University Project Proposals – August 2010

1. MSc/Final year Individual/Group project: Determining a Hedging Strategy for Risk-based Products

Betfair has several risk-based products (Multiples, Sportsbook, Select) which take bets from customers.  Consequently, Betfair can be exposed to a significant profit or loss based upon the outcome of the events on which the customers bet.  The standard method that bookmakers use in order to balance their books is to adjust the prices that they offer in an ongoing manner so as to make each or the runners or selections more or less attractive and hence encourage or discourage more bets.

Since Betfair also operates an exchange, much like the Stock Exchange, where customers can buy and sell bets, it has an additional option for balancing its books.  Hedge bets can be placed in the exchange to reduce the level of risk on a market.  The manner in which these bets are placed has an impact on the cost and effectiveness of the hedging.  Placing large bets at unfavourable prices distorts the market and incurs a significant cost, but does ensure that risks are mitigated.  Placing several smaller bets at competitive prices is more efficient but risks the market moving against us, or the hedge bets being unmatched (unplaced).

The purpose of the project is to determine strategies for the optimal hedging rate for a standard range of markets and circumstances.  It is envisaged that Expected Utility Functions will be used to measure the level of risk associated with a set of bets.  Models will be derived to characterize typical market behaviour and movement.  Finally, these models and functions will be used to simulate the effectiveness of various hedging strategies.

2. Project: Betfair ‘Trusted Circles’ – Developing a working prototype

Betfair is the world’s leading betting exchange which offers an online platform for users to bet against each other on the outcome of sports events. Its Exchange platform has a very large number of active users with diverse betting behaviours, levels of ‘sophistication’, strategies and objectives.

This project will take previous analysis & knowledge around identifying where and how Betfair’s interface features could be improved to enhance the user experience, in particular, through the building of ‘Trusted Circles’ between similar types of user, and facilitating the exchange of information between those users. Users will then be able to share tips with their personalized trusted community, be alerted about tips, create or access statistics made available by their trusted community, and generally communicate via a community forum. This can be Betfair specific, or could be aimed at industries outside of Betfair (e.g. Financial Trading). They would need to assess whether it could be implemented as part of ongoing projects (Betstore & Community), or whether it would need to be considered as a separate entity. Should focus on two key areas – how to enable ‘Trust’ within the group (i.e. ratings, recommendations etc etc) & how groups should be selected (i.e. matched automatically with like-minded users, or user-generated groups).

This project will be about taking the work done previously around implementing ‘Trusted Circles’, & developing several prototypes as to how it could be implemented in practice. Careful thinking should be done around whether to maintain it as ‘Betfair Specific’ or whether it could be a more generic concept applied across other industries, but using some of Betfair’s key principles & technologies.

3. Project: Identifying factors that influence bet sizes amongst recreational punters.

Betfair is best known for operating the world’s leading betting exchange, a concept that effectively allows punters to bet against each other on a wide range of racing, sporting and other events.  Betfair’s wagering product suite has recently expanded to include tote and risk-based products of a type typically offered by bookmakers.  In some jurisdictions Betfair also offers gaming products (eg poker and casino) but those products are beyond the scope if this study.  Importantly for the availability of a broad and credible range of data, Betfair operates an exclusively account-based betting platform, where the identities of all customers are known, as is their funding and betting behaviour.

The vast majority of punters – whether they bet on betting exchanges, with totes or with bookmakers – are net losers.  The scale and rate of their losses will be determined by a range of factors, including but not limited to the punters’ strategies and objectives and the operating margins of the relevant betting agency.  Punters who are, or are destined to be, net losers are normally categorised as “recreational’ punters.  This is, they wager for enjoyment and the hope, but not the reasonable expectation, of it being a profitable pastime.  Like any other form of entertainment, therefore, the benefit comes at a cost.

Whilst most punters have “budgets” that they are prepared to spend on wagering each week, month or year, there is little known about the basis on which recreational punters decide how much of their budget they are prepared to risk on a particular outcome.  For instance, is it determined by betting market factors such as odds on offer, the margin of the operator; or by past results, such as a winning or losing streak; or by external factors such as general economic conditions, consumer confidence, interest rates, etc?  A number of more specific questions arise, such as whether the global financial crises – and its varying impact, say, in Australia compared to the UK – impact the size (if not the number) of bets placed by punters before and after the event?  Did recreational punters deplete their funds faster pre or post the GFC?  Were they faster or slower to re-fund their accounts pre or post GFC?

4. Project: Statistical analysis of the performance of horses at the top of the Betfair win market that significantly drift pre-race

Betfair is the world’s leading betting exchange which offers an online platform for users to bet against each other on the outcome of sports events. Two of its key unique selling points are the ability for customers to ‘lay’ or ‘bet against’ a selection winning and the unprecedented information Betfair provides customers on the historic activity within a market.  As such, a customer can go to a Betfair market, check the price and trading history of a selection before deciding to back or lay it, either now at the best price available or in the future at a requested price.

