the bad press and the legal bills for Fab's trial, this no longer looks like such a great trade for Goldman. And the fact that Goldman sold ABN another way to feel safe about the trade, but didn't actually make it safe, is not great for Fab either.

* Though I guess Felix Salmon did, three years ago. New to me! The first comment on that post is particularly worth reading

** By the way this doesn't really help Fab's argument. It is true that ABN buying credit protection on ACA suggests a "focus on ACA credit risk," but the fact that the "directly proportionate" amount of protection it bought was only 3 percent of the potential amount due suggests that ABN was untroubled by the risk on the mortgages.

*** "We bought $30 million of CDS on ACA for 50 basis points, why should we now buy $800 million for 500 basis points, they won't really default, this will all blow over soon," etc.