Friday, October 26, 2012

You're Projecting

One thing that humans are notoriously bad at is predicting the future. And the reason is very simple: they tend to project the present into the future. So if you're only predicting what might happen tomorrow, there's a 98% chance that the following will happen:

1. ~40 billion dollars worth of goods and services will be produced.
2. ~6500 people in the US will die.
3. The S&P 500 will lose or add less than 2.5% from its current value.

How did I arrive at those numbers? For the first two, I simply took the official estimate for GDP and deaths per year and divided it by 365. The third number? I looked at a distribution of single day stock market changes.

In other words, I looked at what happened in the past and then projected it into the future. But even if I'm right 98% of the time, it's the 2% of the time that I'm wrong, when something doesn't go as expected, that really changes the whole equation.

Think about it from the perspective of a teenage driver. Just about all of them read and write texts while driving.The vast majority of times that they do it, nothing out of the ordinary happens. But the one time out of a million that they do it and get into a fatal car wreck, their life is permanently altered. And suddenly, their life trajectory changes from one likely scenario to another.

This is what drives me absolutely insane about studies on proposed policy in Washington. All they do is project the present into the future, making a series of assumptions that are unlikely to hold up over any reasonable period of time. It's easy to say, "oh if we just raise/reduce rates by a certain amount, we can expect revenue to change by this amount over 10 years", but it's highly unlikely that that will be the case. We simply don't know how things are going to change in 10 years.

The difference between 2% annual growth and 3% annual growth is substantial. In relative terms, it's a pretty big change. In absolute terms? Not at all. These kinds of changes are, from day to day, relatively small but they have an incalculable effect on various people. Those intangible effects become tangible realities. The game changes and then our models, based on things that happened in the past, get wrecked.

If you ever look at an investment prospectus, the number one thing that they try and impress on you is this: past performance is no guarantee of the future. That phrase, or some variation of that, appears at least 50 times in any Federally regulated prospectus. And yet, we can't help but look at past performance to gauge how the stock will do in the future. Even the people who set up the investment fund do the exact same thing.

Long Term Capital Management was a hedge fund that traded based on a complex mathematical model based on the history of asset price movements. For a period of 3 years in its 5 year history, it produced outsized returns on investment. But the fund completely collapsed 2 years after that. Why? The Asian Financial Crisis happened and made price movements happen in such a way that the model couldn't predict.

The model then recommended actions that proved financially disastrous in the new reality, and it took a consortium of Wall Street's biggest banks to sort through the ruins. Whenever a person develops a trading algorithm, the first thing they do is test the algorithm by feeding it historical data on the stock market. Most of these algorithms tend to do very well, until they don't. And then they get completely screwed up by an unplanned event and everything they worked for gets wiped out in the space of 2 weeks of bad trading.

This is why life is so unpredictable. Because it only needs to get messy once before your life gets permanently altered, even if the vast majority of time, things proceed as planned. But as generals are always eager to fight the last war, governments are always eager to regulate the last problem.

Collectively, we as a society have to accept the fact that shit happens. It's such an immediately obvious truism but we don't follow it. Enron and Worldcom happen and we get Sarbanes-Oxley. The 2008 financial crisis happens and then we get Dodd-Frank. Modern society seems intent on trying to remove as much uncertainty as possible by regulating every possible action. That is ultimately a futile and self defeating task, as new rules inevitably beget new rule evaders and changes the incentives and actions of millions of people.

Ideally, government should operate by a fixed, transparent, and universally recognized set of rules. And when bad things happen, sometimes they have to shrug their shoulders and say "tough luck, kid". If that bad thing happens repeatedly, then it might be worth regulating at a later time when heads are cooler and minds are clearer. Reactionary policy is usually the worst type of policy.

No comments:

Post a Comment