4.

The Application for Moneyball in Football (Soccer)

a. Case Study: Brentford FC (English Premier League)
In 2021, Brentford FC made it to the Premier League after a new challenge that was their
competitor clubs, which had significantly larger financial resources. However, this was where
Brentford took the "Moneyball" philosophy by considering advanced data analytics and machine
learning in finding undervalued players. (Merchant, 2024)
Large sets of data in performance by leagues worldwide are used through machine
learning algorithms by the club. One of the crucial aspects of these analytics understood by
Hewitt and Karakuş (2023) were: passing accuracy distance covered expected goals, and
defensive actions, amongst other things; hence, determining if a player could fit in at Brentford,
tactically speaking. They used different aspects of this technique for instance:

1. Using historical data, machine learning models predict future performance, helping
Brentford identify talent from lower leagues that larger clubs might overlook.
For instance, regression models estimate how a player's contributions in lower-tier
leagues (e.g., the Danish Superliga or French Ligue could translate into Premier League
success.

2. One of the important techniques is the use of clustering algorithms, which group players
by similar playing styles and metrics. Reinforcement learning models optimize transfer
decisions based on player performance data, contract status, and market dynamics to
maximize value during transfer windows.
A good example of Brentford's data-driven recruitment strategy is the signing of striker
Ivan Toney from Peterborough United in 2020 for about £5 million. Toney scored 24
goals in League One, and Brentford took notice. Although the transfer fee was relatively
low, Toney became an important player in Brentford's promotional campaign. In 2024,
the club sold Toney to Al Ahli for around £40–45 million, showing the return on
investment, their approach generates.

3. Lastly their data-driven approach to goalkeeping needs. Machine learning models will
judge goalkeepers such as Illan Measlier and Sam Johnstone on their shot-stopping,
saving percentages, and involvement in attacking phases: final-third passes.
Johnstone has 26 passes into the final third, compared to Mesler’s. This might make him
the more suitable contender, especially if Brentford seeks a Premier League-proven
goalkeeper. Considering different metrics and market conditions, Brentford would
probably choose Johnstone due to his performance metrics and the lower cost.

Wigmore (n.d.) identifies Brentford's success in not only spotting underpriced talent but also
in making a consistent profit from player sales. For instance, Brentford signed Saïd Bernama and
Ollie Watkins at relatively low fees and sold them for £27 million and £26 million, respectively.
In this way, the club ensures its growth as it reinvests the profit into new talent.

b. Case Study: FC Midtjylland (Denmark)

“In an industry where the gap between the rich and the poor is growing apart to a ridiculous
extent, outsmarting the big spender has never been more crucial.” (Ritmeester, 2018, p. 2) Thus
giving more weightage to making financially viable decisions in sports also becomes a necessity.
FC Midtjylland is the benchmark of how data analysis has revolutionized football, especially for
smaller clubs competing with financially superior opponents.

Exhibit 1: Player Performance vs Salary

This method has proven highly successful in giving Midtjylland a competitive

edge against larger clubs, fulfilling Ankersen's goal to "outthink" rather than outspend
their rivals. The method included:

1. Ranking Data: As per Wolf (2011) the club utilizes an innovative model that ranks all
clubs in Europe as though they compete in a unified league, cross-referencing results
from various leagues to assess a team's true standing.
For instance, although Grether Fürth is a German team playing in the second division, its
performances against teams such as HSV (Hamburg) can be statistically correlated to top-
level European teams like Bayern Munich and Manchester United. This model reveals
the real value of players in lesser-known leagues, thereby stripping away bias associated
with the more prominent leagues like the Premier League or La Liga.

2. Excepted Goals: A key statistical tool used by FC Midtjylland is the expected goals (xG)
model, which is considered more accurate than traditional metrics such as goals scored.
 Ankersen argues that goals alone are a poor reflection of a team’s quality due to the
randomness and luck involved in matches.
The xG model strips away this randomness by calculating the probability that a specific
shot results in a goal. This metric allows analysts to determine the expected goals scored
and conceded, providing a net xG score that offers a better indication of a team’s quality
and future performance. This also helps in identifying undervalued players whose
contributions may not be immediately obvious based on traditional statistics.

