Successes and failures of prediction - How to use data and prediction in an increasingly complex world
I recently finished reading Signal and the Noise by Nate Silver. Nate is well known for his website: FiveThirtyEight, which covers sports and politics in a data driven way and has really fantastic visualizations. His website was user pretty heavily in the 2016 presidential election where they aggregated published polls and predicted very publicly who they thought would win the election. (Spoiler: It predicted Hillary Clinton would win by what I thought was a large margin on election day)
It’s one of the best books I’ve read in the last year. Nate covers a wide range of topics in a fascinating way, keeping you hooked even though the subject: how to get signal (insight) from noisy data and where prediction and models fail, does not normally generate the same excitement as a crime novel. Covering:
- credit reporting agencies failure to see the problems that caused the 2008 financial crisis (instead actively masking the risks with their erroneous ratings)
- how television pundits don’t make very accurate predictions and are not held to account
- baseball (Moneyball fans will enjoy this)
- weather prediction - we’ve gotten much better at it than you think
- earthquakes - we are not very good at predicting earthquakes
- sports betting - how betters use scientific thinking to come out ahead in a fast paced world
- the poker bubble of 2005 - how Nate personally participated and came to create FiveThirtyEight
- The stock market - Why there is more trading than is rational in the stock market and how the efficient market theory works.
I’m not a believer in the efficient market theory, but Nate covers it fairly well as it’s believed. It’s a good example of an unprovable hypothesis or pseudoscience, how people can build up lots of fancy jargon and formulas around an idea and still be completely incorrect:
To have the market be unpredictable is separate from all available information being priced into markets, even if it takes longer for those prices to walk to their correct position in a weak or semi-strong version. If prices move randomly, that doesn’t necessarily mean that markets are following some special information based pricing and it has nothing to do with them being hard to beat. For a longer and more thorough refutation of the Efficient Market Hypothesis, check out Irrational Exuberance by Robert Shiller.
Transaction costs are given as a reason most strategies are not profitable to “beat the market”. Transaction costs are very low because of competition, much below 0.25% that Nate cites (Robinhood, Interactive Brokers, etc.), but this might not hold true forever. These products might be priced below their true price and rise later. The major volume creators (High frequency traders) don’t pay 0.25% per transaction and so many strategies are available. Taxes might make up for this someday too, if a 3% tax on investing transactions is created by Bernie Sanders and friends it would make many more “efficiency creating transactions” unprofitable (and making using information to make the market more efficient less possible).
I’m very passionate about investing and I think a lot about doing it better. Nate gives a fair hearing to a lot of investing ideas and after reading the chapter a second time I came away much more positive opinion of it than the first time. For most people, the stock market is the only way they can invest their savings in productive places and beat inflation (outside getting raises above inflation every year). It’s important that people start investing simply with easy to understand index funds, but I’m a bit believer that to make the world truly better, we need to get our hands dirty, do some learning, and figure out how to invest our money and donations in effective ways. Picking productive investments is no different than picking productive charities or building effective and accountable governments. They have a big effect on people around the world. If everyone buys index funds, we are making it so that fewer people determine where our money will go next month based on what they buy or recommend, and we’re all stuck following the indexes off a cliff as they get more and more expensive to decades of below normal returns.
We live in a world with more data everyday. are becoming much more numerous. But more data doesn’t equal more insight, infact, it probably makes finding truth more difficult and our future more challenging. As an example I’ve included a graphic from Wikipedia and from The Economist from this article (https://www.economist.com/briefing/2017/05/06/data-is-giving-rise-to-a-new-economy)
These difficulties with increasing information and less clear signal in many fields unfortunately will probably increase over time. While Nate doesn’t mention the phrase “fake news” in his book, the problems for humans dealing with more and more information over time from more and more voices won’t get any better. As someone who’s very interested in data and discovering what’s true, even if it doesn’t lead to conclusions that are popular, this book was thoroughly enjoyable to read and increased my knowledge on the state the analysis for these problems, many of which have very big impacts on the world.
Overall, the book was engaging, educational, and fun to read. Nate does a good job in FiveThirtyEight style of writing in a way that appeals to a wide audience, but teaching you a lot and making you think at the same time. I was not familiar with weather forecasting and earthquake forecasting before the book and some thoughts of his let me to come up with some cool ideas on investing. It’s not always easy to find datasets to analyze, but his examples create great motivation for tackling some of these big problems, especially at the large organization level. There are many areas where better forecasting can dramatically improve our lives. I’m optimistic that if more people took Nate’s approach to thinking about interesting problems, we’d can make life on earth a lot better for a lot of people.
If you read this book, I think you’ll come away with lots of knowledge and appreciation for the challenges and opportunities that using data and scientific thinking can add to your life. I’m a big fan of Nate’s site and the ideas presented, and I hope in the future we can have more people who take risks, come up with better insights and bets that help us better use our resources, and predict and hedge against disasters so not so many people suffer from preventable problems.
More reviews coming soon:
Antifragile - Nassim Taleb
Irrational Exuberance - Robert Shiller
The Intelligent Investor - Benjamin Graham
Check out my other websites:
www.tulipscreener.com - Have our robots watch the stock market for you.
www.visacalculators.com - Learn more about the world and the people you share it with while traveling. I share information on visa requirements and I’m building other calculators for travel that are coming soon!
www.ferryscan.com - Book a ferry in the Baltic Sea on your travels this summer!