q-fin.STOct 21, 2026

What Prediction-Market Contracts Reveal at Launch: Trading Volume and Dispute Risk

Authors: Ahmed Mahrous, Roberto Di Pietro

Organizations: King Abdullah University of Science and Technology Thuwal, Saudi Arabia

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Using a near-census of 802,863 resolved markets listed on Polymarket between January 2023 and May 2026, and only information available at listing (the market's title and description, named resolution source, and contract metadata), we predict lifetime trading volume and whether settlement is later disputed. Contract information predicts trading volume well (out-of-sample R2=0.747R^2 = 0.747 for log volume), and most of the signal is interpretable: market topic provides the largest gain (+0.289+0.289 R2R^2 over a 0.2200.220 contract-feature baseline), wording and resolution-rule design add smaller but measurable improvements (+0.027+0.027 combined), and neural text models add little beyond the interpretable features. This performance does not survive a period of platform change coinciding with a shift in listing formats: accuracy in predicting whether a market reaches $1,000 falls from AUC 0.9400.940 under same-period evaluation to 0.5770.577 when training on past quarters and testing on later ones. The loss concentrates in markets whose question forms were absent from training (76.9% of test markets; on familiar forms the model still reaches 0.7780.778), and a decomposition attributes more of it to a changed relationship between contract characteristics and demand than to an outdated mapping from text to features. Dispute predictors are more stable but less precise: launch-time features rank dispute risk far better than chance (AUC 0.780.78), yet at a 0.31% base rate precision is too low for case-by-case prediction (average precision 0.0120.012), so the scores suit ranking groups of markets rather than forecasting individual disputes.

Figures & tables

Figure 1. Each point is one NMF topic, sized by market count. Topic separates high- from low-traction segments. Scatter plot with one point per dominant NMF topic, plotting topic prevalence against the share of markets reaching one thousand dollars of volume; colors indicate the broad topic family (Crypto/Finance, Sports, Esports, Weather, Politics/Media, Other) and bubble area indicates the number of markets in the topic.
Panel
Analysis sample (Sec. 3.1 )802,863
Events (clusters)391,955
Markets per event (median / mean / max)1 / 2.05 / 144
Creation window2023-01 – 2026-05
Share created in 2025 / 202623.6% / 74.3%
Trading volume
Table 1. Analysis sample and target variables. The market is the unit of analysis; the event is the cluster used for all splits and standard errors.
Figure 2. Out-of-sample fit as feature blocks are added. Topic contributes the largest increment. Model ladder showing out-of-sample R squared for log-transformed volume and AUC increasing from basic contract features to text models, with neural text benchmarks near the same ceiling.
Figure 3. Out-of-fold predicted vs. observed volume (M5). Dashed line: perfect calibration; dots and bars: decile means and interquartile ranges. Hexagonal-binned density of out-of-fold predicted log-transformed volume on the horizontal axis against observed log-transformed volume on the vertical axis, shaded by the number of markets per hexagon on a logarithmic scale. The mass lies along the diagonal. Ten white dots mark decile means of observed volume with white interquartile bars, and a white dashed diagonal line marks perfect calibration.
Example recurring title templateMarketsVolume ≥\geq $1,000Dispute rate
Crypto up/down, 15-min windows343,09870.5%0.40%
Sports over/under totals67,14955.3%0.12%
Daily temperature ranges60,26088.2%0.39%
Tennis set-game totals11,44824.1%0.07%
Trump/political-media mentions10,99886.0%1.00%
Table 2. Selected recurring title templates. Counts and volume shares use all markets; dispute rates use the dispute-observable subsample.
Figure 4. Contract-design factors with topic and launch-quarter fixed effects on a one-standard-deviation scale. Forest plot of six continuous contract-design factors ordered by effect size, each shown as a point estimate with a 95 percent confidence interval, all on a one-standard-deviation contrast. Planned duration has much the largest positive estimate; edge-case coverage, conditional language, reading grade and specificity form a narrow positive band; description length is negative.
Excluded family% panelR2R^{2}AUC
None (full model)—0.7470.940
Crypto/Finance45.40.6680.896
Sports28.00.7850.957
Esports9.70.7520.945
Other events8.10.7550.941
Weather6.30.7540.940
Table 3. Leave-one-family-out robustness: each row recomputes the reference model’s out-of-sample fit with one broad topic family excluded.
Evaluation settingAUC
Train and test in the same period (cross-validation)0.940
Train on the past, test on the future (cutoff 2025Q4)0.577
question form never seen in training (77% of test)0.575
question form seen in training (23% of test)0.778
Diagnostics on later-period data: 
before / after monotone rescaling (same 90% subset)0.597 / 0.597
Table 4. Predicting whether a market reaches 1,000(AUC;chance1,000 (AUC; chance =0.5$ ), boosted model. Recalibration is fit on the earliest 10% of the test period.
 Relationship (features →\to volume)
Text representationfrom the pastre-learned
From the past0.0110.772
Re-learned0.2490.773
Table 5. Attributing the 2025Q4 loss: out-of-sample R2R^{2} on the same later markets, with each part kept from the past or re-learned from recent data.
Figure 5. Contract characteristics of disputed markets: matched conditional-logit odds ratios with 95% event-cluster bootstrap intervals; grey marks intervals crossing 1. Unstarred rows: topic-by-quarter-by-source-type match; starred row: topic-by-quarter match, Wald interval. Forest plot of matched dispute odds ratios for resolution-source presence, neg-risk structure, edge-case coverage, conditional markers, readability, description length, title-to-description ratio, and edge-case clauses.
More in this topic

Prediction Markets

Prediction markets aggregate beliefs about uncertain events through tradable contracts whose prices encode implied probabilities.

