Full Research, The Accuracy of Betting Markets: Issue 2

🌟 Editor's Note

Our full research posts will be deep dives into a specific subject with lots of research and we intent to make all of the research and papers into an easy-to-read article meant to greatly expand your knowledge on any given subject with our reading TOO much like a research paper. (and sorry for the lack of images)

Betting markets have a mystique as crystal balls 🔮 for the future. From presidential elections to Super Bowl winners to economic turns, many people watch the odds closely, hoping the “wisdom of the crowd” will foretell outcomes. But how accurate are these markets really as forecasting tools? In this article, we’ll explore what betting markets do well, where they fall short, and why. By the end, you’ll have a nuanced understanding of how these markets can be both helpful and misleading in gauging public sentiment and real-time probabilities.

🪙 Betting Markets as Forecasting Tools: How They Work

- At their core, betting markets (or prediction markets) let people wager on specific outcomes. If a share tied to an outcome pays $1 if the event happens (and $0 if not), then the trading price of that share reflects the crowd’s perceived probability. For example, a candidate’s share trading at 0.33 (33¢) implies roughly a 33% chance of that candidate winning. The idea is that people with information or strong beliefs will “put their money where their mouth is,” driving the price up or down. In theory, with enough participants and trading, the market price becomes an aggregated forecast incorporating all available information​.

- Economists love this concept. With sufficient volume and incentives, a betting market can distill the collective knowledge of diverse individuals into a single probability. This is sometimes described as the “wisdom of crowds” in action. Instead of relying on one expert or one poll, a prediction market lets anyone trade on what they think will happen. If new information emerges – an injury to a star player, a scandal for a politician, a surprising economic report – traders quickly buy or sell, and the odds adjust in real time. In essence, the market price continuously updates to reflect the current consensus on likelihood.

- Importantly, these markets can be designed for almost any kind of prediction: binary yes/no events (win vs. lose), elections with vote share contracts, even questions about when something will happen or how much (using special contract structures). Companies have even run internal prediction markets (sometimes with play money) on things like project completion dates or sales figures, often finding they beat traditional internal forecasts​. The allure is clear: if structured well, a betting market gives an instant, crowdsourced forecast backed by real stakes.

✅ When Betting Markets Get it Right

So, do they work? Evidence suggests that often they do remarkably well. Academic studies and real-world trials have found that prediction markets can be as accurate as, or more accurate than, other forecasting methods in many domains. Early experiments showed that markets performed about on par with expert opinions, and better than opinion polls or simple crowd averages​. This has held true in areas from political elections to sports betting to even predicting movie box-office results​. In other words, letting people trade on outcomes often beat just asking people what they think will happen.

One famous example comes from U.S. presidential elections. The Iowa Electronic Markets (IEM), an academic real-money prediction exchange, has been forecasting elections since 1988. **In presidential races from 1988 through 2004, the IEM’s final market prices were closer to the actual vote results than the polls about three out of four times. In fact, one analysis comparing hundreds of polls to the IEM found the market’s projection was more accurate about 74% of the time​. Not only did these markets tend to nail the final outcome, they often provided a better forecast even months before Election Day, long before many pollsters converge on a result. The average error in the IEM’s presidential vote-share prediction was only around 1.1 percentage points – impressively small​.

Sports betting markets have also shown forecasting prowess. The odds set by bookmakers (or the trading prices on betting exchanges) contain a wealth of knowledge about team strength, injuries, weather, and more. It’s no coincidence that you rarely hear of someone consistently beating the Vegas betting line – the odds are usually very well calibrated. For instance, if a football team is given a 70% chance to win by the market, you’d expect them to win about 7 out of 10 times in reality. In practice, that tends to hold true. Studies find that sports betting odds, while not perfect, are good probabilistic forecasts of game outcomes​. (There are small biases we’ll discuss later, but by and large a team favored by 3 points really does usually win by about a field goal.) The market odds integrate information from sharp bettors who study teams intensely. The moment new information (like a star player’s injury) comes out, odds shift to reflect the changed outlook. In this way, sports books and exchanges are perpetually updating predictions in real time.

Financial and economic events are another arena where markets shine. In a sense, stock prices and commodity prices are themselves prediction markets – they represent collective expectations of future earnings, product successes, or economic demand. For example, traders in futures markets bet on central bank interest rate moves, and those probabilities (such as the chance of a Federal Reserve rate hike next month) are closely watched indicators in finance. In fact, well-known economic research has shown that market-generated forecasts are typically fairly accurate, often outperforming moderately sophisticated benchmarks for prediction​. When a broad swath of investors and analysts trade on financial outcomes, the resulting prices tend to incorporate all known information. It’s not magic: it’s many minds (and wallets) processing news and data, which often yields a better forecast than any single mind.