Betfair’s proactive work with administrative and regulatory bodies in sport has led to a number of high profile investigations around events that have been flagged for suspicious betting patterns.  Integrity is a key value for Betfair, every bit as important operationally as it is from a brand perspective.  Over time, detractors have tried to cite Betfair as the primary source of a new wave of corruption in sport, ignoring the indelible audit trail that Betfair now offers to prevent such corruption. There are some (from a cross section of customers, competitors, racing stakeholders, media etc.) who would claim that a significant drifter in a Betfair horse racing market is in itself evidence of some level of insider trading and/or corruption, something  that would be less prevalent if betting exchanges and Betfair specifically never existed.  It may be useful from a PR and regulatory perspective, if we had access to a confidential statistical study of the relationship between horses that drift significantly and their performance in those markets.  It should be possible to get all of the historic trading information via the Betfair API and would then just be a case of setting the parameters for the study (should it just be UK & IRE racing? How many horses per race – just the fav, top 3 in the market, any min/max odds?  What would represent a statistically significant drift at different bands of odds? Should there be a specific pre-race timeframe applied? Etc). We have done a similar study before in 2005 and some details of it can be seen here http://promo.betfair.com/tactemails/camb1a.htm and here http://betting.betfair.com/horse-racing/general/steamers-and-drifters-a-myth-worth-busting-120708.html.

What I would expect to be delivered is a definitive & statistical view on how relevant the pre-race drift of horses on Betfair is to their eventual performance.  Ideally, the study would show that despite drifting pre-race, a high % of horses go on to win their races meaning that the fluctuations in price that can be seen pre-race, in the vast majority, represent no more than varying degrees of customer opinion, standard behaviour to be seen in a liquid trading market. A lot has happened since the last study in 2005 so it would be useful to get an updated view here.

5. Project: A betting integrity monitoring tool

Betfair is the world’s leading betting exchange which offers an online platform for users to bet against each other on the outcome of sports and other events. The Integrity Team has a remit to prevent, detect and investigate unusual activity across the Betfair betting platforms. Betfair now has 47 Memorandum of Understanding (MOU) agreements in place with sporting regulators; over 4m registered users and offers approximately 11,600 exchange markets per week across 40+ sports. These MOU agreements allow Betfair to share customer account details and customer transaction data with the relevant sporting bodies should there be concern around the betting behaviour of particular accounts. With the increasing number of Betfair accounts and the growth in the numbers of betting markets offered by Betfair it is becoming a greater challenge to monitor the website without more technological assistance. The recent media reports about the alleged spot fixing by Pakistan have shown that integrity in sport and in betting markets is a high profile subject and has the potential to be an area of large reputational risk for Betfair.

In order to assist in continuing to fulfil its remit successfully the Integrity Team would greatly benefit from an integrity monitoring tool that maximises the technological capabilities of the company. At present a significant amount of betting market monitoring is undertaken by the analysts in the team manually watching the live Betfair site for price and volume changes. A monitoring tool would improve the efficiency of this task. Additionally the market monitoring carried out by the team requires the manual execution of SQL and Business Objects queries, which is both time-consuming and dependent on the Data Warehouse being up-to-date. The Integrity Team envisage the monitoring tool using event processing technology to generate real-time parameter-driven alerts that allow the team to monitor accounts and markets in real-time. This would assist the team in coping with an increased workload in an efficient and cost-effective way. Additionally, a monitoring tool will help the team continue to offer MOU partners the best integrity service in the industry. The Integrity Team can provide examples of reports that are currently used to manually analyse live betting data.

This project will produce a working prototype of the tool that will include a group of first phase alerts. These alerts will be of a more basic nature and will trigger on things such as an account depositing a certain % more than its average deposit, an account taking a liability a given % more than its average liability (lifetime or within a given period), an alert for when a certain % of a market has been taken by one account or an alert to highlight a given % change in the odds of a selection for selected markets and sports. The tool should be user-friendly and display the data in a format that is clear and engaging to the user. Most of the account alerts and some of the market alerts will require comparison against historical data. Once the team have a working prototype encompassing basic alerts the team would look to increase the scope and sophistication of the tool to help with more advance monitoring, this however would be in stage two of the project.

6. Project: To categorize the different types of gaming and establish how betting could be applied to these categories

Betfair is the world’s leading betting exchange which offers an online platform for users to bet against each other on the outcome of sports events and play against each other in Poker and other games combining chance and skill

This project will investigate other peer-to-peer online gaming and explore the suitability and feasibility of applying wagering to games currently played more commonly for bragging rights, both by individuals and self-organised teams,

This project will categorise the most popular online gaming models and propose a betting mechanism that could be applied to each, taking into account skill grading or handicapping to appeal to a braod range of player abilities.

7. Project: Modelling Betfair’s “eco-system”

Betfair is the world’s premier sports betting exchange where thousands of people come together to bet against each other on sporting outcomes. The betting exchange platform is a complex ‘ecosystem’ where a multitude of factors and circumstances come into play and which drive a market’s liquidity.

This project will seek to understand how some of these factors influence markets. For example, the addition or subtraction of specific customer groups to enable us to model the impact. A “model ecosystem” would be invaluable in helping to predict a multitude of scenarios. For example:

· We have a number of liquidity provision deals in place with specific individuals. To quantify incremental return we currently ask the provider to cease activity in a subset of target markets, allowing us to perform a statistical analysis of the trial and control groups. A market simulator would help us to quantify the impact and proactively identify markets that need help, potentially replacing our current trial and error approach.

· Modelling the impact of removing court-siders (customers betting with a temporal advantage over the market). This is something that we believe will increase the sustainability of the exchange and whilst we believe the market would be positively impacted, we do not have a robust way to model ahead of turning off a group of customers.

· Modelling targeted pricing changes – commission ladder optimisation, discriminatory profits charging etc

· Quantifying the impact of ring-fencing customer groups (Italy, Greece, Denmark etc).

This project will deliver a prototype capable of modelling the Betfair ‘ecosystem’ in respect of the above mentioned factors.