Betting On Elections Poses a Minimal Threat To Election Integrity
By: Solomon Sia and Pratik Chougule
One of the most important reasons why regulators are reluctant to liberalize political betting regulations is due to the fear that wagering on elections would affect the integrity of elections.
To assess this risk, we analyzed the historical and theoretical adverse effects of prediction systems on election integrity and interviewed traders and platform operators of existing election-based markets to understand election manipulation risks specific to election markets.
We consider the following integrity risks:
- Prediction markets may serve as a mechanism to sway election outcomes through self-fulfilling (or self-defeating prophecies). This includes manipulating markets to sway voter sentiment and hence election outcomes and vote buying.
- Prediction markets may serve as a direct financial incentive to manipulate elections (by means other than prediction markets).
- Prediction markets may be subject to manipulation for profit, e.g. via the publication of false polls, which manipulates election outcomes as a byproduct.
- Prediction markets may affect the perception of election integrity.
- Prediction markets may facilitate violations of campaign finance laws.
- Election insiders may manipulate outcomes to create profits on the market or trade on insider information.
Ultimately, we believe the risk of election-based contracts on election integrity is negligible relative to the risks that already exist.
Further, these risks are mitigated by effective regulatory oversight of these markets, and small relative to the economic and social utility of these contracts.
Self-Fulfilling Prophecies
One category of risk is the self-fulfilling prophecy—or its inverse, the self-defeating prophecy—where knowledge of the prediction affects the result.[1] The 2016 US presidential election offers plausible evidence for the self-defeating theory, where overconfident win predictions for Hillary Clinton, including in the prediction markets, may have lowered turnout enough to tip the election.[2]
The converse risk is the self-fulfilling prophecy.
One could imagine a hypothetical scenario in 2016 where overconfident win projections for Hillary Clinton lowered turnout for Republican voters sufficiently to tip the election in her favor.
The arguments here speak more to the integrity risks of faulty prediction systems generally rather than the proposed contracts specifically.
The incentives for prediction markets traders to seek and disseminate truth are greater than is the case for the most commonly-referenced predictors in mainstream discourse such as pundits.
Because of their active mechanisms to combat overconfidence, we consider one of the primary benefits of election markets to be a decrease of unrealistic projections, thereby reducing the potential for incorrect electoral modeling to affect election integrity.
Deliberately Swaying Election Outcomes
We searched for historical attempts to use political event contracts to manipulate the outcome of elections.
We did not find any instances of manipulation in Congressional control markets but did discover examples in other markets.
An aide on a presidential campaign in the 2016 primaries informed one of us on background that he and his colleagues placed bets on their candidate on PredictIt as part of the campaign’s strategy. They did so both in order to respond to media coverage that their candidate’s price was slipping in the markets as well as to garner favorable news coverage about the supposed prospects of their campaign. The betting limits and low liquidity on PredictIt made this a relatively inexpensive decision in the short-term, but it proved impractical as their campaign failed to gain traction and traders became increasingly bearish on its prospects.
Another likely instance of attempted manipulation occurred in the UK in markets in the London mayor race. Anecdotal evidence suggests that allies of an obscure candidate, Brian Rose, may have facilitated bets in the market to pump his price, and therefore, support notions that he may be a viable candidate. The gambit failed when media and political observers treated the candidate’s briefly inflated price as noise.
The relatively inconsequential cases of manipulation we found in our research are consistent with the academic literature on this topic.
Studies on the so-called “Romney whale” in the 2012 Intrade markets, a single pro-John Kerry manipulator in the 2004 TradeSports market, as well as surveys of earlier political stock markets indicate that manipulation can be detected by traders, media, and researchers, and that systematic manipulation is difficult beyond short time periods.[3]
As the efficient market hypothesis indicates, given sufficient interest and liquidity, traders can be expected to bring the market price to a more efficient level relatively quickly.
Our thesis in this respect is informed by our conversations with UK-based colleagues, who have monitored Congressional control markets with large amounts of liquidity over many decades. They reported no clear, consequential cases of manipulation in these markets. They observed, moreover, that while allegations of manipulation in sports betting in the UK have led to the creation of a nationwide integrity unit, nothing comparable exists in politics and there appears to be no meaningful demand for one even by the most vocal advocacy groups.
We believe that the existing evidence of failed manipulation is reason for cautious optimism. It suggests that prediction markets are considerably less likely to mislead the public than the less transparent mechanisms already available today such as push polling, reporting based on background sources, election analysis platforms, and proprietary models.
Lying in a market that has an active mechanism to counter noise and fake news is a dubious strategy when considering the alternatives.