To summarize, betting markets can be very effective forecasting tools. They’ve correctly called numerous elections, sports championships, and other events – frequently matching or beating expert predictions and opinion polls. They provide a real-time snapshot of what informed people collectively think, updating fluidly as new data rolls in. This is incredibly useful for taking the pulse of public sentiment and perceived odds at any given moment.

🚀 The High-Profile Misses: When Markets Get it Wrong

Yet, for all those successes, betting markets are not crystal balls. They sometimes misfire – spectacularly so in a few high-profile cases. Perhaps the most famous recent example was the 2016 U.S. presidential election. Going into Election Day, virtually all major betting markets had Hillary Clinton as the clear favorite to win. Some markets gave Donald Trump only around a 20–30% chance. The market consensus, like most experts and polls, was pointing to a Clinton victory. We all know what happened: Trump pulled off a surprising upset. As PBS NewsHour put it, “The long shot won. And yet pundits, pollsters and punters in the prediction markets had all been so sure.”​. Indeed, prediction market traders – the “punters” – were caught off guard just like the pollsters. These markets, which “had been a good predictor” in prior elections, got it wrong in 2016​.

The Brexit referendum earlier that same year (2016) was another black eye for betting markets. In the United Kingdom’s vote on whether to leave the EU, most betting firms and exchanges consistently had “Remain” as the odds-on favorite. Many bettors (and observers) assumed the UK would ultimately stay in the union. But when votes were counted, “Leave” won with about 52% – an outcome that shocked those who had been following the betting odds. As one British bookmaker admitted afterward, bookies had been overconfident about Remain’s chances​. Ladbrokes (a major UK bookmaker) had made Remain a heavy favorite for months, even when polls were neck-and-neck​. Why? In a post-mortem, Ladbrokes’ head of political betting noted that the betting volume was overwhelmingly on Remain. Because the bookmaker’s goal is to balance the books and manage risk (not necessarily to “predict” the result), they had to shorten the odds for Remain as money poured in​. In his words, “bookies do not offer markets on political events to help people forecast the results. We do it to turn a profit... If most of the cash went on Remain... [we] make Remain the favourites”. In short, the lopsided betting demand made the odds misleading. The market seemed to imply a high probability of staying, but that was partly an artifact of how bookmakers set prices amid skewed betting interest.

These examples highlight that betting markets can and do get things wrong. Sometimes the crowd overlooks late-breaking factors or collectively underestimates an underdog. It’s worth noting that in both the Trump upset and Brexit, the true outcome was not a 100-to-1 longshot – the odds of the surprise were maybe 20–30%. So while the favorite lost, one could argue the market did indicate some chance of the upset (just not enough to overcome people’s expectations). Still, many took those odds as gospel and were caught off guard.

Beyond politics, we see misses in sports and finance too. A heavily favored team can be upset despite the odds; an “sure thing” fight can end in a shocking knockout against the betting lines. And financial markets, for all their foresight, sometimes signal recessions or events that never materialize. There’s an old joke in economics that the stock market has predicted nine of the last five recessions – meaning it often flashes warning signs that turn out to be false alarms​. Market prices can swing on fears or hype that later prove unfounded. In 2020 and 2021, for example, prediction markets on topics like COVID-19 trends or election outcomes showed huge volatility and occasional error as people grappled with unprecedented situations.

💹 Why Markets Can Be Both Helpful and Misleading

What causes these markets to be so prescient in some cases and off-base in others? The answer lies in the behavioral and structural forces that shape market outcomes. Several key factors can boost or undermine the accuracy of betting markets:

• Liquidity and Participation: The more people involved and the more money at stake, the more robust a market’s prediction usually is. High liquidity means no single trader can easily sway the odds – instead, the price reflects a broad consensus. In a deep market (like a well-traded sports league or a large election exchange), if one person places a wild bet that Team X will win the championship, others will quickly bet against that if they disagree, keeping prices rational. But in a shallow market, a few trades can move the odds significantly​. For instance, an investigation showed that on one online political market, just 4 large accounts wagering $30 million managed to shift the odds notably​. Low participation makes markets more fragile and easier to manipulate. Niche prediction markets or new platforms often suffer from this – with few bettors, the “price” might reflect only a couple of opinions rather than a true crowd wisdom.

