Skip to content
All library documents

Pairs Trading in Election Betting Markets with a Mean-Reverting Probability Model

Article arXiv papers · Author: Haoyu Liu et al.
Read the original

Summary

This case study adapts pairs-trading logic to political betting markets by modeling the combined implied probability of the two major-party nominees. A latent Ornstein-Uhlenbeck process with a time-varying mean-reversion level represents the underlying probability, while additive noise accounts for observed odds. Parameters are estimated from regularly sampled data with a state-space likelihood; repeated consecutive quotes are compressed while preserving their elapsed time in the continuous-time transition.

Parametric-bootstrap upper prediction bounds generate signals when the combined probability is expected to decline, and a Bradley-Terry-type model chooses which candidate’s odds quote to use. The candidate-selection model is trained on 2020 election data and evaluated on 2024 data, producing 130 signals, 95.1% one-step coverage, a 1.86% mean synthetic odds-price return, and a 1.12 unannualized per-trade Sharpe-type ratio. These are frictionless descriptive measures, not executable betting profits; the authors present the method as a proof of concept for two-candidate markets.

Key ideas

  • The framework models the combined implied probability of two nominees with a noisy, time-varying mean-reverting process.
  • A state-space likelihood estimates model parameters while accounting for irregular elapsed time between retained observations.
  • Bootstrap upper prediction bounds identify potential declines in combined implied probability.
  • A Bradley-Terry-type model selects the candidate-specific odds quote.
  • The out-of-sample 2024 results are descriptive and exclude betting-exchange frictions.

Tags

Mean reversionStatisticsEvent-driven
Full text
# Adapting Pairs Trading to Gambling Markets A Case Study of the U.S. Presidential Election


# Adapting Pairs Trading to Gambling Markets A Case Study of the U.S. Presidential Election









Pairs trading exploits mean reversion in the relationship between related assets. We adapt this idea to political betting markets by modelling the combined implied probability of the two major-party nominees with a latent Ornstein-Uhlenbeck process whose mean-reversion level varies over time and whose observations contain additive noise. Model parameters are estimated from regularly sampled odds data using a state-space likelihood, with consecutive repeated values represented by a single retained observation and the elapsed number of sampling intervals preserved in the continuous-time transition. Parametric-bootstrap upper prediction bounds identify signal times at which the combined implied probability is likely to decline, and a no-intercept Bradley-Terry-type model selects the candidate-specific odds quote. The candidate-selection model is trained on 2020 U.S. presidential-election data and evaluated out of sample on 2024 data. The 2024 analysis produced 130 signals, empirical one-step coverage of 95.1%, a mean synthetic odds-price return of 1.86%, and an unannualized per-trade Sharpe-type ratio of 1.12. These returns are frictionless descriptive quantities rather than executable betting-exchange profits. The results support the integrated framework as a proof of concept for two-candidate electoral markets.

Shown in full with attribution under the source's licence. Licence: abstract CC0

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.