Even if cases were to arise of market manipulation, calls for outlawing election contracts on this rationale should be weighed against the benefits that isolated attempts at manipulation have from an academic/research perspective. They would further knowledge on when and under what circumstances traders seek to manipulate election markets and how consequential these efforts are.
Using Event Contracts To ‘Buy Votes’
In the Nadex Ruling, the CFTC declared that,“Political Event Contracts can potentially be used in ways that would have an adverse effect on the integrity of elections, for example by creating monetary incentives to vote for particular candidates even when such a vote may be contrary to the voter’s political views of such candidates.”[4]
A related argument is that, if the political event contracts truly worked as an efficient economic or emotional hedge, a voter could theoretically put enough money on one side or another such that they became wholly ambivalent to the outcome and abstain from voting.
This is not a well formulated integrity concern.
It is in the nature of democratic elections that voters have the prerogative to weigh myriad personal incentives—including financial ones—in their choice of candidate.
In an era in which the government has a profound impact on individuals’ financial future through tax, spending, and regulatory policy, the relatively small amounts of money at stake in an election market can be expected to be a secondary concern at most.
The CFTC’s Nadex statement suggests that voters might voluntarily shape their own preferences and “steal votes from themselves”, which does not constitute an election integrity risk.
These concerns, moreover, are speculative, abstract, and almost entirely absent from our experience with political prediction markets.
In large part due to the difficulty of generating high profits in election markets relative to other types of betting markets with more frequent and consistent events, traders tend to participate in election markets because of their preexisting interest in politics.
While traders routinely acknowledge that they are trading against candidates who they personally support, we are unaware of traders who consciously base their personal political activities on their investments in the market.
Indeed, discussions in the political prediction community are replete with traders who disengage from election markets when they have a strong opinion about one of the involved parties and do not trust themselves to place an objective bet.
A more coherent example of the CFTC’s concerns is as follows: a manipulator who wants people to vote Democrat could put a lot of money on Republicans winning, with the expectation of losing that money. Republican voters would see the easy money, and start betting on a Democratic victory, and thereby become incentivized to vote Democrat. Ultimately the Democrats win, and the manipulator has lost a lot of money on the prediction market but has effectively ‘bought votes’ and hurt the election integrity as a result. This mechanism may appear dangerous, not least because it is indistinguishable from hedging behavior by an actor who hopes for a Democrat win but is hedging against a Republican win. However, this method of ‘vote purchasing’ is extremely impractical because there is no way to make the right amount of money go to the right people. A single individual, or even a dispassionate corporate entity that has no voting power could take all the money without providing any return on investment.
There exist far more direct and reliable ways to sway election outcomes today.
Direct Financial Incentive to Manipulate Elections
Another integrity risk is that election-based contracts, by introducing a profit motive, may incentivize individuals with a stake in those markets to alter election outcomes in order to make money on the markets.
If an entity has a large position on an outcome, it has a financial incentive to make that outcome come to pass.
This concern does not make sense given the size of typical event contract positions relative to the incentives already at stake.
Individuals and organizations already have strong reasons to sway an election and the policy outcomes at stake far outweigh any market gains available in the vast majority of political betting contracts. We do not see direct financial incentives as an issue even at several multiples of the current proposed limits being considered at the CFTC.
The CFTC may have recognized in 2012 that the election integrity fears based on additional incentives created by political event contracts were frivolous as it did not elect to mention them in the Nadex ruling.
Nevertheless, we sought historical examples of individuals attempting to manipulate elections to make money on prediction markets. The closest one we found were rumors that death threats were being leveled against Andrew Yang during his presidential campaign from an anonymous trader who was attempting to manipulate PredictIt’s briefly operating market on how many times Yang would tweet.
PredictIt’s decision to offer the market in the first place went against advice from veteran political prediction market traders who reasoned, correctly, that this type of niche market was on dubious regulatory grounds and was more likely to incentivize foul play than the election contracts proposed by Kalshi.[5]
Perception of the Integrity of Elections
We considered how the proposed contracts might affect perceptions of election integrity.
As a meta point, considerations pertaining to the perception of election integrity hold much less weight than considerations of actual election integrity risk. Given logical analysis and reasoning, perception will approach reality—that is, that the proposed contracts have an insignificant impact on election integrity.
Nevertheless, we do not take it on faith that the public will automatically take the same reasoned analysis and come to the same conclusions we have described above. We discuss some reasons why prediction markets might be perceived as threats to election integrity.
First, the financial incentives caused by prediction markets are more direct than conventional political incentives, as there is a direct payout in response to one side or another winning. This direct mechanism could be perceived as higher risk relative to its actual risk.
Second, although public perception of gambling has improved in general and prediction markets are meaningfully distinct from gambling, a minority may take offense at a financial incentive mechanism they consider to be gambling. In this case there is a focus on the mechanism of election integrity violation rather than the likelihood of the actual violation.
