Review of Nate Silver’s The Signal and The Noise

Nate’s book, The Signal and The Noise, immediately goes into the pantheon of great books about risk and prediction.

Any illusions that this book will put the reader on the path to quickly copy the success of Nate or his varied protagonists are dispelled in the early pages. Nate writes:

That is why this book shies away from promoting quick-fix solutions that imply you can just go about your business in a slightly different way and outpredict the competition. Good innovations typically think very big and they think very small. New ideas are sometimes found in the granular details of a problem where few others bother to look. And they are sometimes found when you are doing the most abstract and philosophical thinking, considering why the world is the way that it is and whether there might be an alternative to the dominant paradigm. Rarely can they be found in the temperate latitudes between these two spaces, where we spend 99 percent of our lives.

In Nate’s brilliant chapter on poker, he references the “Pareto principle of prediction in competitive environments,” which hypothesizes that 80% of predictive accuracy is achieved with the first 20% of effort. This level of effort, in poker or any other field, takes one only up to “water level”, well short of the standard required to make a living. Nate’s view is that, to succeed in any predictive field, one must work very hard, perhaps 60 hours a week, but all the value relative to the competition comes in hours fifty through sixty.

Nate’s careful tone, familiar to readers of his blog (fivethirtyeight.blogs.nytimes.com), is partly temperamental, but partly, I suspect, related to his poker experience. Poker has a way of beating certainty out of people. I don’t think it’s an accident that two of the best books on risk written in the past couple of years, Nate’s The Signal and The Noise, and Aaron Brown’s Red-Blooded Risk, were written by poker players.

You would think that our best writing about risk would come from guys who run hedge funds, but this is emphatically not the case. It could be that hedge fund managers value their time and knowledge highly and don’t want to share, but it might also have to do with selection. Every intelligent poker player has gone through periods where they’ve run insanely poorly, and has also, more painfully, played in games where they started off assuming they were a winner but then ultimately had to conclude that they likely were not. Many hedge fund managers simply haven’t had the painful lessons; the fact that we observe them currently managing a lot of money means that if they are, like most humans, overconfident in their beliefs, they’ve yet to pay the price for it. Victor Neiderhoffer wrote a pretty good on speculation back in 1998, when he was on top (The Education of a Speculator); he would probably write a damn good one now, having been broke two times since, but there’s no market for busted hedge fund managers who want to write books and so few get the benefit of their experience. The consensus Wall Street pick for the best book on speculation, Reminiscences of a Stock Operator, was the rare Wall Street book narrated by someone who lost his fortune several times over.

Nate’s book covers statistical prediction as it relates to the mortgage crisis, political prognostication, baseball, weather, earthquakes, macroeconomic forecasting, infectious disease, basketball, chess (Deep Blue vs Kasparov), poker, money management, climate change, and terrorism. I suspect that the areas I know the least about (terrorism, climate change, infectious disease, earthquakes, and weather) were also the weakest chapters in the book. His chapters on poker, basketball, and money management, the areas I know the most about, struck me as nearly perfect, getting nearly every detail correct. I think the classic books in economics are defined by the little choices about what to include or not to include; a classic book, for example, Thomas Schelling’s Micromotives and Macrobehavior, will make the tradeoffs so perfectly as to inspire and teach newcomers and masters in the field equally, meeting Hilary Putnam’s test for a philosophical classic: “The smarter you get, the smarter it gets.”

In the three-plus years Nate spent writing this book, he clearly borrowed a bit of the best from every field. Those seeking a summary of the current state of the literature on market efficiency could hardly to better than Nate’s chapter, “If You Can’t Beat Them…” I can see the thumbprints of the genius Richard Thaler all over that chapter, and indeed the influence of another book from the pantheon — Richard Thaler’s The Winner’s Curse – is evident throughout Nate’s book.

