Profiting from Election Arbitrage: A Data-Driven Approach to Betting Opportunities
Introduction
The betting market for elections has grown exponentially, offering not just a platform for placing bets but also a rich dataset for analysis. By examining the differing odds across various bookmakers, we can identify arbitrage opportunities and gain valuable insights into market dynamics. These price discrepancies appear to be driven by the different client bases served by each provider. This article details our successful automated trading strategy during the 2024 U.S. Election and demonstrates how data-driven approaches can reveal inefficiencies in election betting markets.
Understanding Arbitrage Betting
Arbitrage betting involves placing bets on all possible outcomes of an event across different bookmakers to guarantee a profit. This is possible when bookmakers have differing opinions on the likely outcome, leading to discrepancies in the odds they offer.
There are two primary approaches to arbitrage betting:
- Dutching: When Bookmaker A offers higher odds for Candidate X while Bookmaker B offers higher odds for Candidate Y, strategically placing bets on both candidates can ensure your total payout exceeds the total amount wagered.
- Back-Lay Strategy: This involves backing (betting on) a candidate at higher odds with a bookmaker while simultaneously laying (betting against) the same candidate at lower odds on a betting exchange.
Gathering Odds Data
We gathered odds data from multiple reputable platforms:
- Betfair Exchange
- PredictIt
- Polymarket
- Oddschecker (aggregating multiple traditional bookmakers)
Our data collection approach utilised both web scraping and API integrations:
- API Integration: Direct access to Polymarket and Betfair data through their respective APIs (gamma-api.polymarket.com and betfairlightweight package)
- Web Scraping: Utilised Python tools including requests, BeautifulSoup, and Selenium for platforms without API access
- Infrastructure: Automated collection via Python scripts hosted on Google Cloud Platform (GCP)
I have detailed this data retrieval process in greater depth in previous articles.
Understanding Bookmaker Margins: The Overround
Bookmakers profit through the “overround” — where the sum of implied probabilities for all outcomes exceeds 100%, with higher overrounds generally meaning a worse deal for bettors. This margin allows bookmakers to earn money by balancing the bets placed on all outcomes, ensuring a profit no matter which outcome occurs.
The traditional bookmakers, along with PredictIt, were priced very defensively over the election, with an average margin amount ranging from 3.5% for William Hill to 6.1% for PredictIt. This allows them to be slower in updating prices alongside new information, as their margin acts as a buffer against better-informed bettors.
The Polymarket and Betfair Exchanges were, unsurprisingly, far tighter, with average overrounds well under 1% for both. However, Betfair does charge a commission on winning bets.
Market Inefficiencies and Trading Strategy
Our analysis revealed constant arbitrage opportunities in the week leading up the the election, which occur when the net overround is below 100%. This highlights the significant inefficiencies across different bookmakers, likely caused by the different flow of bets that each bookmaker receives.
We’ll take a look at a very simplistic Dutching arbitrage strategy to highlight how profitable these inefficiencies can be, although using the backing and laying strategy proved to be substantially more profitable. The results of this strategy were:
- $100 hourly bets on Trump with proportional Harris bets
- Total investment: $14,400
- $8,610 placed on Trump at average odds of 1.779
- $5,790 placed on Harris at average odds of 2.645
- Guaranteed return: $915 (6.4% ROI in one week)
PredictIt consistently had the likelihood of Harris winning the election far greater than other market participants, with the plot below highlighting one of the inefficiencies observed.
Market Dynamics Analysis
An attempt to investigate whether any of the bookmakers led another highlighted again how disconnected the markets were, with little correlation between the price changes on a minute-by-minute basis. This is likely due to the activity at each being driven by somewhat independent traders rather than broader market information.
The correlation between Betfair Exchange and Polymarket did appear to be slightly stronger than others, as seen in a cross-correlation chart aiming to identify if one was lagging behind another. If the highest correlation occurred at, say, 5 minutes, it would indicate that the prices on Betfair generally realigned themselves with Polymarket after 5 minutes, providing another way to capitalise on the inefficiencies. However, we can see that the spike is at 0, suggesting they were generally in sync.
In contrast, cross-correlation between Betfair and PredictIt is somewhat flat, suggesting that PredictIt prices tended to move independently. This is further confirmed by the overall correlation between the two being far lower than between Betfair and Polymarket.
The traditional bookmakers were similarly uncorrelated at the minute-by-minute level, as their price updates were very infrequent due to their large margin acting as a buffer.
Risk Factors and Considerations
- Market Volatility: Election markets can be highly volatile, with odds changing rapidly due to news events, polls, or debates. This volatility can make it challenging to execute arbitrage strategies promptly.
- Liquidity Constraints: Not all bookmakers or exchanges offer sufficient liquidity, especially for larger bets.
- Transaction Costs: Fees and commissions, such as Betfair’s commission on winning bets, can eat into arbitrage profits.
- Regulatory Restrictions: Betting regulations vary by jurisdiction, and some platforms may not be accessible in certain regions.
- Operational Challenges: Implementing automated trading strategies requires technical expertise and reliable infrastructure to collect data, process it in real-time, and execute trades efficiently.
Final Thoughts
Election betting markets present unique opportunities for data-driven traders. While our analysis demonstrates clear profit potential through arbitrage, success requires:
- Robust technical infrastructure
- Real-time data processing capabilities
- Understanding of market microstructure
- Careful risk management
The inefficiencies we identified suggest these markets are still maturing, potentially offering continued opportunities for sophisticated traders.
If you found this analysis valuable, please share it with others interested in quantitative approaches to betting markets. For questions or discussion, feel free to reach out.