3. Data Collection Model: Midtjylland’s success relies on continuous access to extensive

player and team performance data, which is collected by major data providers such as
Opta, Wyscout, and Squawka. These companies track key metrics related to player
development, performance, and team strategies, selling this data to football clubs like
Midtjylland. By using this data, Midtjylland has created its own proprietary statistical
models to assess player effectiveness and recruitment potential.

However, while the current models used by Midtjylland and other data-driven
clubs have proven valuable, they still have limitations. Existing models can assess how
good a player is today and predict future performance, but they fall short in determining
the player’s overall contribution to team success. Additionally, no model yet exists to
measure how specific attributes per position can affect the team’s overall strategy.

5.AI and Data Analytics in Identifying Undervalued Basketball Players

The use of artificial intelligence (AI) and data analytics in basketball has changed the
way players are evaluated, scouted, and developed strategies. Through the examination of huge
datasets that reflect player efficiency, biometrics, and game patterns, AI allows teams to make
better recruitment and development choices.
Digital Defynd (2024) identified one of the most pivotal applications of AI in basketball
has been performance grading, wherein intricate metrics evaluate an individual's shot accuracy,
defense, and efficiency. Through a player's in-game decision-making, movement, and flexibility,
AI enables teams to spot talent that goes unseen by traditional scouting. By comparing players
between leagues and past data, AI can make future performance predictions based on playing
style, fitness, and statistical consistency.
The San Antonio Spurs are one such team that has been able to use data-based scouting to
draft and develop players who were underappreciated at first but turned out to be valuable
players. Moreover, real-time game analysis and injury prevention have become major areas
where AI promotes player longevity and performance. Wearable sensors and AI-based health
monitoring systems monitor player fatigue, workload, and risk of injury, enabling coaching staff
to make data-driven decisions regarding playing time and conditioning. By reducing risks of
injury
a. Case Study: Houston Rockets (Moreyball)
Daryl Morey, Houston Rockets General Manager, transformed basketball strategy by
utilizing analytics to optimize team efficiency. His statistical modeling-based approach
determined inefficiencies in conventional gameplay and capitalized on these to become
competitive. Five outstanding features studied by Sarnoff (2018), of Morey's analytical
philosophy that has transformed the Rockets' playing:

1. Three-Point Optimization and Shot Selection

Morey's most basic observation was valuing three-point attempts over mid-range two-
pointers. His examination found that, as harder shots to make, three-pointers gave 50%
greater scoring possibility than typical two-point shots. It resulted in the Rockets
achieving a record number of three-pointers made per season. Furthermore, Morey
determined that "corner threes" were especially desirable because they were closer to the
basket and therefore created specialized plays to generate open shots for them.

Exhibit 2: Shot efficiency (most common shot locations)

Exhibit 3: League-wide three-point shots attempted over time

 2. Elimination of Low-Value Shots
One key aspect of Morey's plan was the removal of inefficient mid-range jump attempts,
which, statistically speaking, yield fewer points per attempt. He pushed attempts instead
either from outside the three-point line or from inside the paint area around the basket,
where layups and dunks provide a better chance of scoring. Consequently, 82% of the
Rockets' overall shot attempts were made in these two locations—well ahead of the
second-most efficient team in the league at 71%.

3. Data-Driven Roster Building

Morey's approach went beyond shot choice to how he constructed the team. He looked
for players who matched the analytical model instead of using conventional scouting.
This involved bringing in players with good three-point shooting skills or players who
could thrive in finishing close to the rim. Without the appropriate personnel, the approach
would not work, which underscores the significance of data-driven recruitment in player
signings and team construction.

