Cairo University – Faculty of Engineering
MTHS204 0 Advanced Probability and Statistics
Fall 2024
Team Name: In the Net
Submitted to: Prof. Maha Amin Hassanein
Ahmed Medhat Abdelaziz 4230142
Abdelrahman Mohamed Abdelfattah 1220065
Abdelrahman Mohamed Kamal 1220255
Fayrouz Ahmed Magdy Moharam 1220136
Lujina Shawky Ahmed Shehata 1220138
Omar Waleed Elsayed Abo-Elnasr 1220070
Mazen Mohamed Sayed Khalifa (Leader) 1220018
1
Table of contents:
Abstract……………………………………………………………………….3
Introduction……………………………………………………………...……3
General Methodology……………………………………………………..….4
Statical Questions and numerical Results……………………………………6
Question 1……………………………………………………………………7
Question 2 …………………………………………………………………...9
Question 3……………………………………………………………………11
Question 4……………………………………………………………………13
Question 5……………………………………………………………………15
Conclusion……………………………………………………………..…….18
Appendix………………………………………………...…...………..…….18
References…………………………………………………………….……. 19
2
Abstract:
Penalties in football are high-stakes events that often decide the outcome of matches, presenting
unique tactical and psychological challenges. This research investigates the factors influencing
penalty success by analyzing data from six seasons. A multi-disciplinary approach was
employed, combining statistical analysis, psychological evaluation, and contextual modeling to
explore the interplay of physical attributes, mental resilience, and external influences.
Key findings reveal that the goalkeeper’s positioning and strategy, the penalty taker’s
psychological state, and situational factors such as match importance and crowd dynamics
significantly impact outcomes. The study highlights the complexity of these interactions and the
difficulty in isolating individual influences.
The conclusions emphasize the importance of tailored training regimes for penalty takers and
goalkeepers, as well as the role of psychological preparation in high-pressure scenarios. This
work provides actionable insights for coaches and analysts, aiming to refine penalty strategies
and enhance performance in critical moments.
Introduction:
Football penalties represent pivotal moments in the game, often determining the outcome of
matches and becoming a focal point for tactical and psychological considerations. These high-
stakes situations demand exceptional skill and composure, as they test not only the physical
capabilities of players but also their mental resilience. Key factors influencing penalty outcomes
include the goalkeeper’s reaction time and positioning, the penalty taker’s technique and
psychological state, and external elements such as crowd influence and match context.
This study delves deeply into these aspects by analyzing penalty data over six seasons, aiming
to uncover the multifaceted factors that contribute to success. By doing so, it provides a
nuanced understanding of how physical, psychological, and contextual variables interact to
shape outcomes in penalty situations.
Motivation
Understanding penalties is crucial for enhancing performance and decision-making in football.
This research offers valuable insights for improving penalty strategies, selecting the most
reliable penalty takers, and refining training methodologies. Moreover, it sheds light on the
psychological pressures and environmental dynamics that players face during these critical
moments, offering a broader perspective on human behavior under stress.
3
Objectives
1. Identify the primary factors that influence penalty success.
2. Explore the interplay between physical attributes, psychological preparedness, and
contextual influences.
3. Develop actionable recommendations for coaches, players, and analysts to optimize
penalty-taking and defending strategies.
Challenges
Solving the complexities of football penalties involves navigating several challenges:
1. Interplay of Variables: The interaction between physical, psychological, and contextual
factors is intricate and dynamic, making it difficult to isolate the impact of individual
elements.
2. Limited Data Context: Penalty outcomes are influenced by unique match situations such
as scorelines, player fatigue, and weather conditions, which are often challenging to
quantify comprehensively.
3. Psychological Pressure: The mental state of players under high-pressure scenarios varies
significantly, introducing unpredictable variables that are hard to measure or simulate
accurately.
4. Goalkeeper Influence: Goalkeepers employ diverse strategies, such as feints or
positional adjustments, to unnerve penalty takers, adding another layer of complexity.