Third, there is a natural inherent distrust of any new potential mechanisms of abuse, regardless of its risks relative to the mechanisms already available.
Fourth, by adding “skin in the game” for market participants, the proposed contracts increase the emotional and financial investment in the outcome. When the outcome does not go according to their wishes or expectations, it increases the emotional response, which leads to stronger, albeit unfounded perceptions that the election integrity has been compromised. For example, in the 2020 elections, millions of Americans went to the polls believing that their preferred candidates would win by a comfortable margin. When the results defied their expectations, many “blue wave” traders lost money on PredictIt while suspicions about election fraud gained traction.
Conspiracy-oriented traders flooded political prediction markets with bets on Republican candidates, only to suffer losses as more sophisticated traders took the other side of their bets.
Conversely, there are strong reasons to believe prediction markets will be a net positive to both election integrity and perceived election integrity, as follows.
First, because prediction markets are inherently non-partisan, aggregate perspectives democratically and have strong incentives towards accuracy, they are less likely to be demonized by one side or another. This is coupled with the insight and social consensus building incentives of prediction markets.
Doubts about the integrity of U.S. elections have risen in the past few years for reasons that have little if anything to do with political prediction markets. Because of the transparency of prediction markets and its active mechanisms to combat falsehood, we consider one of the primary benefits of a political prediction market to be the reduction of incentives and effectiveness of current methods to interfere with election integrity.
Second, mainstream understanding and acceptance of gaming and the benefits of forecasting have improved since the 2012 Nadex contracts.
Prediction market platforms such as Kalshi and influencers within the forecasting and rationalist community are strongly incentivized to educate the public.
Third, the opportunity to trade on election outcomes in the context of PredictIt has created powerful incentives for the public to become informed about the political process and more cognizant of one’s own ignorance and biases. This is easily observed in discussions in the political prediction market community, which are often far more sophisticated than those in the mainstream or even professional discourse. This creates financial incentives for market participants to be rational, which in turn moves their perceptions of election integrity closer to reality.
We expect election betting to continue to produce a new generation of citizens whose interest in political prediction markets leads them to engage constructively in the political process and to have reasoned opinions about election integrity.
Fourth, prediction markets themselves give signals on election integrity.
In 2020, at a time when the president of the United States and a major political party were seriously entertaining the possibility that the election was “stolen”, that Trump would serve a second term, and that key Senate race calls would be reversed, market prices indicated that traders understood better than many members of Congress that the election was conducted without a meaningful amount of fraud and that the United States would see a transfer of power to Joe Biden.
Reflecting widespread concerns about election integrity among the electorate, candidates since Trump have decried election fraud after losing their congressional races, but election markets on PredictIt and elsewhere have hardly moved on this news. At the same time, Congressional markets are among the most valuable sources available today to assess whether and how federal and state inquiries into election integrity will proceed.
Although political prediction markets play a limited role currently in shaping perceptions of election integrity, recent history shows that they are more likely to increase rather than decrease confidence in U.S. elections when the public at large sees that the “smart money” is betting on the assumption of fair elections.
Facilitate or Violate Campaign Finance Laws
Over the course of extensive interviews with historians, practitioners, and industry leaders in both the United States and the United Kingdom, we did not come across any evidence that political prediction markets have been or are being used to facilitate violations of, or otherwise undermine, federal campaign finance laws or regulations to any meaningful degree.
Relative to the existing mechanisms and loopholes by which parties may sidestep prohibitions governing coordination between candidate campaign committees and political action committees, the contracts do not offer a feasible mechanism to facilitate violations of, or otherwise undermine, federal campaign finance laws or regulations.
Insofar as election markets carry the risk of undermining campaign finance laws, however, law enforcement officials are more likely to determine if this is occurring on a regulated exchange with a responsible stakeholder rather than a decentralized or offshore site with less incentive to police its site in line with American legal and political norms.
Market Manipulation for Profit
As part of our research, we sought examples of manipulation by insiders on existing prediction platforms.
