Varsity Football Insights - Wisdom of Crowds
Enter our prediction competition ahead of the Women’s & Men’s Varsity football matches at the Cledara Abbey Stadium on Friday 21 March 2025
Make 3 predictions on the Women’s and Men’s Varsity matches
Win 10star merch if your predictions are correct – or the closest of all entries
Enter here before KO at 16:00, Fri 21 Mar – full T&Cs are on this page
The 10star Summer Intern Program is designed to introduce undergraduate and graduate students to how we apply data science and quantitative modelling within the sports betting industry.
Through this program, interns will have the opportunity to work on real projects, collaborate with experienced professionals, and gain insights into the workings of a constantly evolving and fast paced industry.
Timeframe
- June - September 2025
Location
- Hybrid working, with regular onsite sessions at our offices in Camden, London
Eligibility
- Strong academic performance in a Mathematical discipline
- Experience using a coding language such as R, Python, C#, VBA
- Ability to work independently and in a team environment
- Interested in sports and thinking about them in an analytical way
Application Process
- Interested candidates should submit their resume and a cover letter outlining their interest in the internship program to careers@10star.com with the subject line "Summer Intern Program Application"
- Closing date for applications 30 April 2025
- Shortlisted candidates will be contacted for interviews
Benefits
- Professional Development
Work closely with experienced professionals and participate in learning sessions tailored to enhance your skills - Networking Opportunities
Connect with professionals across various departments and build relationships that could benefit your career in the future - Hands-on Experience
Engage in meaningful projects and tasks that contribute to the company's goals - Remuneration
Competitive salary offered for the duration of the internship - Flexibility
Work around your term dates, study requirements and holidays
The Wisdom of Crowds is a theory around how the collective opinion of a diverse and independent group of individuals will provide a more accurate answer than that of a single expert.
To showcase this we ran some predictive analysis on the match at 10star and compared it to the data collected from the predictions made on our Varsity Football Prediction Competition.
Match Result
We would expect that the crowd for this type of event would not be particularly diverse or independent and that we would likely see biases in the responses.
Men's 1sts
We can see from this data that the crowd was much more skewed towards Cambridge than we analytically expected and this is common across any home team bias. Crowd analytics of football results also tends to see the Draw under represented as people prefer predicting something rather than nothing to happen. A major reason for this is that teams are not independent and loss aversion can often cause teams to settle for a draw rather than push for a win. A popular paper by Dixon & Coles provides more insight into this.
Total Goals
Goals, being events over time, tend towards a Poisson distribution. Overall expectancies have fluctuated over time as rules, playing conditions and even play styles change. A common theme throughout however is that knockout games generally have lower goals than the equivalent league based games.
Home advantage is well known throughout sport and traditionally teams could expect an improvement of around 0.2 of a goal on average. During Covid the observed results showed home advantage across major football eradicated to almost nothing, highlighting the impact of a positive supporting crowd. Interestingly this year in the Premier League we are seeing almost no home advantage, although this is a small sample.
Women's 1sts
Correct Score
Extreme results are generally overestimated by people as they are harder to predict. This is a well known phenomenon called the Favourite-longshot bias
Women's 1sts (Prediction Difference)
Time of First Goal
Time of goal is heavily correlated with the goal expectancy and using this as a start point there are simple ways we can calculate a distribution. Nuances come into play with the variability of scoring rates throughout each half with a very obvious spike at 45+ mins and 90+ mins due to added on time. For games with technology such as VAR it not uncommon for games to go to around 100 mins which has a huge impact on this analysis. It brings a whole new meaning to the term "last minute goal" when this is actually accounting for around 10% of the match itself.
Men's 1sts
Method of Victory
Predictions for Method of Victory should be consistent with those on other related areas such as Match Result.
Men's 1sts
In sports analytics a major area of research is around momentum. Generally, Method of Victory is a good example of this. If a game does go to Extra Time it is likely that the underdog is overperforming vs the favourite and therefore we should update our ongoing beliefs. This cannot be lazily applied though. In certain scenarios a team might have fielded an understrength lineup to begin with, knowing that if the match is close they can bring on higher quality players. A good example of this is Premier League teams v lower league opposition in the early rounds of the cups. This momentum tends not to carry over into penalties where instead we would expect to see a very small bias for the home team. This then gets outweighed by a medium bias for the team going first in the shootout, showcasing the effect of pressure on performance.
