Paper: arXiv 2407.14844
Authors: Hongzhou Chen, Xiaolin Duan, Abdulmotaleb El Saddik, Wei Cai
Abstract
Harnessing the transparent blockchain user behavior data, we construct the Political Betting Leaning Score (PBLS) to measure political leanings based on betting within Web3 prediction markets. Focusing on Polymarket and starting from the 2024 U.S. Presidential Election, we synthesize behaviors over 15,000 addresses across 4,500 events and 8,500 markets, capturing the intensity and direction of their political leanings by the PBLS. We validate the PBLS through internal consistency checks and external comparisons. We uncover relationships between our PBLS and betting behaviors through over 800 features capturing various behavioral aspects. A case study of the 2022 U.S. Senate election further demonstrates the ability of our measurement while decoding the dynamic interaction between political and profitable motives. Our findings contribute to understanding decision-making in decentralized markets, enhancing the analysis of behaviors within Web3 prediction environments. The insights of this study reveal the potential of blockchain in enabling innovative, multidisciplinary studies and could inform the development of more effective online prediction markets, improve the accuracy of forecast, and help the design and optimization of platform mechanisms. The data and code for the paper are accessible at the following link: https://github.com/anonymous.
Complexity vs Empirical Score
- Math Complexity: 4.5/10
- Empirical Rigor: 8.5/10
- Quadrant: Street Traders — empirical and applied, lower math complexity
Why this score: The paper relies heavily on feature engineering and empirical validation (e.g., 800+ features, cross-comparisons with polls, public code) but uses relatively standard statistical and ML methods without advanced mathematical theory or derivations.
Research Flowchart
flowchart TD A["Research Goal<br>Measure political leanings in Web3 prediction markets"] --> B["Data Collection<br>Blockchain data: 15k addresses, 4.5k events, 8.5k markets"] B --> C["Methodology Development<br>Construct Political Betting Leaning Score PBLS"] C --> D["Computational Analysis<br>Regression + 800+ behavioral features"] D --> E["Validation<br>Internal consistency & external comparison"] E --> F["Key Findings<br>Decoded political vs profit motives<br>Enhanced market mechanism design"]
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes the Political Betting Leaning Score (PBLS) as a novel quantitative measure of individual political leanings derived from Web3 prediction market betting behavior.
Source sentence
we construct the Political Betting Leaning Score (PBLS) to measure political leanings based on betting within Web3 prediction markets.
Assumes 5
- Blockchain anonymity prevents collection of demographic data about users. data
Source sentence
we cannot obtain demographic data about Polymarket users due to blockchain anonymity.
- Polymarket users may not represent the general population, introducing sample bias. other
Source sentence
Polymarket users may not represent the general population but rather a subsample with biases.
- The Discrete Log Score (DLS) is an appropriate metric for measuring prediction accuracy of political markets. model
Source sentence
As the variant of Brier score [47], the DLS is widely applied to score continuous binary outcomes that resolve out-of-bounds.
- Linear regression can capture the relationship between EWMA DLS and market characteristics. statistical
Source sentence
Linear regression analyses were performed to estimate the coefficients and the 𝑅2 scores, which indicate the proportion of variance in the EWMA explained by each characteristic.
Show 1 more
- Features engineered from on‑chain behavior contain information predictive of political leanings. model
Source sentence
Among the 825 features, 533 exhibited significance with PBLS (𝑝< 0.05), suggesting that these features contain valuable information for predicting users’ political leanings.
Data
- Polymarket event and market data crypto-onchain public
Source sentence
We collected Polymarket data from 4,592 “events” (collections of related “markets”) and 8,742 “markets”, spanning the first market on October 2, 2020, until February 27, 2024.
- Betting records crypto-onchain public
Source sentence
Our dataset includes over 1.8 million betting records for 15,555 unique blockchain addresses participating in political betting on Polymarket.
The Graph (subgraphs)
Costs
Not applicable
Finds 5
- Political markets on Polymarket achieve an average Discrete Log Score of 0.74, surpassing some comparable platforms. predictive-accuracy
Source sentence
the average DLS for political markets on Polymarket is 0.74, a little surpassing some famous counterparts, like Metaculus (0.645)
- Betting volume and number of unique participants together explain 50% of the variance in predictive accuracy (R² = 0.5). statistical
Source sentence
The regression yielded an 𝑅2 score of 0.5, indicating that these two variables collectively explain 50% of the variance in predictive accuracy.
- Higher betting volume is associated with lower accuracy (coefficient -5.18) while more participants improve accuracy (coefficient 5.55). statistical
Source sentence
The coefficients for betting volume and participants exhibit opposite signs (-5.18 and 5.55, respectively), suggesting that increased trading volume may introduce noise or speculation, while a larger participant improved accuracy
- 533 out of 825 engineered features are significantly correlated with PBLS (p < 0.05). statistical
Source sentence
Among the 825 features, 533 exhibited significance with PBLS (𝑝< 0.05), suggesting that these features contain valuable information for predicting users’ political leanings.