• Information Asymmetry: One of the strengths of betting markets is that they incentivize people with private information to trade, thereby revealing that info in the price. If you somehow know a candidate is about to withdraw or a CEO is about to resign (and it’s legal to trade on that knowledge), you stand to profit by acting before others. In sports, when a rumor of an injury or an unexpected lineup change leaks to a few insiders, those insiders might quickly bet, causing the odds to move before the news is widely known. This way, markets can incorporate whispers and insider insights faster than public news channels. However, the flip side is if nobody in the market has the crucial piece of info, the market can’t magically predict it. Or if information is misleading or rumors swirl, the market might react to essentially false information. Most of the time, widely followed markets respond and adjust to real news rapidly (often within minutes), making them useful real-time indicators. But they are only as good as the information bettors have.

• Biases and Emotions: Betting markets are driven by humans (or algorithms programmed by humans), so they’re not immune to biases. One well-documented phenomenon in sports wagering is the “favorite–longshot bias.” In simple terms, bettors tend to overvalue longshots – outcomes with low probability – and undervalue big favorites. This has been observed in horse racing and other sports: the odds for longshot horses aren’t long enough (they should be even longer given how rarely those horses win)​. As a result, bets on longshots lose money more often than bets on favorites​. Why would this happen? Possibly because some bettors just love the thrill of a longshot and are willing to bet despite the poor odds (a bit of risk-loving behavior), or because casual bettors don’t distinguish small differences in low probabilities well. On the other end, heavy favorites can sometimes offer slightly better value than the odds imply, maybe because people find it boring to bet on a near sure-thing with low payout. Over time, these tendencies create small inefficiencies – essentially, a bias in the crowd’s predictions.

  In politics and financial betting, biases can creep in too. Wishful thinking is a big one: people may bet on the outcome they want to see. If a particular candidate or team has an especially enthusiastic fan base with money to spend, the odds might tilt in that direction beyond what objective analysis would warrant. (Bookmakers often report that local fans will overwhelmingly bet their home team to win championships, no matter the actual odds – skewing those markets.) In U.S. political markets, there’s evidence that certain demographics participate more – for example, one critique noted that men (who historically leaned toward certain candidates) are far more likely to wager on elections than women​. If the pool of bettors isn’t representative of the overall electorate, the market could reflect those biases. During some recent elections, it’s been observed that online crypto-based betting markets had a user base tilted toward particular political leanings, possibly overstating those candidates’ chances​. The bottom line: markets aggregate opinions, but if those opinions are systematically biased, the market will be, too.

• Profit Motives and Market Structure: It’s crucial to understand what a betting market’s odds really represent. On a betting exchange (peer-to-peer market), the price purely reflects what traders are willing to buy or sell at – essentially a direct crowd consensus. But with traditional bookmakers, the “odds” are set by the bookmaking company, and they have a built-in profit margin. Bookmakers adjust odds based not just on probability, but on how money is flowing in. Their goal is to balance their book and ensure a profit regardless of outcome, if possible. So odds can be skewed by betting volume. We saw this with the Brexit case: bookies made Remain the favorite because most bettors were putting money on Remain, and the bookie needed to shorten those odds to limit liability. That doesn’t mean the bookie truly believed Remain was that likely – it means the bettors collectively put more money there. In a sense, the odds got biased by the weight of money. This can happen in any market: if one outcome is heavily favored by public betting, the odds will move that way. If that weight of money is smart (informed bettors), it’s fine. But if it’s just passionate fans or even a few deep-pocketed players, the odds can mislead. In thin markets, a wealthy individual could even deliberately bet large amounts to push the price and create a false impression of confidence (though they’d risk losing money if they’re wrong). Studies of political markets have found little evidence of sustained manipulation – attempts to push prices usually get corrected quickly by other traders​. Still, short-term distortions are possible, especially if few people are trading.

• Herding and Momentum: Traders are humans who can be influenced by others. If odds start moving strongly in one direction, others might jump on the bandwagon, assuming someone knows something they don’t. This can lead to overshooting. A classic worry in election markets is a “bandwagon effect”: if voters see a candidate heavily favored in betting odds, could it influence their vote (either to join the winner or to turn out as underdogs)? The evidence on voters being swayed by odds is inconclusive​, but the perception of momentum can certainly drive more betting. Thus, a market can sometimes get caught in a feedback loop of optimism or pessimism that isn’t fully justified by fundamentals – much like stock market bubbles or panics. The difference is that in a prediction market, eventually reality will hit (the event outcome), and anyone who bet irrationally will lose money. That tends to keep things more in line than not, but doesn’t eliminate short-lived runs of overconfidence.

