Out-Of-Sample Risk Is (Always) The Biggest Threat To Expectations

In the 1974 film “Chinatown,” Noah Cross tells Jake Gittes, the nosy detective: “You may think you know what you’re dealing with, but believe me, you don’t.” The advice isn’t out of line for investors, portfolio managers and economists who’ve studied economic and market history and assume that the future will play out with similar twists and turns by dropping robust signals ahead of the next crisis.

Case in point: the current blowback triggered by the coronavirus. Let’s start with the economy. At this point it’s unclear if the moderate expansion that prevails (based on data published to date) will continue. But let’s assume the US falls into recession at some point in the near future, a forecast that’s gaining traction among analysts. If a coronavirus-triggered downturn is fate, the shift could be one of the more rapid changes in the macro trend in history.

Consider that recent economic indicators are still reflecting that the moderate pace of growth reported in last year’s fourth quarter remains on track to continue in Q1. Payrolls data for February, for instance, suggests that forward momentum remains solid. The big picture trend, not surprisingly, remains upbeat, based on using the latest numbers to anticipate the near-term outlook. The Atlanta Fed’s GDPNow model, for example, was especially upbeat in the March 6 update: output is expected to rise 3.1% in the first three months of this year, a solid acceleration in growth from Q4’s 2.1% gain.

Other nowcasting models are projecting softer growth, but overall most models are still pointing to an ongoing expansion in Q1 that will sidestep recession.

The question is whether the coronavirus represents an exogenous shock that will derail the macro trend? “We’ve reached the tipping point where things are feeding off of each other,” advises Julia Coronado, president of MacroPolicy Perspectives. “We’re going to lose a chunk of activity and then we’ll grow out of it. That’s the good news. But are we going to boom out of it or crawl out of it? Crawling is looking more likely.”

Is Recession Risk Rising? Monitor the outlook with a subscription to:
The US Business Cycle Risk Report

The stock market in the throes of a sudden wave of discounting softer and perhaps negative growth in the near term. The rapid fall from grace is among the sharpest reversals on record. A few weeks ago (Feb. 19), the S&P 500 closed at a record high. As of yesterday (Mar. 9), the S&P is down nearly 19% from that peak – a stunning reversal of fortunes in just 13 trading sessions.

When the S&P last  month first posted a 10% drawdown in the current selling wave, the decline marked the fastest 10% correction on record.

To be fair, there have been warning signs for the economy and the market all along, including an inverted Treasury yield curve (a widely respected warning for the economy) and a stock market that, until recently, seemed to defy gravity and prudent valuations with non-stop exuberance. There are, of course, always warning signs in some corner or another and so the challenge was deciding if limited threats were turning into a larger problem. The answer, we know now, is “yes,” but for reasons few of us expected.

Before the coronavirus become everyone’s obsession, as recently as a month ago, you could be forgiven for thinking that the economic risk remained low and the US stock market’s biggest threat was a garden-variety correction. A careful study of past crises, recessions and market corrections offered minimal reasons to think otherwise. The in-sample testing, in other words, offered convincing support for expecting that risk in the broad sense was low.

But the economic gods and Mr. Market seem to be forever conspiring to fool the crowd by coming up with new ways to deliver unexpected confusion and chaos. If a coronavirus-based bear market and recession is upon us, the decline and fall is destined to be the new poster child for out-of-sample risk.

How many robust modeling efforts and careful reviews of history have been torn asunder in the current crisis? The failure rate is surely high. Par for the course. Just as generals are always fighting the last war, investors and economists are prone to relying on the historical record to divine the future and assess the state of risk. But the future rarely unfolds as expected.
We can debate if the coronavirus is a black swan risk (an unforeseeable event) or a gray rhino (an obvious but neglected risk). Whatever you call it, the current crisis has caught many if not most of the world’s governments, policymakers, economists and investors by surprise. In the process, many if not most models that appeared robust have been damaged, perhaps fatally in some cases.

A familiar theme, however, remains intact: heavily if not blindly relying on history for guidance on the future has hit a wall—again. That crunching sound you hear is the breaking of elegant in-sample modeling that’s groaning under the weight of out-of-sample results.

The future remains uncertain and so there are no formal solutions to out-of-sample risk. But there are degrees of oversight and error and some of the more egregious blunders are currently in the process of revealing themselves.

The lesson: we can all do better and presumably some of us will, in part through studying how the coronavirus blowback surprised us. But human nature being what it is, most of the lessons will be ignored.

Perhaps the bigger danger is that some of us will continue to assume that working harder and going deeper with in-sample testing will remove all the uncertainty and risk that awaits in the out-of-sample future.

Don’t fall into that trap. We can start with maintaining a bit of humility when it comes to modeling the markets and the economy. But as the world has rediscovered (again), absorbing this wisdom is, for most of us, a cyclical rather than a cumulative process.

Learn To Use R For Portfolio Analysis
Quantitative Investment Portfolio Analytics In R:
An Introduction To R For Modeling Portfolio Risk and Return

By James Picerno

One thought on “Out-Of-Sample Risk Is (Always) The Biggest Threat To Expectations

  1. Pingback: Current Blowback Triggered by the Coronavirus - TradingGods.net

Comments are closed.