Sports Analytics Weekly by kubeia.io - 21/2024

Spain are favorites to win what looks to be the most hotly contested group at Euro 2024. At the time of the draw, Group B contained the sixth, seventh and eighth highest-ranked European teams in the FIFA rankings, with the Spanish at the front of that queue.

On Tuesday, sportsbook operators made their feelings known without saying a word regarding the famous sports betting practice of limits.Long-term winning sports bettors know the battle to defeat the bookies isn't restricted to developing a winning strategy. A bigger duel occurs between winning sports bettors and sportsbooks — the limiting of players.

Anthony Bruno and Scott Cullen preview the NHL Conference Finals, sharing their best bets and predictions for the Eastern Conference Final between the Florida Panthers and New York Rangers and the Western Conference Final between the Edmonton Oilers and Dallas Stars. They also discuss Stanley Cup Futures and give their picks to win it all based on Pinnacle's odds

As Manchester City closed in on their fourth Premier League title in a row (and their sixth in seven years), their outstanding manager Pep Guardiola was asked whether this might be a little, well, boring.The response was forthright: “It's boring? It's not. It is so difficult. Before do you know what it was? It was the money.”

At StatsBomb, we are committed to providing education and resources to support the development of the next generation of analysts. One way we've been doing this is through regular releases of our industry-leading data to enable analysis of interesting datasets, teams and players. We're excited to bring you this latest release: Bayer Leverkusen's unbeaten Bundesliga title win under Xabi Alonso – the first league title in the club's history.

In 2004, Goalimpact (or its predecessor) was established. The algorithm's founder, Jörg Seidel, accepted an invitation to wager on the 2004 European Championships from his old coworkers who worked as investment bankers. Jörg was not deeply interested in soccer (at the time; he now supports Werder Bremen), but he was skilled at creating models thanks in large part to his training in finance and physics as well as his proficiency with computer programming. To fill up his betting sheet, he ultimately developed an algorithm for the tournament that determined the likely outcomes of each game.

On-Ball Value (OBV) is a Possession State Value (PSV) model developed by StatsBomb. Unlike other models such as expected threat (xT), On-Ball Value not only measures the contribution a player makes to their team's chances of scoring. Learn more about On-Ball Value - https://statsbomb.com/articles/soccer/introducing-on-ball-value-obv/

By hedging systematic risk, we can increase the number of independent bets and improve breadth. This increase in breadth leads to better performance and higher returns. That’s why managing systematic risk is important for optimizing the Sharpe Ratio.

We investigate whether an LLM can successfully perform financial statement analysis in a way similar to a professional human analyst. We provide standardized and anonymous financial statements to GPT4 and instruct the model to analyze them to determine the direction of future earnings. Even without any narrative or industry-specific information, the LLM outperforms financial analysts in its ability to predict earnings changes. The LLM exhibits a relative advantage over human analysts in situations when the analysts tend to struggle. Furthermore, we find that the prediction accuracy of the LLM is on par with the performance of a narrowly trained state-of-the-art ML model. LLM prediction does not stem from its training memory. Instead, we find that the LLM generates useful narrative insights about a company's future performance. Lastly, our trading strategies based on GPT's predictions yield a higher Sharpe ratio and alphas than strategies based on other models. Taken together, our results suggest that LLMs may take a central role in decision-making.

A better understanding of inductive biases could lead to a more creative and data-generating process-friendly way of modeling.This newsletter series (you are reading part 1), is my attempt to catch my thoughts before they elude me again. And to manifest my newfound enthusiasm for inductive biases.

Kolmogorov-Arnold Networks (KANs), inspired by the Kolmogorov-Arnold representation theorem, are promising alternative to neural networks (NNs). Coming out of MIT, KANs have been making waves everywhere you look, from Twitter to forums. The authors have some strong claims and it seems like everyone has boarded the hype train, but do they live up to the claims made? What are they and how do they work? Well, I will answer all of the following in this post and hopefully demystify some of the horrible jargon and notation that comes with them

When learning a tensor programming language like PyTorch or Numpy it is tempting to rely on the standard library (or more honestly StackOverflow) to find a magic function for everything. But in practice, the tensor language is extremely expressive, and you can do most things from first principles and clever use of broadcasting.This is a collection of 21 tensor puzzles. Like chess puzzles these are not meant to simulate the complexity of a real program, but to practice in a simplified environment. Each puzzle asks you to reimplement one function in the NumPy standard library without magic.

For years, we’ve used percentage of passes completed as an evaluation tool for how good a passer a player is. The problem is that basic passing percentages are meaningless for player evaluation.

Recently, I attended AAAI ‘24 to present a paper that got accepted titled “Robust Uncertainty Quantification using Conformalised Monte Carlo Prediction”. In this blog post I will walk you through the paper in a more informal manner and how our method, MC-CP works. Take a look at any of the links above before you read the rest of this post.MC-CP is a conformal prediction algorithm that produces sets/intervals of smaller size without any reduction in prediction performance. So as a nice easy walkthrough I will cover a brief introduction into conformal prediction, the MC-CP method, our results, and finally where future work could look towards. I encourage you to check out our GitHub implementation also!

Here are the probabilities (in %) -generated over 10.000 simulations- with all match-results based on NT prediction formula's for goals scored in a match depending on elo win expectancy.

European football’s governing body said that multi-club ownership “has the potential to pose a material threat to the integrity of European club competitions, with a growing risk of seeing two clubs with the same owner or investor facing each other on the pitch.”