5. External Influences: Factors such as crowd noise, match location, and referee decisions
can significantly affect outcomes but are difficult to control or replicate in studies.
General Methodology: -
Dataset Overview:
• Source: Six seasons of penalty data collected manually. Each penalty was reviewed by
video replay to obtain the necessary information.
• Sample Size: 628 penalties.
Key Variables:
1. Player Information: Includes the name, region, and footedness of the penalty taker.
2. Match Context: Captures details such as match location, time remaining, match
condition, and team category.
3. Penalty Characteristics: Encompasses the angle chosen for the shot and the penalty
outcome (goal, saved, missed).
4. Goalkeeper Information: Covers goalkeeper-specific data such as name and height.
4
Procedure:
1. Data Collection: Video replay was used to meticulously record each penalty’s details
across six football seasons.
2. Data Preprocessing: The data was cleaned and standardized to ensure consistency across
variables. Missing values were imputed where applicable.
3. Variable Categorization: Variables were categorized into physical, psychological, and
contextual factors for detailed analysis.
4. Model Development: Statistical models and machine learning algorithms were
developed to identify patterns and relationships between variables.
5
General rules used:
6
Statical Questions: -
Question 1: - Does the Relationship Between Goalkeeper Height and Penalty
Success Vary with the Shot Angle?
Methodology: We performed a multivariate regression analysis to assess how goalkeeper
height and penalty shot angle impact penalty success. Analyze interaction effects by including a
term for height × angle.
Dependent Variable: Penalty outcome (binary). (e.g., 1 = scored, 0 = missed).
Independent Variables: Goalkeeper height (numeric).
dummy variables penalty angles (e.g., right top, left bottom).
R² = 1.56%: The model explains only a small portion of the
variability in penalty success rates.
Significance F = 0.565 suggesting that the combination of goalkeeper height
and penalty angles does not reliably predict penalty success
in this dataset.
Scatter-plot
Left- Bottom Angle: Left -Top Angle:
Middle – Bottom Angle: Middle Top Angle:
7
Right – Bottom Angle: Right – Top Angle:
Hypothesis Testing:
Null Hypothesis (H₀): Goalkeeper height and penalty shot angles do not
significantly affect penalty success.
Alternative Hypothesis (H₁): Goalkeeper height and penalty shot angles significantly
affect penalty success, with potential interaction effects.
Hypothesis Result: we fail to reject the null hypothesis as the overall model and
individual predictors were not significant.
Implications: penalty success is likely influenced by factors other than
goalkeeper height and shot angle.
Limitations: -
Low R² Value: limited explanatory power, highlighting the need for additional
variables
Linear Model Assumptions outcomes may not follow a simple linear relationship with
goalkeeper height and angle, potentially requiring non-linear
models
Unexplored Factors Key variables like penalty taker experience, match pressure, and
goalkeeper positioning were not included
Conclusion: -While goalkeeper height and penalty angles show minimal influence on penalty
success in this dataset, this study provides valuable insights into the limitations of these factors.
The findings underline the complexity of penalty outcomes and the need for further research
incorporating additional variables and advanced analytical technique conclude to 3 sentences.
8
Question 2: How does match location (home/away) impact penalty success
rates for Big Six, average, and small teams?
Methodology: ANOVA Analysis to determine whether penalty success rates are influenced by:
1. Match location (home vs. away).
2. Team category (Big Six, Average, Small).
alpha (significance level) 0.05
Hypothesis Testing:
1. Match Location
Null Hypothesis (H₀): The mean penalty success rate is the same for home and away
matches. Match location has no effect on penalty success
rates.
Alternative Hypothesis (H₁): The mean penalty success rate is different for home and away
matches. Match location has an effect on penalty success
rates.
2. Team Category
Null Hypothesis (H₀): The mean penalty success rate is the same across all team
categories (Big Six, Average, Small). Team category has no
effect on penalty success rates.