Given these forces, it’s no surprise that betting markets are a mixed bag. When the right conditions are met – plenty of diverse, informed traders and balanced action – they synthesize information amazingly well, giving a kind of probabilistic wisdom of crowds. When those conditions aren’t met, they can stray, reflecting more noise, bias, or even strategic betting than true probability. They’re neither omniscient oracles nor useless noise; reality lies in between.

Gauging Public Sentiment and Uncertainty in Real Time

One clear advantage of betting markets is how they capture public sentiment in real time. Unlike a poll that might be taken days ago or an expert prediction that updates infrequently, market odds change by the minute. This makes them powerful for monitoring the ebb and flow of expectations. News outlets and observers often watch prediction markets during election nights or major events as a kind of instant barometer: if a candidate’s odds suddenly spike, it signals a surge in confidence (perhaps early results coming in strong); if they plummet, something has likely gone wrong for the favorite. In finance, you’ll see traders say things like “the market is pricing in a 75% chance of a rate hike” – essentially summarizing a lot of complex information into a single percentage. That helps communicate uncertainty: if that rate hike market suddenly swings to 40%, you know sentiment shifted dramatically, say, due to a weak jobs report.

Betting markets also excel at synthesizing many factors. Consider an upcoming championship game. The odds will implicitly factor in team statistics, injuries, weather, coaching, even intangibles – because bettors weigh all those in making their bets. Similarly, an election market wraps together polling data, economic indicators, debate performances, scandals, and more into one moving number. For a casual follower, this can be a convenient shorthand: a candidate at 90% likely has a comfortable lead; if you see a market at 50/50, you know it’s truly up in the air. In a way, these odds communicate what everyone collectively thinks at that moment, which can be more informative than one person’s opinion.

That said, using markets to gauge sentiment should come with a big asterisk: understand the context behind the numbers. A 90% favorite is not a guaranteed winner – it just means 1 in 10 times, we’d still expect an upset. If that favorite loses, it doesn’t mean the market “lied” – it means a 10% chance event happened. Likewise, a sudden swing in odds doesn’t cause the reality (just as a thermometer doesn’t cause the temperature). It’s easy to get caught up in the excitement of watching odds jump around (many of us saw the live betting odds oscillate on election nights or during games and felt the drama). They are valuable precisely because they distill sentiment and information swiftly. Just remember they are fallible.

For policymakers, journalists, and the public, prediction markets can serve as one tool among many to understand what might happen. They give a sense of the range of possibilities and how confident people are. This can actually enhance our understanding of uncertainty. Traditional news might say “Candidate X is leading,” but a market might say “Candidate X has a 60% chance” – which highlights that there’s a 40% chance of the opposite. In complex situations, that framing can be more honest about the unpredictability inherent in real life.

Conclusion: A Nuanced View of Market Predictions

Betting markets occupy a fascinating middle ground between insightful foresight and human folly. On one hand, they harness collective wisdom, incentives, and information aggregation in a way that often yields impressively accurate predictions. They’ve proven their mettle in many cases, offering smarter-than-polls election forecasts, well-calibrated sports odds, and meaningful readouts of public expectations. On the other hand, they are no silver bullet – they reflect the people participating, with all their biases and blind spots, and can be skewed by structural quirks like low liquidity or uneven betting interest.

For a general reader, the takeaway is this: betting markets are useful but imperfect thermometers of what people think will happen. They’re great at taking the temperature of current sentiment and odds, and they usually incorporate new information faster than any other forecast method. Just don’t treat them as determinative prophecies. Like any forecast, they can be wrong. A wise approach is to use them alongside other indicators (polls, expert analysis, data models) to get a fuller picture. If all signals agree, you might feel more confident. If they diverge, dig into why – the differences can be instructive (is it something the market knows that others don’t, or vice versa?).

In the end, prediction markets remind us that uncertainty is always part of the equation. No market odds can eliminate the fundamental unpredictability of the future. What they can do is help quantify that uncertainty in real time, and reveal the collective judgment (and misjudgment) of the crowd. That is valuable – as long as we interpret it with eyes open. In a world enthralled by predictions, betting markets offer a compelling, ever-updating insight into our best guesses about what tomorrow holds, warts and all.

Quick Fact: The Iowa Electronic Markets, a betting market run by the University of Iowa, correctly predicted U.S. presidential elections more accurately than traditional polls 74% of the time between 1988 and 2004.

Till next time,

Open Terms Team