Thursday, April 15, 2021

Are you a Superforecaster?; The Big Lessons of the Last Year; Job Training That's Free Until You're Hired; A Vast Web of Vengeance

By Whitney Tilson

1) One of my all-time favorite investing books is Philip Tetlock's Superforecasting: The Art and Science of Prediction.

Even though it's not technically about investing, the book's findings on why some people are able to better predict the future than others relate directly to investment success.

So I read with great interest this e-mail I received recently about Tetlock's free website, Good Judgment, which allows anyone to test their forecasting skills:

Our goal, then and now, has been a simple one: to give aspiring forecasters a place to improve their foresight skills and to interact with each other. In Prof. Tetlock's words:

To get better at this, you need to believe that forecasting is a skill that can be cultivated – and is worth cultivating. That is step 1. 

Step 2 is training and practice and timely feedback – and then more of it.

That's exactly what we offer on Good Judgment Open.

Making predictions

Every week (mostly on Fridays) GJOpen launches forecasting questions about current events in politics, economics and finance, science and technology, public health, sports, popular culture, and other fields. A typical question asks, "Will X happen by date Y?" where X is an outcome such as the result of an election, two countries reaching a trade agreement, a key development in a new industry, or a sports team winning a championship. 

Our forecasters make their predictions about these events using the same language that meteorologists use when making weather forecasts, the language of probabilities; such as there is a 60% chance that an incumbent politician will win re-election, or there is an 80% probability that an automaker launch a new vehicle by date X. This enables us, among other things, to measure forecasting accuracy using Brier scores.

I decided to sign up and answered a number of questions like these (here's a full list):

  • Before 1 May 2021, will it be officially announced that the Tokyo 2020 Summer Olympics and/or Paralympics will be canceled?
  • When will the Transportation Security Administration ("TSA") next screen two million or more travelers in a single day? 
  • Will the percentage of Global Business Travel Association survey respondents planning to resume international business travel in the near future (next 1-3 months) reach 20% or higher before September 2021? 
  • When will SpaceX's satellite internet service, Starlink, begin offering commercial service in North America?
  • For any seven consecutive day period between 9 October 2020 and 15 June 2021, will there be fewer than 50,000 combined total confirmed new cases of COVID-19 in the United States? 
  • Will there be a complex coordinated terrorist attack ("CCTA") in the United States before 1 September 2021?
  • Before 1 January 2022, will the U.S. Senate expand the scope of matters for which a filibuster cannot be used?
  • What will be China's year-on-year GDP growth rate for the second quarter of 2021?
  • What total percentage of global vehicle sales in 2022 will be battery electric vehicles ("BEVs") and plug-in hybrid electric vehicles ("PHEVs")? 
  • What will be the delinquency rate on commercial real estate loans (excluding farmland) in the United States in the second quarter of 2021? 
  • Which will happen next regarding the price of a bitcoin ($25,000 or less, $100,000 or more, or neither by July 1)? 
  • How many total Model 3 and Model Y vehicles will Tesla (TSLA) deliver to customers in 2021 and 2022?
  • What will be the 2022 industry-wide average cost of Li-ion batteries used in battery-powered electric vehicles?
  • Before 1 January 2023, will Amazon (AMZN) or a subsidiary use a Zoox autonomous vehicle to deliver a package and/or for a public ride-hailing service?
  • When will a Carnival Cruise Line (CCL) cruise next depart from the continental U.S.?
  • What will be the closing yield for the 30-year U.S. Treasury on 31 December 2021? 
  • What will Uber's (UBER) end-of-day market capitalization be on 31 December 2021?

If you're interested in trying it out, I suggest watching this video, Introduction to Forecasting (5:01), and this one, which describes how Good Judgment keeps score.

2) I enjoy the writings of blogger and venture capitalist Morgan Housel, who last year published an excellent book, The Psychology of Money: Timeless Lessons on Wealth, Greed, and Happiness. Here's his latest blog post: The Big Lessons of the Last Year. Excerpt:

History is one damned thing after another. A war ends, a boom follows, then a crash, then an uprising, then a pandemic, a breakthrough, a new boom, a new war. On and on, from agony to awe. 

A question that always arises after a terrible event is why haven't we learned our lesson?