Alternative Hypothesis (H₁): The mean penalty success rate differs between team
categories. Team category has an effect on penalty success
rates.
Results (1):
F-value 1.4299812
P-value 0.3543170863
9
Interpretation The p-value (p > 0.05) indicates that the difference in penalty success
rates between home and away matches is not statistically significant.
Hypothesis Result: We fail to reject the null hypothesis
Implications: Data indicates it is not enough to claim that Match Location has a
significant effect on penalty success rates
Results (2):
F-value 1.433269707
P-value 0.4109696501
Interpretation The p-value (p > 0.05) indicates that the difference in
penalty success rates among Big Six, Average, and small
teams are not statistically significant.
Hypothesis Result: We fail to reject the null hypothesis
Implications: Data indicates it is not enough to claim that team category
has a significant effect on penalty success rates.
The ANOVA results suggest:
1. Match location: does not significantly affect penalty success rates.
2. Team categories: does not significantly affect penalty success rates.
10
Question3: What is the impact of psychological pressure (time remaining and match
condition) on penalty success?
Methodology: we used logistic regression, where one category of each categorical variable was
omitted and treated as the reference category. This approach prevents redundancy and
multicollinearity in the model, ensuring that the independent variables are not perfectly
correlated.
Dependent: penalty outcome as binary (1 = scored, 0 = missed).
Independent: Time remaining and score difference (e.g., drawing, Winning,
Losing).
Reference: Losing
Intercept: The baseline odds of scoring (when trailing and with 0
minutes remaining) are significantly higher than not scoring,
with an odds ratio of 4.30.
Winning: Drawing:
Losing:
11
Null Hypothesis (H₀): The mean penalty success rates are the same across the different
match conditions (drawing, winning, and loosing). Time
remaining does not significantly affect the probability of scoring a
penalty
Alternative Hypothesis The mean penalty success rates differ across match conditions.
(H₁): Time remaining significantly affects the probability of scoring a
penalty
Tied Condition: The odds of scoring are about 32.53% lower than when trailing,
but the effect is not statistically significant (p-value = 0.1555).
Leading Condition: The odds of scoring are slightly higher (3.18%) when leading, but
this effect is also not statistically significant (p-value = 0.9134).
Hypotheses Result: We fail to reject the null hypothesis that: There is no significant
difference in penalty success based on match conditions (Tied,
Leading, Trailing). Time remaining does not significantly affect
the probability of scoring a penalty.
Implication: The coefficients for D-drawing and D-wining reflect the
differences in penalty success likelihood relative to the reference
category (loosing), while the baseline odds ratio (4.30) represents
the odds of scoring when trailing with zero minutes remaining.
Limitation:
Multicollinearity: The creation of dummy variables for all match conditions (Losing,
drawing, Winning) can lead to multicollinearity because the sum of the
dummy variables equals 1. This can make the regression coefficients
unstable and unreliable if not addressed.
Reference To avoid multicollinearity, one category (Trailing) is excluded as the
Category reference group. This means the model only provides insights about
Restriction: differences relative to the reference category, limiting the ability to
interpret absolute effects for all categories simultaneously.
12
Question4: How do penalty success rates differ across players from different countries,
and does this depend on their team’s reputation (Big Six, average, and small teams)?
Methodology: Chi-Square Test for Independence
1- Player’s region and team’s categories (Big Six, average, and small teams).
2- Dependent variable: Penalty success rate. Application
Steps: -
1-Group players based on their countries/regions (e.g., Europe, South America)
2- Create a contingency table of success rates by region and team categories.
3- Perform a chi-square test to assess independence.
13
Results: -
P-value 0.05%
Significance Level (α) 5% (0.05)
Degrees of freedom 25
Test Value 54.77
Critical Value 37.65
Hypothesis Testing:
Null Hypothesis (H₀): Penalty Success rate will not change upon changing the team
league and region
Alternative Hypothesis (H₁): Penalty success rate will change upon changing the team
league and region
Hypothesis Result: Test value > Critical value meaning, it’s in the rejection
region, reject H0
Interpretation penalty success rates vary significantly across countries,
influenced by the team’s reputation.