- In the 2022 U.S. Senate election case study, higher PBLS (more Democratic‑leaning) is positively associated with higher realized profit. performance
Source sentence
The regression in the scatter plot revealed a significant positive relationship between PBLS and P&L, suggesting that Democratic-leaning bettors might have better judgment.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 21 related papers). Coverage is partial, so this is neither replication nor refutation.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.
Cited References (73)
- The economic approach to human behavior (1976)
- A note on the pure theory of consumer’s behaviour (1938)
- Social choice and individual values (2012)
- Interpreting the predictions of prediction markets (2006)
- The wisdom of crowds (2005)
- Life cycle of smart contracts in blockchain ecosystems (2017)
- Political activism, information costs, and stock market participation (2013)
- Political polarization on twitter (2011)
and 65 more
- Markets vs. polls as election predictors: An historical assessment (2012)
- Interpreting prediction market prices as probabilities (2006)
- Combinatorial information market design (2003)
- Why people vote: Free riding and the production and consumption of social pressure (1991)
- The serial position effect of free recall. (1962)
- Are political markets really superior to polls as election predictors? (2008)
- Habit formation in voting: Evidence from rainy elections (2016)
- Trading strategies and market microstructure: Evidence from a prediction market (2016)
- Political misinformation (2020)
- Why are self-report and behavioral measures weakly correlated? (2020)
- “It could turn ugly”: Selective disclosure of attitudes in political discussion networks (2018)
- Does Political Participation Contribute to Polarization in the United States? (2022)
- Analyzing polarization in social media: Method and application to tweets on 21 mass shootings (2019) — link
- The herfindahl-hirschman index (1993)
- Job engagement: Antecedents and effects on job performance (2010)
- Work engagement, performance, and active learning: The role of conscientiousness (2012)
- Xgboost: A scalable tree boosting system (2016)
- Tree boosting with xgboost-why does xgboost win" every" machine learning competition? (2016)
- XGBoost model for chronic kidney disease diagnosis (2019)
- Feature selection: A data perspective (2017)
- The 2022 Elections: A Test of Democracy’s Resilience and the Referendum Theory of Midterms (2023)
- Quality, not quantity: Strategic politicians in US Senate elections, 1952-1990 (1994)
- Fundamental models for forecasting elections at the state level (2014)
- The Blue Wave: The 2018 Midterms and What They Mean for the 2020 Elections (2019)
- Prediction markets (2004)
- Web3 Metaverse: State-of-the-art and vision (2023)
- When digital economy meets web3. 0: Applications and challenges (2022)
- Decentralized society: Finding web3’s soul (2022)
- Results from a dozen years of election futures markets research (2008)
- Wishes, expectations and actions: a survey on price formation in election stock markets (1999)
- Corporate prediction markets: Evidence from google, ford, and firm x (2015)
- Distilling the wisdom of crowds: Prediction markets vs. prediction polls (2017)
- Use of Brier score to assess binary predictions (2010)
- The power of prediction markets (2016)
- Prediction market accuracy in the long run (2008)
- Explaining variance in the accuracy of prediction markets (2019)
- How crowdsourcing improves prediction of market-oriented outcomes (2016)
- How to motivate people to put their money where their mouth is: What makes employees participate in electronic prediction markets? (2011)
- Bet Out the Vote: Prediction Markets as a Tool to Promote Undergraduate Political Engagement (2019)
- Evidence of in-play insider trading on a UK betting exchange (2012)
- Personality and the strength and direction of partisan identification (2012)
- Accessibility and the political utility of partisan and ideological orientations (1999)
- The social structure of political echo chambers: Variation in ideological homophily in online networks (2017)
- Interpersonal networks and political behavior (1982)
- Quantifying political leaning from tweets, retweets, and retweeters (2016)
- Social desirability bias (2010)
- Quantifying political leaning from tweets and retweets (2013)
- Fine-grained prediction of political leaning on social media with unsupervised deep learning (2022)
- Prediction markets for corporate governance (2006)
- Information transparency in prediction markets (2015)
- The Regulation and Value of Prediction Markets (2014) — link
- Blockchain based prediction markets (2022)
- Determinants of social desirability bias in sensitive surveys: a literature review (2013)
- Sensitive questions in surveys. (2007)
- The personalization of modern politics (2007)
- Correlation not causation: The relationship between personality traits and political ideologies (2012)
- Electoral and partisan cycles in economic policies and outcomes (2002)
- PageRank for ranking authors in co-citation networks (2009)
- Use of R2 in accounting research: measuring changes in value relevance over the last four decades (1999)
- A discussion of low r-squares: Concerns and uses (2000)
- The wisdom of amateur crowds: Evidence from an online community of sports tipsters (2019)
- Aggregation and social choice: a mean voter theorem (1991)
- Measurement validity: A shared standard for qualitative and quantitative research (2001)
- Understanding voters’ preferences: how the electorate’s complexity affects prediction accuracy and wishful thinking among politicians with respect to election outcomes (2012)
- Geneva: Evolving Censorship Evasion Strategies (2019)
0 of 73 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.