Financial crises keep happening, again and again. People have been making the same investing mistakes for hundreds of years. The same military blunders, over and over. The characters change but the plot rarely does. 

Jason Zweig explained years ago that part of the reason the same mistakes repeat isn't because people don't learn their lesson; it's because people "are too good at learning lessons, and they learn overprecise lessons." 

A good lesson from the dot-com bust was the perils of overconfidence. But the lesson most people took away was "the stock market becomes overvalued when it trades at a P/E ratio over 30." It was hyperspecific, so many of the same investors who lost their shirts in 2002 got up and walked straight into the housing bubble, where they lost again.

The most important lessons from a big event are usually the broad, 30,000-feet takeaways. They're more likely to apply to the next iteration of crisis.

COVID-19 is far from over, but we're now more than a year into this tragic mess. Enough has happened that we can start to ask: "What lessons have we learned?" If you're a doctor or a health regulator, some of those lessons are hyperspecific. But for most of us the biggest lessons are broad.

A few that stick out: 

  1. Big risks are easy to underestimate because they come from small risks that multiply...
  2. A lot of undue pessimism comes from underestimating how quickly and firmly people adapt...
  3. History is only interesting because nothing is inevitable... 

Whatever your view of the world was a year ago, it's different now. Maybe COVID's biggest lesson is accepting how true that will be going forward, too.

Daniel Kahneman says that when you experience a surprise the correct takeaway is not to assume that event will happen again; it's to accept that the world is surprising. The big lesson is to realize that you will again be hit by things you didn't see coming, that no one was talking about, and that will move the needle more than all the things you expected to happen combined.

Just one damned thing after another.

3) Cheers to Social Finance, which is pioneering a new pay-only-for-results model of job training (full disclosure: the Pershing Square Foundation, on whose board I sit, has funded Social Finance).

Here's a New York Times article about it: Job Training That's Free Until You're Hired Is a Blueprint for Biden. Excerpt: 

American Diesel Training is part of a new model of work force training – one that bases pay for training programs partly on whether students get hired. Early results are promising, and experts say the approach makes far more economic sense than the traditional method, in which programs are paid based on how many people enroll.

Right now, there are only a relative handful of these pay-for-success programs that train low-income Americans for better-paying careers. The challenge has been to align funding and incentives so that students, training programs and employers all benefit.

But Social Finance, founded a decade ago to develop new ways to finance results-focused social programs, is showing how the idea could grow quickly just as the pandemic made job-training programs more important than ever. The coronavirus put millions of people out of work, upended industries and accelerated automation. 

State and federal officials are now looking for new ways to improve work force development. President Biden's $2 trillion infrastructure and jobs plan, announced last week, includes billions for work force development with an emphasis on "next-generation training programs" that embrace "evidence-based approaches."

The Social Finance effort is powered by a fund of more than $40 million raised from philanthropic investors. The money goes toward paying for low-income students, as well as minority candidates and veterans, to enter the training programs.

4) Jeers to the Internet companies who allowed a vengeful woman to wreck the lives of numerous people by posting false and malicious things about them: A Vast Web of Vengeance. Excerpt: 

Mr. [Guy] Babcock, a software engineer, got off the phone and Googled himself. The results were full of posts on strange sites accusing him of being a thief, a fraudster and a pedophile. The posts listed Mr. Babcock's contact details and employer. 

The images were the worst: photos taken from his LinkedIn and Facebook (FB) pages that had "pedophile" written across them in red type. Someone had posted the doctored images on Pinterest, and Google's (GOOGL) algorithms apparently liked things from Pinterest, and so the pictures were positioned at the very top of the Google results for "Guy Babcock." 

Mr. Babcock, 59, was not a thief, a fraudster, or a pedophile. "I remember being in complete shock," he said. "Why would someone do this? Who could it possibly be? Who would be so angry?" 

Then he Googled his brother's name. The results were just as bad.

He tried his wife.

His sister. 

His brother-in-law.

His teenage nephew.

His cousin.

His aunt. 

They had all been hit. The men were branded as child molesters and pedophiles, the women as thieves and scammers. Only his 8-year-old son had been spared. 

Guy Babcock was about to discover the power of a lone person to destroy countless reputations, aided by platforms like Google that rarely intervene. He was shocked when he discovered the identity of the assailant, the number of other victims and the duration of the digital violence.

Best regards,