14
Question 5: How do penalty success rates differ across players from different penalty
angles and footedness (left-footed vs. right-footed)?
Methodology: A Chi-Square Test for Independence was conducted to determine whether penalty
success rates (scored vs. missed) depend on:
1-Penalty angle (Left Bottom, Left Top, Middle Bottom, Middle Top, Right Bottom, Right Top)
2-Footedness (Left-footed vs. Right-footed)
15
Results:
Chi-Square test statistic 22.048
P-value 0.1065
Significance Level (α) 5% (0.05)
Degrees of freedom 15
Test Value 22.05
Critical Value 24.996
Hypothesis Testing:
Null Hypothesis (H₀): Penalty success rate is independent of the penalty angle and
footedness.
There is no relationship between the penalty angle and
footedness with penalty success (scored/missed).
Alternative Hypothesis (H₁): Penalty success rate is dependent on the penalty angle and
footedness.
Penalty success varies significantly based on the penalty angle
and footedness.
Hypothesis Result: P-value is greater than 0.05, we fail to reject the null
hypothesis.
Interpretation Observed frequencies of penalty outcomes (scored or missed)
for various angles and footedness do not deviate significantly
from the expected frequencies, meaning the two variables
(penalty angle and footedness) do not influence the success
rates of penalties in this analysis.
Conclusion:
Based on the Chi-Square test, penalty success rates are not significantly influenced by the
penalty angle or footedness of the player in this dataset. Therefore, there is insufficient evidence
to suggest a relationship between penalty success and the angle/footedness variables.
16
Conclusion:
This study thoroughly explores the factors influencing penalty success in football, offering
several critical insights. It finds that goalkeeper height has minimal interaction with penalty
angles, and match location shows no significant effect on success rates across team categories,
challenging assumptions about home advantage. Psychological pressures, such as time
remaining and match conditions, also appear to have limited impact, emphasizing the primacy
of player skill and preparation. However, significant variations in success rates across players
from different countries, influenced by their teams’ reputations, highlight the role of cultural
and contextual dynamics. Furthermore, penalty success at specific angles is not linked to
footedness, suggesting technical precision outweighs biomechanical advantages.
These findings underscore the complexity of penalty scenarios, driven by an intricate interplay
of physical, psychological, and contextual factors. The insights are highly practical: coaches can
design tailored training regimens and optimize player selection, analysts can enhance predictive
models for strategy development, and teams can address cultural influences to improve
performance. Future research should investigate additional variables, such as goalkeeper tactics
and decision-making under pressure, to deepen understanding. Overall, this study offers
valuable guidance for refining strategies, enhancing player preparation, and improving
outcomes in crucial match situations.
Appendix:
Tools:
Excel sheet
Google Sheets: Utilized for data collection, statistical analysis, and visualizations. Google
Sheets was instrumental in creating pivot tables and charts for analyzing the dataset
XL Miner Analysis Tool Pak: Add-on in google sheets
Data Collection The dataset was collected manually. The process involved:
1. Gathering initial basic data about penalties from Transfer market.
2. Watching penalty videos one by one to record the required detailed information.
3. This labor-intensive process spanned approximately 40 hours.
Statistical Analysis
Google Sheets was used for:
○ Performing statistical analysis with pivot tables.
17
○ Generating charts to visualize the dataset.
Python Code: Python Script: A Python script was used to enhance the statistical analysis of the
dataset. This script performed advanced operations, such as calculating correlations and
generating heatmaps to visualize relationships between variables
References:
Dataset: Self-collected data, accessible via this
[Link]
9g/edit?usp=sharing
Transfer Market: [Link]
league/topErhalteneElfmeter/wettbewerb/GB1
18