Losing Investment Models

Ever since the financial crisis, and especially in the aftermath dominated by the Fed’s quantitative easing, I’ve noticed that old investment models generally don’t work. There are many mutual funds and investment firms that built models that worked very well for a long time. Some had models to pick stocks or bonds, Others had models that told them how to allocate their portfolios among different assets. But those models were based on decades of market and economic data from a completely different environment. The Fed’s policies distorted markets since 2008. The old relationships between inputs and outputs don’t work. Only a few managers I’ve seen that use models have done well, and those managers adjusted their models for the new environment.

Here’s a more detailed example. The Bloomberg article discloses talks with bond market managers who relied on models and found they don’t work any more. I’ve always been suspect of purely quantitative investment methods. Data and formulas are helpful. But markets are dynamic and evolving. You have to notice changes, understand them, and adjust models appropriately.

Just last month, researchers at the Federal Reserve Bank of New York retooled a gauge of relative yields on Treasuries, casting aside three decades of data that incorporated estimates for market rates from professional forecasters. Priya Misra, the head of U.S. rates strategy at Bank of America Corp., says a risk metric she’s relied on hasn’t worked since March.

After unprecedented stimulus by the Fed and other central banks made many traditional models useless, investors and analysts alike are having to reshape their understanding of cheap and expensive as the global market for bonds balloons to $100 trillion. With the world’s biggest economies struggling to grow and inflation nowhere in sight, catchphrases such as “new neutral” and “no normal” are gaining currency to describe a reality where bonds are rallying the most in a decade.