Some of you might be familiar with ZeroHedge, a stock analysis site that until very recently, was regarded as something of a tinfoil-hat haven.

That is, until they were at the vanguard of the coverage of the theft of Goldman Sachs' automated trading platform software, which was apparently frontrunning their client traders by executing Goldman's orders slightly ahead of the smaller frys' orders, and pocketing the resulting rise in prices as pure profit.

As an example, they were using their advance knowledge of, say, a large hedge fund's desire to buy 10,000 shares of MSFT (Microsoft), to then place a 100,000 share order, which their computers executed ahead of the hedge fund's order. Since stock prices go up when people are buying, Goldman's order then caused the price to rise, and the hedge fund basically paid an extra commission directly to Goldman for the 10K shares of BillCo.

The ZeroHedge coverage of that event, ahead of Goldman's admission that the software in question had indeed been comprised and stolen, and was "out there" in the wild, got Wall Street's attention, and today the site is enjoying an increasing level of respect not just from "bankster" conspiracy theorists, but also regular traders and the financial press.

Well, here they're back with a piece describing how Goldman's 98.5% success rate on its stock picks is very clearly not a natural phenomenon that can be explained by the brilliance of its analysts, even though those analysts are generally the pick of the litter.