I cheated and did not read this book straight through. I immediately jumped to see what Nate had written about my buddies Tom Dwan and Haralabob Voulgaris. Haralabob was clearly a bit gun shy with Nate, sparing him the details of his analytical methods or at least not allowing Nate to write about them. The chapter is nonetheless immensely entertaining. The exact backstory of how Haralabob came to risk all of his money on a 6.5-to-1 shot was not known to me, and it provides some context for our first meeting, when Haralabob and I played heads up $25-50 poker all night in…

The Policy Wonk’s Guide to the Presidential Betting Market

This October, as the presidential election nears, we witness the strange intersection of the worlds of the gambler and the policy wonk. Daily, our best political observers reference the current prices of the presidential betting market. Unfortunately, I think their lack of knowledge of gambling mechanics leads them astray.

For starters, why do we care about what gamblers think about the election? The power of simple consensus estimates is remarkably powerful in a wide variety of settings (see James Surowieki’s The Wisdom of Crowds for examples). Each individual’s guess contains a kernel of truth and a lot of idiosyncratic error; when we average together individual guesses, the idiosyncratic errors filter out. Market prices are a much more powerful form of consensus estimate in which the more informed participants tend to be given much greater weight in the market. Indeed, a stronger condition holds: one perfectly informed individual can fully determine price.

So, in theory, market prices can tell us quite a lot about the election. In practice, we have to look at the institutional peculiarities of the market to determine if it is in fact well functioning and informative. This is where you have to ask yourself some simple, common-sense questions. Is the market big enough to matter? Are people staking large amounts of money on the outcome? The answer to these questions, in the case of the presidential betting markets, is yes. The amount bet on this election is in the ballpark of $100 million.

In betting markets, it’s wise to be conspiracy-minded. We should ask: Is there an endogeneity to the market, such that movements in market prices actually influence the outcome? The answer to this is “yes, probably”. This opens up the possibility that there is gaming of the market. You then have to ask, “How costly is such gaming?” The answer is that it’s somewhat costly, but the dollars are quite small relative to total campaign spending, so we have to leave this open as a possibility. I personally think that market gaming is exceedingly unlikely, but it’s a possibility that one must consider in an examination of market mechanics.

Another crucial betting mechanics question is: can a lot of money be moved at relatively low spreads? You want a lot of money to be moved so that sophisticated speculators have an inventive to dig deeply in an attempt to understand the underlying reality; if it’s a small dollar market, no one can be bothered. A low friction, low spread market is going to react much more quickly to news. Day-to-day, hour-to-hour fluctuations will tend to have more meaning in a low spread market than a high spread market.

This brief introduction to betting mechanics brings us to the first uncomfortable tension between gamblers and policy wonks: policy wonks love to quote Intrade, and gamblers think it’s by far the least important and least informative presidential betting market. It’s easy to get a small amount of money on Intrade but hard to get a large amount of money on or off; as a result, you have a huge number of people betting small amounts, resulting in a middling level of total volume (some of the other sites I will discuss are far bigger). You don’t want the gambling to look like the voting, where everyone casts a vote and all votes count the same. You want the most sophisticated market participants to have more money on the site and much more influence in determining marginal price. Intrade lacks these high balance, sophisticated accounts. One way to see quickly that Intrade is lacking these large accounts is to note that a no-arbitrage condition does not hold between Intrade and sites such as Pinnacle and Betfair that are booking a huge amount of presidential action. The prices between Intrade and Pinnacle are often dramatically different; this remains true because it is difficult to get large amounts of money on or off Intrade.

The people that know the most about the presidential race are, presumably, American analysts, but, since operating online sites is illegal in the US, almost all the action is booked on offshore betting sites. These sites are difficult for Americans to deal with, and so the prices on US elections are much less informative than, for instance, the prices of British elections of comparable import (it’s very easy for Brits to get money on and off of betting sites).

There are three betting sites worth paying attention to: Pinnacle Sports, Betfair, and Matchbook. Betfair and Matchbook are both exchanges that match bets and take a commission. Pinnacle (www.pinnaclesports.com) is more of a traditional sports book. Among traditional sports books, Pinnacle takes the most action on the presidential race by a huge margin. All other sports books are anchoring their prices to Pinnacle.

When a sports book like Pinnacle is taking the lead on a market like the presidential race, they can choose to set their line with a heavy hand, where they have strong opinions about what the price should be and are slow to move the line as new money flows in (they will as a result sometimes accumulate big positions on one candidate or the other), or they can set their line with a light hand, where they are quick to move the line as new money…