Questions arise about Goldman's blowout quarter

(Fortune) -- The glitter is already coming off Goldman Sachs' golden quarter.

Goldman (Charts, Fortune 500) wowed just about everyone when it reported very strong earnings for its fiscal third quarter, a period when rival investment banks did poorly because of the steep downturn in bond markets, from which investment banks try to generate trading profits.

More from Fortune

Will Mmmhops be a hit? NBA confirms L.A. Clippers sale to ex-Microsoft CEO Steve Ballmer FBI and SEC probe into Carl Icahn and golfer Phil Mickelson

FORTUNE 500 Current Issue Subscribe to Fortune

However, Goldman's blow out quarter benefited from large gains in hard-to-value financial instruments, and its trading results in the period were particularly volatile, according to data contained in a Goldman filing of quarterly financial results with the Securities and Exchange Commission.

Goldman's stock has gained 13% since its earnings came out, as investors have bought into the notion that the bank is a cut above its peers and is able to weather, and even profit from, tough market conditions.

But that view could get revised, now that it can be seen in the numbers that a large proportion of its third quarter profits were 'unrealized' - i.e. paper gains, and not hard cash payments from fully closed out trades - and came from financial instruments that Goldman values largely according to its own estimates.

"The opaqueness of Goldman's balance sheet makes us immediately question how they made money in the quarter," says Charles Peabody, analyst with Portales Partners.

Friday, Goldman stock was up $3.54 , or 1.56%, to $232.55.

So what do the numbers actually say?

Much of the focus is on Goldman's trading revenue, which totaled a spectacular $8.23 billion, up 70% on the year-earlier quarter. Part of that increase was due to a bold bet that made money if mortgage-backed bonds and financial instruments tied to mortgage values fell in price. Of course, because of the credit crunch, they did plunge in value, netting gains for Goldman that the banks said "more than offset" the losses it saw on the mortgages it was holding.

It's impossible to trace exactly how that bet against mortgages was made, but the financial filing does describe some very telling details about what made up the enormous $8.23 billion of trading revenue.

The interesting data comes from disclosures in the filing about 'level 3' assets and liabilities, which are securities and derivatives that can't be valued according to observable prices in liquid public markets. Because of their illiquidity, Goldman has to attach values to them chiefly according to in-house models and estimates.

Investors typically prefer banks to make money from liquid assets and liabilities that trade regularly because they have greater confidence that they are valued on the balance sheet at their real worth. This is why level 3 gains have recently become a hot topic for the brokerages, and it is a subject Fortune has looked at closely.

And Goldman reaped huge gains within the level 3 pot in the third quarter. For example, it made a net gain of $2.94 billion from level 3 derivatives, financial instruments whose value is based on the value of underlying securities. And get this: $2.62 billion of that gain was unrealized. Was that amount unrealized because there's no way illiquid level 3 derivatives could be cashed out at the prices Goldman attached to them?

"Common sense tells me that a lot of their losses were real and a lot of their gains were paper, and that's something we'd like to know more about," says Portales' Peabody.

Indeed, if that level 3 derivatives gain does include the stupendously prescient bet against mortgages, it deepens the mystery over what type of institution is on the other side of that trade, effectively holding the losses. In other words, if hedge funds - which operate with thin capital and high leverage -- are on the other side of a large part of this mortgage bet holding the losses, it may not be easy for Goldman collect all it is owed.

Asked about the derivative gain, Goldman spokesman Lucas van Praag responded that the level 3 derivative gains "did not come from level three inputs," but from "observable" data taken from more liquid markets.

Why not classify the derivatives in the theoretically more liquid level 2 and level 1 pools, then? "The rules preclude us from doing so," says van Praag.

Okay, let's say Goldman does end up making cash gains from all its trading gains in the third quarter. How likely is it that the bank can do it again?

That's the question Bernstein analyst Brad Hintz is asking after looking at the filing. Often, aggressive trading strategies that result in large gains can also lead to large losses.

And Hintz notes that in the third quarter, Goldman Sachs lost money on just over 25% of their trading days, despite having the most diverse trading business on Wall Street (diversification is supposed to reduce volatility of returns). That 25% number is substantially above the average annual number of 17.5% for 2002 through 2006.

Van Praag says this particular 'loss day' approach is unreasonable. It doesn't make sense to compare one quarter's loss days with annual numbers for loss days, since volatility is dampened over time, and he notes that the third quarter was a very volatile period for every broker.

Hintz isn't convinced, responding: "Goldman Sachs has increased financial leverage, added illiquid assets and has the highest percentage of level three assets in the industry. This might also explain why trading volatility would logically increase."

All that glitters...