2. Macro as a filter, not as a prediction

Macro 6: From macro to strategy

This is the most important module in the course and also the least intuitive, because it contradicts the reason almost everybody studies macroeconomics in the first place. People study macro to know what is going to happen. And it turns out that knowing what is going to happen is a business with a dismal track record, while there is another use for macro, far less glamorous, that does add value consistently.

Forecasters have a poor record, and not for want of talent

The psychologist Philip Tetlock spent two decades recording specialists' predictions about political and economic events and then scoring them against what actually happened. The sample ran to hundreds of experts and tens of thousands of forecasts. The result, put plainly: average accuracy barely beat what you would have got by assigning probabilities blind, and it was frequently worse than elementary extrapolation rules. And an even more uncomfortable finding: the confidence with which an expert pronounced was negatively related to how often they turned out to be right. The most quoted in the press were, on average, the least reliable.

It is not incompetence, and it is not bad faith. Institutional forecasters have the best information available and enormous teams, and even so the economists' consensus has almost never called a recession a year ahead, and in 2021 inflation was described as transitory for months.

Why it fails, and why it will carry on failing

There are three structural reasons, and none of them is fixed by trying harder:

• The price already embeds the consensus. As you saw in level 3, the market does not react to the data but to the gap between the data and what was already priced in. To make money it is not enough to get CPI right: you have to be right about where everybody else is wrong.
• You need to be right twice, not once. Even if you guess the number, you still have to guess the reaction. There are days when an excellent payrolls print sinks the stock market and days when it lifts it, depending on what is worrying the market that week.
• The economy is a reflexive system. Andrew Lo describes it as an adaptive environment: participants change their behavior in response to what they learn, so regularities erode as they are discovered. A planet does not change orbit because you have worked it out; a market does.

The question that does have an answer

Change the question and macro turns from a crystal ball into a tool. Instead of what is going to happen, ask what am I exposed to. The first demands guessing the future; the second, understanding what you already hold, which is information available today and does not depend on luck.

The practical difference is enormous. Forecasting gives you a position and no defense when it fails. Knowing your exposure gives you the list of things that have to happen for you to lose money, and with that list you can size properly, avoid stacking the same bet three times over and spot a change of regime sooner than someone who only watches their equity curve.

Every strategy is a macro bet, whether you declare it or not

This is the heart of the module. Almost no strategy is macro-neutral, however firmly its author swears they do not do macro:

• A growth screen is a position in long-duration assets, and therefore a bet that rates do not rise (level 1).
• A portfolio of banks is in large part a bet on the slope of the curve and on the credit cycle: the business consists of borrowing short and lending long, so a flat curve eats the margin (levels 3 and 5).
• Selling volatility is an explicit bet that there will be no shock: you collect a small, steady premium in exchange for taking on a huge, concentrated loss.
• Buying high, stable dividend payers is, in sensitivity terms, buying something rather close to a long bond dressed up as a share.
• Currency carry pays while the calm lasts and hands back several years of gains in a week when it breaks (level 4).
• Trend following is a bet that regimes persist; it works badly at exactly the sharp turns.

Antti Ilmanen has argued for years that the expected returns of apparently different strategies come from a small handful of common risk factors. The practical consequence is direct: if your five strategies share the same factor, you do not have diversification, you have concentration in disguise, and you will find out on the day correlations go to one (level 5).

What to do with this, concretely

• Write each strategy's exposure in a single sentence beginning 'this makes money as long as...'.
• Add those sentences up and see whether they repeat. If they all say the same thing in different words, you now know what your real risk is.
• Size by exposure, not by conviction.
• Decide in advance which macro observation would make you cut or close. Writing it down beforehand is what stops you reinterpreting it afterwards.

A real case: February 5, 2018

In the preceding years, selling volatility had been one of the most profitable trades in the market. There was a listed product, the ETN known as XIV, which tracked the short volatility position and which more than doubled in value during 2017. Any historical check from 2012 to 2017 gave a spectacular result.

On February 5, 2018 the VIX index went from around 17 to around 37 points in a single session, the largest daily percentage rise in its history. XIV lost roughly 96% of its value and the issuer wound it up a few days later. Nobody predicted the date, and nobody needed to: anyone who had written down their exposure in one sentence would have had the exact shape of the ending in front of them. I make a little almost every day and I lose the lot on the day there is a shock. That sentence requires forecasting nothing. It only requires having written it.