StatsBomb Data Analysis of 1,085 Matches Finds Late Soccer Goals Follow Predictable Patterns

·1 views

A data analyst examined 1,085 professional soccer matches from 2017 to 2022 using StatsBomb's publicly available open data, focusing on goals scored in the final 15 minutes of regulation. The study found that late-game scoring correlates at 79.3% with specific pre-match and in-match conditions, suggesting these goals are far from random. A key finding was the 'Desperation Window' between the 70th and 80th minutes, during which teams trailing by one goal sharply increased attacking intensity. In 312 such matches, 68.6% showed an immediate shift to attacking play, and teams generating shots in that window were more likely to either equalise or concede a second goal late on. The analysis concludes that factors like fatigue, tactical desperation, and defensive adjustments create measurable, repeatable frameworks around late-match scoring.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

AI Speeds Up Dev Work, But Leaves Some Engineers Questioning Ownership of Their Code

A software developer reflects on how AI-assisted coding has transformed their workflow, enabling them to ship complex billing systems in under a week that would previously have taken much longer. The author contrasts this with an early career memory of spending four days debugging a bank financing pipeline alone, describing the intense personal satisfaction that came from solving the problem independently. While acknowledging that AI catches edge cases and accelerates delivery, the developer feels a growing disconnect from the work being produced. The concern is not about competence or output quality, but about the loss of the hard-won, hands-on learning that once made shipped code feel personally meaningful. The piece raises broader questions about authorship and identity in software engineering as AI tools take on more of the problem-solving role.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Analyst Claims UFC Betting Market Has Systematic Mispricings After 500-Fight Study

A data analyst spent six months building a dataset of 500 UFC fights from January 2022 to June 2023, cross-referencing striking accuracy, takedown defense, fight duration patterns, and historical betting odds against actual outcomes. The study claims UFC sportsbooks systematically misprice certain underdog profiles, pricing fighters based on public perception and recent results rather than granular statistical analysis. The analyst cites Sean Strickland's +340 underdog victory over Dricus du Plessis at UFC 287 as a case study in market inefficiency. According to the findings, specific underdog profiles generate consistent positive ROI that would be unlikely if odds accurately reflected true win probabilities. The research draws on publicly available data from UFCStats.com alongside opening and closing lines from at least two major sportsbooks per fight.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Key principles engineers must follow when developing software with AI tools

A guide published on DEV Community outlines essential practices for software engineers integrating AI into professional development workflows. The article stresses that AI is a productivity tool, not a replacement for engineering judgment, architecture knowledge, or decision-making. It recommends using top-tier models such as Claude Opus and GPT with high-effort reasoning, paired with dedicated local agents like Claude Code or Codex for better output quality. Engineers are advised to always analyze and plan before implementation, maintain a structured project context file, and validate all AI-generated code through builds, tests, and static analysis. Ultimately, the article emphasizes that accountability for every line of code reaching production rests solely with the engineer, not the AI.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Why a Single Train/Test Split Can Mislead Your ML Model's Accuracy

A single 80/20 data split can produce misleading accuracy scores because the result depends heavily on which data points land in the test set. Cross-validation addresses this by dividing data into k equal folds, training the model k times, and validating on a different fold each round. This ensures every data point is used for both training and validation, producing a more reliable mean accuracy with a standard deviation. Best practices include fitting preprocessors inside each fold to prevent data leakage, stratifying splits for imbalanced datasets, and preserving a separate final test set untouched during tuning. The trade-off is k times the computational cost, but the payoff is a statistically honest performance estimate.

0 comments Read more at DEV Community