I’ve never made a secret of my opinion that acting on market forecasts is destructive to investors. Nate Silver’s fascinating new book, The Signal and the Noise: Why So Many Predictions Fail—But Some Don’t includes some telling examples from the world of finance, but he drives home this idea even more forcefully with his insights about the weather.
Silver explains that meteorological forecasts are quite accurate if they’re made just a few days in advance, but the further out you go, the less helpful they become. Forecasts made eight days in advance are useless, and beyond that they’re actually harmful: “They are worse than what you or I could do sitting around at home and looking up the table of long-term weather averages,” Silver writes. Yet despite being aware of this evidence, The Weather Channel and AccuWeather make forecasts for 10 days and 15 days into the future, respectively.
The book also describes how for-profit weather services are more concerned with the perception of accuracy than with accuracy itself. This gives them an incentive to be bolder than they should be. If their models forecast a 50% likelihood of rain, “which might seem wishy-washy and indecisive,” they are likely to report either a 40% or 60% chance. Doing so makes the forecast less accurate, but it sounds more confident than simply calling it a coin flip.
Finally, and most troubling, commercial weather forecasters tend to forecast precipitation more often than they truly expect it to occur. When the government-run National Weather Service forecasts a 20% likelihood of precipitation, it really does rain about one time in five. But historically, when The Weather Channel calls for a 20% chance of rain, it only rains about 5% of the time, Silver writes. This is not because their models are less accurate: it’s because the numbers are fudged intentionally. Meteorologists call this the wet bias, and Silver says it’s the “worst-kept secret in the weather industry.”
Why do forecasters have a wet bias? Because people don’t mind being surprised with good news, but they get angry when reality turns out worse than expected: “If it rains when it isn’t supposed to, they curse the weatherman for ruining their picnic, whereas an unexpectedly sunny day is taken as a serendipitous bonus.” In other words, the companies tell people what they want to hear, rather than what they know to be true. “Forecasts ‘add value’ by subtracting accuracy,” Silver writes.
The lessons for investors
There are some lessons here for investors who listen to the forecasts of market gurus, because many of the same economic incentives and human biases come into play.
For-profit weather services issue forecasts with negative value because people will pay for them, regardless of their inaccuracy. In the same way, financial forecasts have a terrible track record that’s obvious to anyone who looks at the evidence, yet the urge to know where markets are headed is so strong that investors instinctively listen to anyone who claims to have this knowledge.
Second, the financial industry loves to sell the perception of accuracy with their risk models. I’ve seen tactical portfolios calling for a 19.1% allocation to this asset class, or a 1.4% allocation to another, as if this kind of precision were meaningful. But a well-diversified portfolio is not a chemical formula: only the broad strokes matter. No one needs an asset allocation with decimal points except research teams who are trying to impress you.
Finally, the “wet bias” among active money managers has been rampant since the financial crisis. Many have been calling for rain the whole time, whether it’s the secular bear market that won’t go away, a bond Armageddon caused by rising interest rates, a Canadian housing bubble, or any number of other calamities. Like the weather forecasters, they know investors feel losses about twice as acutely as they feel gains. That means we’re unlikely to berate our financial advisor for playing defense at the wrong time, but we get angry when she doesn’t “know when to get out” and avoid losses. Yet the fact is, anyone who avoided equities or shunned bonds has missed a bull market in both asset classes that is now almost four years old.
Except in extreme cases, we rarely stay at home because of weather forecasts: it’s best to simply plan ahead by dressing appropriately and carrying an umbrella. Investors can do the same by building a diversified, all-weather portfolio that will hold up in every season.
The lessons can be extended further by examining the difference between weather and climate. Weather is impossible to predict with any useful degree of accuracy one year ahead, but climate can be predicted with relative ease. We can say with confidence that next winter will be colder than next summer, for example.
The climate is changing in much of the world, but because climate and weather are naturally variable it’s hard to see the signal when you’re living in the noise of year-to-year variation. An exceptionally warm year might be followed by an exceptionally cool year, even as the global temperature overall is rising. Over the long term, though, the trend is toward warmth: the curve is rising. We can predict with high confidence that the global average temperature in 2050 will be higher than it is today, even if we wouldn’t dare to try to predict the weather for 28 December 2050 today.
Similarly, investments in index funds can be volatile from year to year, but over the long term the trend is toward growth. We can say with a fair degree of confidence that a balanced portfolio of stocks and bonds will be worth considerably more in 2050 (or even 2020) than it is today, even though it would be foolhardy to predict the value of any individual stock or ETF that might still exist in 2050.
The bottom line is that short-term behaviour in markets and weather is hard to predict, but long-term trends can be predicted with relatively high confidence.
http://socioecohistory.files.wordpress.com/2010/01/time-magazine-april-1977.jpg
Check out the cover of this issue of Time magazine from 1977. Global cooling was the concern at the time. It appears we cannot recognize a long term trend. The media can not be our source information. There are too many conflicts. Long term data is the only true source of information. Unfortunately most do not have the patience to stay the course.
This is a brilliant article. Thank you Dan. I’m a big Nate Silver fan too.
@Kiyo: Thanks, glad you liked this. I enjoyed the book, especially the way Silver made connections between disparate fields like investing, poker, baseball and climatology.
Thanks for this, Dan. You’re a great source of calm perspective. I’m going to have to get the Silver book — investing, poker, baseball, climatology… I’m interested in all four of those topics!
Another very useful article and Brads analogy is right on.
As an empiricist by trade I really appreciate Dans comment about the silliness of the financial industrys use of precision – garbage in, garbage out is a precept of modelling.
Yet precision implies confidence and authority over the data.
Actually I’m not so sure my analogy is right on, because future climate depends on physics whereas future stock prices depend ultimately on human behaviour. The latter is much harder to predict than the former ;-)
Actually, on a macro level, future climate has a lot to do with human behaviour. We are producing all that carbon and warming the planet!
@Noel: yes, but much of the warming over the next 30-40 years has already been committed: if we stopped emitting CO2 today the atmosphere would continue to warm because there’s a time lag involved. The amount of warming that occurs in later years (toward 2100) depends much more strongly on the emissions scenario you assume, but over the next few decades the expected warming doesn’t differ too dramatically from scenario to scenario. In the near to mid term, it pretty much all boils down to physics.
Whether we understand the climate system better than we understand how economies work is another question, of course.
the earth may keep warming. co2lags earth temps a hundred years or more. rising temperature increases co2. gore forgot to mention that part. ha
@Dale: I don’t mean to hijack this thread into a debate on global warming, but your argument is outdated. A study published in Science in 2012 showed that 90% of global warming during the last glacial-interglacial transition occurred after the CO2 increase, not before it as previously thought.
Silver’s book has a whole chapter on global warming, and he correctly raises questions about the accuracy of climate model projections. But some prominent climatologists took issue with a few of his conclusions and pointed out some inaccuracies in his arguments; see for example http://www.huffingtonpost.com/michael-e-mann/nate-silver-climate-change_b_1909482.html.
I forgot to provide a link summarizing the Science paper on CO2 leading vs. lagging temperature increases: http://www.skepticalscience.com/co2-lags-temperature.htm
I recall at least one study (there are several) that tracked the accuracy of predictions by financial and stock market experts. A recent study conclude that experts’ predictions were no better (and no worse) than the weather is as a predictor for markets. Does anyone know of this study and its source? Would much appreciate any info. THANKS