See all 21 papers

Explore similar work

CardsList
Apr 22, 2026

Unlocking the Forecasting Economy: A Suite of Datasets for the Full Lifecycle of Prediction Market: [Experiments & Analysis]

Prediction markets are markets for trading claims on future events, such as presidential elections, and their prices provide continuously updated signals of collective beliefs. In decentralized platforms such as Polymarket, the market lifecycle spans market creation, token registration, trading, oracle interaction, dispute, and final settlement, yet the corresponding data are fragmented across heterogeneous off-chain and on-chain sources. We present the first continuously maintained dataset suite for the full lifecycle of decentralized prediction markets, built on Polymarket. To address the challenges of large-scale cross-source integration, incomplete linkage, and continuous synchronization, we build a unified relational data system that integrates three canonical layers: market metadata, fill-level trading records, and oracle-resolution events, through identifier resolution, on-chain recovery, and incremental updates. The resulting dataset spans October 2020 to March 2026 and comprises more than 770 thousand market records, over 943 million fill records, and nearly 2 million oracle events. We describe the data model, collection pipeline, and consistency mechanisms that make the dataset reproducible and extensible, and we demonstrate its utility through descriptive analyses of market activity and two downstream case studies: NBA outcome calibration and CPI expectation reconstruction.

Huaiyu Jia, Luofeng Zhou, Wentao Zhang +3

Nanyang Technological University Singapore, Singapore

Data IntegrationPrediction Markets
AskPDF
Dec 2, 2025

Agentic AI for Clustering, Relationship Discovery, and Semantic Trading in Prediction Markets

Prediction markets allow users to trade on outcomes of real-world events, but are prone to fragmentation with overlapping questions, implicit equivalences, and hidden contradictions across markets. We present an agentic AI (AAI) pipeline that autonomously recovers cross-market structure from contract text before prices enter the analysis. The workflow first clusters markets into coherent topical groups using natural-language understanding over contract text and metadata, and then identifies contracts within each cluster, but from different event markets, that exhibit strong dependence or leader--follower relationships. We evaluate this system, along with a natural language inference (NLI) benchmark, on a large prediction market dataset from early 2026. Using resolved outcomes to evaluate identified relations, we find that AAI-identified relations are 62.8% consistent with exchange-recorded settlements, whereas the NLI benchmark only achieves 40.6% accuracy. Within clusters, the AAI output is sparse and also remarkably compatible as a signed graph with a frustration rate of 0.324%. As an application, we show how discovered relations inform semantics-based trading strategies on prediction markets. One such strategy yields 14.12% net ROI after fees in a two-month period in 2026. Overall, we demonstrate the potential for agentic AI as a structural discovery layer for prediction markets.

Agostino Capponi, Alfio Gliozzo, Brian Zhu

Columbia University, Columbia Business School · IBM · Columbia University

Relation ExtractionClusteringAgentic AIPrediction Markets
AskPDF
Apr 3, 2026

PolyBench: Benchmarking LLM Forecasting and Trading Capabilities on Live Prediction Market Data

Predicting real-world events from live market signals demands systems that fuse qualitative news with quantitative order-book dynamics under strict temporal discipline -- a challenge existing benchmarks fail to capture. We present PolyBench, a multimodal benchmark derived from Polymarket that records point-in-time cross-sections of 38,666 binary prediction markets spanning 4,997 events, synchronously coupling each snapshot with a Central Limit Order Book (CLOB) state and a real-time news stream. Using PolyBench, we evaluate seven state-of-the-art Large Language Models -- spanning open- and closed-source families -- generating 36,165 predictions under identical, timestamp-locked market states collected between February 6 and 12, 2026. Our multidimensional framework assesses directional accuracy, our proposed Confidence-Weighted Return (CWR), Annualized Percentage Yield (APY), and Sharpe ratio via realistic order-book execution simulation. The results reveal a pronounced performance divergence: only two of seven models achieve positive financial returns -- MiMo-V2-Flash at 17.6% CWR and Gemini-3-Flash at 6.2% CWR -- while the remaining five incur losses despite uniformly high stated confidence. These findings highlight the gap between surface-level language fluency and genuine probabilistic reasoning under live market uncertainty, and establish PolyBench as a contamination-proof, financially-grounded evaluation standard for future LLM research. Our dataset and code available at https://github.com/PolyBench/PolyBench.

Pu Cheng, Juncheng Liu, Yunshen Long

College of Electrical Engineering, Sichuan University, Chengdu, Sichuan, China · School of Economics, Sichuan University, Chengdu, Sichuan, China

Forecasting BenchmarksLLM EvaluationPrediction Markets
AskPDF