A form of manipulation is the creation of fake polls by traders to move betting markets. Our British colleagues were not aware of fake polling being used to manipulate UK-based markets, but the phenomenon appears to be more common on PredictIt. FiveThirtyEight’s report “Fake Polls Are A Real Problem” notes, as an example, that the price for one share — which is equivalent to a bet that Senator Debbie Stabenow will be re-elected — fell from 78 cents to as low as 63 cents due to a fake poll before finishing the day at 70 cents. Market motivations may have been secondary to the trolling factor, but the mere fact that the markets can be so easily manipulated—as seen by numerous other examples—is arguably noteworthy.[6]
Ultimately, the phenomenon of manipulation via fake polls is of some concern to certain types of political prediction markets with limited information, few public polls, and low liquidity. Even in such markets, the incentives for market correction and exposure tend to override any attempts to manipulate the market. The proposed markets would be even more difficult to manipulate through fake polls due to the abundance of information available to market participants, frequent polling by reputable firms, and the high liquidity they draw. Manipulation is also possible through sound polling. We interviewed one PredictIt trader who commissioned a real poll to move the markets. The trader told us that the poll was real with a sound methodology, and was commissioned to correct what he believed to be an inefficient market. Ultimately the trader financially benefited from the process of discovering truth via his poll and taking a position before releasing the polling results. We take this example as evidence that prediction markets may also reward truth seeking and truth dissemination by financially motivating the commissioning of accurate polls.
Rules Against Insider Trading
Prediction markets may incentivize insiders to put money on an unlikely outcome and make the outcome occur. For example, a frontrunner candidate may bet against themselves and then intentionally lose the election to reap a profit. We have not found any historical examples of candidates throwing an election in order to make a profit from prediction markets.
Prediction markets may also enable insider trading of non-public information. We learned of several instances of campaign aides in the 2016 primaries trading on PredictIt while working for presidential candidates. Often, aides were simply trying to profit personally, calculating (often incorrectly as it turned out) that their experience on the campaigns would give them an edge. Insofar as we are interested in political prediction markets that express efficient pricing, we would oppose prohibitions on any entity’s participation in these markets given that they may have valuable information. A promise of election markets is that they will elicit knowledge from many market participants that wouldn’t have otherwise been shared and that this knowledge will be used to make better decisions.
If, however, the Commission determines that such a prohibition would alleviate concerns among regulators and/or the public regarding campaign finance law, manipulation, and surveillance, it may be worth enacting such a policy. This prohibition, in combination with Know Your Customer laws, may not completely prevent insider trading, but it would give regulators advantages in monitoring and taking action against the practice that they would not necessarily enjoy on unregulated exchanges.
[1]Herbert Simon, Bandwagon and Underdog Effects and the Possibility of Election Predictions, Public Opinion Quarterly, Volume 18, Issue 3, Fall 1954, Pages 245–253, https://doi.org/10.1086/266513
[2] Nuño Sempere, “Real-Life Examples of Prediction Systems Interfering with the Real World (Predict-O-Matic Problems), LessWrong, 3 December 2020, https://www.lesswrong.com/posts/6bSjRezJDxR2omHKE/real-life-examples-of-prediction-systems-interfering-with; Zeynep Tufekci, “Can We Finally Agree to Ignore Election Forecasts?” The New York Times, 1 November 2020, https://www.nytimes.com/2020/11/01/opinion/election-forecasts-modeling-flaws.html
[3] David Rothschild and Rajiv Sethi, Trading Strategies and Market Microstructure: Evidence from a Prediction Market (November 22, 2015). The Journal of Prediction Markets 10 (1), 1-29, 2016, Available at SSRN: https://ssrn.com/abstract=2322420 or http://dx.doi.org/10.2139/ssrn.2322420; Sethi, “The Romney Whale” 26 September 2013, http://rajivsethi.blogspot.com/2013/09/the-romney-whale.html; Rhode, P.W., Strumpf, K.S.: Manipulating political stock markets: A field experiment and a century of observational data. Working Paper (2008), https://users.wfu.edu/strumpks/papers/ManipIHT_June2008(KS).pdf
[4] CFTC, In the Matter of the Self-Certification by North American Derivatives Exchange, Inc., of Political Event Derivatives Contracts and Related Rule Amendments under Part 40 of the Regulations of the Commodity Futures Trading Commission (Apr. 2, 2012) (“Nadex Order”) at 4. https://www.cftc.gov/sites/default/files/idc/groups/public/@rulesandproducts/documents/ifdocs/nadexorder040212.pdf
[5] See for example Twitter post, 4 August 2022, 6:29 p.m., https://twitter.com/Domahhhh/status/1555320074524770304?s=20.
[6]Harry Eten, “Fake Polls Are A Real Problem,” FiveThirtyEight, 22 August 2017; https://fivethirtyeight.com/features/fake-polls-are-a-real-problem. For additional examples see Yeargain, T. (2020): “Fake Polls, Real Consequences: The Rise of Fake Polls and the Case for Criminal Liability,” Missouri Law Review, 85,140-150
This report is adapted from CFTC, Comment for Industry Filing 22-002, Solomon Sia. Comment No: 70745, 23 September 2022. https://comments.cftc.gov/PublicComments/ViewComment.aspx?id=70745&SearchText=sia