Christopher Langner is a markets columnist for Bloomberg Gadfly. He previously covered corporate finance for Bloomberg News, and has written for Reuters/IFR, Forbes, the Wall Street Journal and Mergermarket.

Goldman Sachs has been loathed and celebrated for its market insights, which have probably made it billions of dollars even in times of crises. Lately, however, the Wall Street bank has been getting it wrong...a lot. Has it lost its Midas touch? Perhaps. But in its defense, it's been as hard as ever to predict where the markets are going using traditional tools.

The firm recently abandoned five of its six top trade calls for 2016. Last year, its top pick for Indian junk dollar-denominated bonds -- the 2020 notes of home builder Lodha Developers -- slipped 3.4 percent. Goldman's rationale for the call may have yielded a better result in a more normal environment.

Take inflation forecasting, one area the bank mistook, as Bloomberg News' Rachel Evans and Andrea Wong reported. This should be a fairly straightforward exercise for economists aided by supercomputers and spreadsheets. It isn't anymore though. The monthly print for the U.S. consumer price index hasn't swung this much since at least the 1960s:

Price Swings
The three-month volatility of monthly U.S. inflation numbers has reached the highest since at least 1960
Source: Bureau of Labor Statistics, Bloomberg

Blame it on a more integrated world economy that makes forecasting any macroeconomic measure a great deal more complex. A slowdown in China's northern Hebei province now has a more immediate and intense effect on prices at a Wal-Mart store in Oklahoma than ever before.

It also means traditional correlations aren't what they once were. Goldman didn't see a recent bout of dollar weakness coming and bet that the euro and the yen would weaken against the greenback. It's not hard to see why that should be the case. As the Fed is raising rates, the Bank of Japan just turned them negative for the first time while the European Central Bank is widely expected to open the spigots wider. Such policy divergence should, in theory, benefit the U.S. dollar as global investors shift their attention to higher-yielding American assets. 

Right now, however, the traditional correlation of the trade-weighted dollar index to yields on 10-year U.S. Treasuries has reversed. Instead of strengthening when Treasuries pay more, the dollar has weakened:

Bottom Up
The traditional pairing of U.S. government bond yields and dollar strength broke last year

The euro, meanwhile, has taken the opposite path:

Top Down
Against history, the euro has strengthened even when U.S. government bond yields rose in the past year
Source: Deutsche Bank, Bloomberg

Goldman's bet on Italian bond yields versus those of the euro zone's largest economy, Germany, has also been upended by correlations that haven't been seen since 2008. So much for generating alpha from idiosyncrasies. It seems the same market imbalances Goldman used to pride itself in spotting have come back to bite it.

Inside Out
Italian and German government bonds haven't been moving so closely since 2008

But while today's headlines put Goldman in it, the shop isn't alone in its missed forecasts. At the end of November, HSBC predicted that emerging-market bonds would be a good investment in 2016 and highlighted local-currency debentures from Brazil and Indonesia as the best buys. Brazilian real 10-year government bonds have lost 1.8 percent so far this year and similar notes in Indonesia are down 8.3 percent.

Last May, Sanford C. Bernstein analysts said an oil supply deficit would push crude to $80 a barrel. Everyone knows what happened there. 

Numerous studies have shown that even throwing darts at stock names on a board may produce a better result than following professional investment advice. Surely that should have changed as smart brains are aided by even smarter scenario-analysis tools? 

The limitation they face, however, is that a new world order is emerging, unlike anything seen in recent history. And in spite of all the disclaimers that say they're not guarantees of gains going forward, any financial forecast uses past data to conjure visions of the future.

Regardless, good analysts are supposed to start by checking if old assumptions still hold true. In this environment, that's clearly false.

This column does not necessarily reflect the opinion of Bloomberg LP and its owners.

To contact the author of this story:
Christopher Langner in Singapore at

To contact the editor responsible for this story:
Katrina Nicholas at