The company's chief investment office originally made a series of trades intended to protect the firm from a possible global slowdown. JPMorgan owns

 
billions of dollars in corporate bonds, so if a slowdown were to occur and corporations couldn't pay back their debt, those bonds would have lost value.

To mitigate that possibility, JPMorgan bought insurance - credit-default swaps - that would go up in value if the bonds fell in value.

But sometime last year, with the economy doing better than expected, the bank decided it had bought too much insurance. Rather than simply selling the insurance, the bank set up a second "hedge" to bet that the economy would continue to improve - and this time, traders overshot, by a lot. [1]

There were two dimensions to the cause of the current credit crisis. The first was that unit risk was not eliminated, merely transferred to a larger pool to make it invisible statistically. The second, and more ominous, was that regulatory risks were defined by credit ratings, and the two fed on each other inversely. As credit rating rose, risk exposure fell to create an under-pricing of risk. But as risk exposure rose, credit rating fell to exacerbate further rise of risk exposure in a chain reaction that detonated a debt explosion of atomic dimension.

The [US] Office of the Comptroller of the Currency and the Federal Reserve jointly allowed banks with credit default swaps (CDS) insurance to keep super-senior risk assets on their books without adding capital because the risk was insured. Normally, if the banks held the super-senior risk on their books, they would need to post capital at 8% of the liability.

But capital could be reduced to one-fifth the normal amount (20% of 8%, meaning $160 for every $10,000 of risk on the books) if banks could prove to the regulators that the risk of default on the super-senior portion of the deals was truly negligible, and if the securities being issued via a collateral debt obligation (CDO) structure carried a Triple-A credit rating from a "nationally recognized credit rating agency", such as Standard & Poor's rating on AIG.

With CDS insurance, banks then could cut the normal $800 million capital for every $10 billion of corporate loans on their books to just $160 million, meaning banks with CDS insurance can loan up to five times more on the same capital. The CDS-insured CDO deals could then bypass international banking rules on capital. To correct this bypass is a key reason why the government wanted to conduct stress tests on banks in 2009 to see if banks need to raise new capital in a Downward Loss Given Default.

CDS contracts are generally subject to mark-to-market accounting that introduces regular periodic income statements to show balance sheet volatility that would not be present in a regulated insurance contract. Further, the buyer of a CDS does not even need to own the underlying security or other form of credit exposure. In fact, the buyer does not even have to suffer an actual loss from the default event, only a virtual loss would suffice for collection of the insured notional amount.

So, at 0.02 cents to a dollar (1 to 10,000 odd), speculators could place bets to collect astronomical payouts in billions with affordable losses. A $10,000 bet on a CDS default could stand to win $100,000,000 within a year. That was exactly what many hedge funds did because they could recoup all their lost bets even if they only won once in 10,000 years. As it turns out, many only had to wait a couple of years before winning a huge windfall. But until AIG was bailed out by the Fed, these hedge funds were not sure they could collect their winnings. [2]