4. Impact on the NBA and Risks Involved

Morey's analytical methodology affected the entire NBA, triggering a league-wide surge
in three-point attempts. The strategy does come with drawbacks, though. Teams without
solid three-point shooters might not be able to apply this strategy to success. More
importantly, detractors say the Rockets' penchant for repetitive shot selection diminishes
the entertainment factor of the game. This prompted the NBA to consider changing the
rules of the game, including altering the location of the three-point line, to even out
gameplay.

Daryl Morey's methodology is a groundbreaking change in the art of basketball

strategy, using statistical know-how to maximize team performance. His data-driven
decision-making not only revolutionized the Rockets but has also impacted the growth of
sport in general. The question now is how teams will keep reinventing while ensuring
efficiency without losing the entertainment value of the game.

b. Case Study: Toronto Raptors (2019 NBA Champions)

Another example in the NBA is how “Fred VanVleet went from undrafted player to
record breaker” Amirazodi (2021) They used data analytics to acquire and develop overlooked
players (e.g., Fred VanVleet, undrafted but became a key player). Raptors bet on VanVleet to
play 4 seasons with average score rates of “10.2 points, 4.5 assists, and 102 starts”.
The numbers for basketball and college level are very low but the stark difference by the
end of his college was a reason for the best bet for Raptors: “VanVleet was ranked first all-time
in school history for assists (637), steals (225) and assists-to-turnover ratio (3.08).”
 Leveraging AI and data analytics in a similar fashion as mentioned previously, allowed
Raptors to hire a passed “undrafted player” who turned out to be a record breaker based on
statistics.

Exhibit 4: Indicators for Injuries

Raptor also attempted to utilize the models for injury prevention that were developed by
Wu (2020). To decrease the uncertainty caused by injuries they identified players performance as
the most vital component in prediction. A lot of factors were excluded from research that were
relative to injuries like weight, height, and intensive routines.

Exhibit 5: Method of Data collection

Reference

Amirazodi, N. (2021, February 6). Reminder: The Toronto Raptors’ Fred VanVleet went from

undrafted to record breaker. Raptors HQ.

https://www.raptorshq.com/2021/2/6/22269218/reminder-fred-vanvleet-undrafted-player-

to-record-breaker

DigitalDefynd, T. (2024, April 18). 10 Ways AI Is Being Used in Basketball & NBA [2025] -

DigitalDefynd. DigitalDefynd. https://digitaldefynd.com/IQ/ai-in-basketball-nba/

Hewitt, J. H., & Karakuş, O. (2023). A machine learning approach for player and position-

adjusted expected goals in football (soccer). Franklin Open, 4, 100034.

https://doi.org/10.1016/j.fraope.2023.100034

Merchant, A. (2024, November 15). Data Over Dollars: How Brentford FC Took on Premier

League Giants with Algorithms. Medium.

https://medium.com/@abbasmerchant60/moneyball-in-football-brentfords-data-driven-

success-in-the-premier-league-66f5c72d8f01

Ritmeester, C. (2018). An elaborate approach to game-winning strategies and player ratings in

football [M.Sc. Thesis, University of Twente AE Enschede the Netherlands].

https://essay.utwente.nl/76361/1/Ritmeester_MA_BMS.pdf

Sarnoff, Z. (2018). Moneyball: The Houston Rockets and Analytics. Harvard Business School,

21–34. https://d3.harvard.edu/platform-digit/submission/moreyball-the-houston-rockets-

and-analytics/

Wigmore, T. (n.d.). Brentford’s Moneyball way to beat football teams with huge budgets.

https://bleacherreport.com/articles/2718752-brentfords-moneyball-way-to-beat-football-

teams-with-huge-budgets
Wolf, D. (2011). Reading The Game Anglo-American perspectives on football fandom in the age

of Premier League football [PhD Dissertation, De Montfort University Leicester].

https://bonndoc.ulb.uni-bonn.de/xmlui/bitstream/handle/20.500.11811/5987/3698.pdf

Wu, W. (2020). Injury analysis based on machine learning in NBA data. Journal of Data

Analysis and Information Processing, 08(04), 295–308.

https://doi.org/10.4236/jdaip.2020.84017