DraftKings FILE
CNN  — 

In the wake of the two national party conventions, a slew of national polling has been conducted that suggests the 2020 race hasn’t changed all that much. Former Vice President Joe Biden maintains a high-single digit lead over President Donald Trump, with voters still very concerned about the coronavirus pandemic – and less than thrilled about how Trump has performed on that front. (I wrote about all of that here.)

Yet, one thing has changed A LOT over the past month: How betting markets see the race. According to Real Clear Politics’ aggregation of the odds being set by various markets, Biden went from a 61% chance of winning August 4 to a 50.5% chance on September 4. Which, if you look at the other available data in the race, doesn’t make a ton of sense!

To answer the “why” behind the movement in betting odds, I reached out to Rufus Peabody, a professional sports bettor and partner at the analytics firm Massey Peabody. (If this stuff interests you, Rufus also has a GREAT sports betting podcast called “Bet the Process.”)

Our conversation, conducted via email and lightly edited for flow, is below.

Cillizza: Can you explain broadly how betting markets like Betfair and Bovada work – for politics and sports?

Peabody: So there’s a distinction here. Betfair is a betting exchange, while Bovada is a sportsbook. If you place a bet at Betfair, you are matching with someone else – often a professional market maker, but possibly just another individual – who wants the opposite side of the bet. Betfair makes money by taking a commission on each transaction. This is a true market, as prices are directly determined by the betting public.

When most people refer to the “betting market.” however, they are not referring to an actual market but rather to prices at sportsbooks, which are *not* directly tied to supply and demand. The key distinction between a sportsbook and a betting exchange is the counterparty. If you place a bet at a sportsbook, the sportsbook is the counterparty. If you bet on Joe Biden to win at +100 (50% implied odds), the sportsbook is taking the opposite side of that bet (Trump at +100). Sportsbooks function as markets in a general sense because books are trying to draw equal action on both sides. While exchanges make money by taking a commission, sportsbooks make money by offering “unfair odds.” If a book thought the true price was Biden 50%, Trump 50%, they might offer Biden 53%, Trump 53%.

Cillizza: Who is setting the baselines for these markets?

Peabody: The prices at an exchange like Betfair are directly set by market participants. Anyone can make an offer, and if there’s a counterparty willing to accept, voila! For traditional sportsbooks, there are a few large offshore books that traditionally act as market makers, setting initial odds, and moving them in response to the betting action they receive.

Cillizza: What makes a market move? Outside events? Momentum?

Peabody: The short answer is money, and usually, though not necessarily, smart money. Sportsbooks want to make money and minimize risk, which they can do by setting odds that draw betting volume proportional to the odds offered. This is referred to as balanced action, or two-way action. If the betting public is wagering a lot more money on one team (or politician) than another at a traditional sportsbook, the book can adjust the odds they offer (for future bets) in hopes of receiving more balanced action. Why? By setting a “price” that equalizes supply and demand, they can use the money from losing wagers to pay off the winning wagers, and collect a risk-free profit. But it’s not exactly that simple.

Let’s say there are two classes of bettors: “Sharp” bettors who will bet large amounts of money if they find a profitable opportunity, and “square,” recreational bettors who do not run sophisticated models, are not especially price sensitive, and are long-term losers (in terms of money made and lost). Squares make up 96% or more of the betting public. If a sportsbook is getting unbalanced action, but all that money is being bet by “squares”, it’s less inclined to adjust its odds, since by doing so, it risks skewing the odds and creating opportunities for “sharps” with large bankrolls to exploit. (Side note: For this reason, books place limits on the size of any one bet, allowing them to adjust the odds before too much damage is inflicted)

All this is to say that the most sophisticated offshore sportsbooks that have the highest bet limits are very good at profiling bettors – identifying the “sharp” bettors from the “squares” – and will adjust their odds in response to large wagers from “sharps.” Other, less sophisticated books (i.e. most US books), will adjust their odds to closely mirror the odds of the sophisticated sportsbooks.

That was a very long-winded way of saying that large bets from sophisticated, successful bettors generally make markets move. But every market is different. The greater the liquidity, the higher the limits, the more likely market moves are to be efficient – moving toward the true probability. Markets with lower liquidity and/or lower limits tend to have less efficient prices. A market like PredictIt, where users can only bet $800 on a position, should have much less efficient odds than a liquid exchange that does not put a cap on position sizes.

Cillizza: There’s very little actual evidence – polling, money etc. – that suggests Trump is much more likely to beat Biden than he was a month ago. But the odds are basically 50-50. What gives?

Peabody: Simply put, there has been an influx of money bet on Trump, which has driven his price up. Does that reflect a true fundamental change in the race, or is it an inefficient market move? I am always inclined to side with liquid betting markets over polls – they have a track record of outperforming polls and pundits. Are there times when markets are inefficient? Just look at the stock market moves in the last six months! Of course there are.

While polls can be a good barometer, they rely on a number of assumptions, including the composition of the electorate. Given how unusual this election will be, with so many fewer people voting in-person, as well as the possibility of voter suppression, I would argue that voter turnout is much more of an unknown this year than in a typical election cycle.

Polls also reflect support now. I’m no political expert, but the perception, accurate or not, seems to be that Republicans have finally found a strong message: law and order. It feels like the election is being fought on different turf than it was a month ago, and while continuing protests may galvanize progressives, they may hurt Biden’s chances with moderates. The law and order message worked for Republicans in 1968, another election year marred by civil unrest.

I’m no political expert, though, so this is really all just speculation.

Cillizza: Finish this sentence: “The best way to use betting markets to understand where a game or a presidential race is headed is _______.” Now, explain.

Peabody: “The best way to use betting markets to understand where a game or presidential race is headed is as the best indicator we have available of the probability. “

Just like stock prices take all available information – including fundamentals such as earnings reports – political betting markets take data, such as polls and fundraising numbers, into account. Markets aren’t always efficient, but liquid markets do a better job of synthesizing available public and nonpublic information than polls alone. (Cillizza note: There is some disagreement on this point.) It’s possible that markets are moving on some information that’s not yet public; we have no way of knowing for sure. I should stress that betting market predictions aren’t always accurate, but historically, they have been better predictors than polls. The predictive value of any market (or pseudo-market) is directly related to its liquidity (how much money is being wagered by all market participants), and any restrictions, such as betting limits.