This is for you — JxmyHighroller
Introduction
Despite my best efforts and careful analysis, my predictions failed to capture the true essence of the 2023 NBA Playoffs.
- What went wrong?
- What did I miss?
- Why is Erik Spoelstra only a two-time NBA Champion?
Yet in the face of defeat, I feel driven to analyze and understand the nuances that have eluded my comprehension.
With a critical eye and a determination to improve, I set out to recalibrate my approach to predicting the 2024 NBA Champion. Learning from past mistakes and embracing the challenge of uncertainty, I am introducing two important changes that promise to improve the accuracy and depth of my prediction.
Lesson learned - 1/2
The most remarkable story of the 2023 NBA Playoffs is the Miami Heat’s meteoric rise, orchestrated by the strategic brilliance of Coach Erik Spoelstra. Against formidable opponents and daunting odds, they defied expectations, showcasing resilience, teamwork, and a relentless pursuit of excellence — And single-handedly destroyed my 2 Top Contender.
It is obvious to me now, that the coach plays a crucial role in a team’s success in the post-season. However, quantifying the coach’s impact is a challenge that goes beyond simply counting championship wins. Therefore, I have developed 3 metrics to evaluate coaching greatness/performance.
A brief breakdown of these metrics can be found directly below the chart.
Don’t miss the opportunity to reveal the identity of…
The NBA GOAT (Playoff-Coach Version) 🏆🤫
The Greatest NBA Playoff Coach of All Time
NBA-C/A:
Job finished — Number of Championships divided by attempts.
PS = Playoff Score:
The Playoff-Score is the sum of all post-season results of a coach. To achieve a positive value, the coached team must finish better in the playoffs than expected after the regular season.
-> Unlock the full potential of your team, consistently year after year
PS/A = Playoff-Score/Attempt:
Simply the median of the Playoff-Score per playoff run.
Upon examining the chart, it becomes apparent that the Top 5 Greatest Playoff Coaches of 2023 outshine Budenholzer and Mazzulla by a significant margin.
Lesson learned - 2/2
Reflecting on last year’s forecasts, it’s evident that Playoff unpredictability can disrupt even advanced models. To enhance accuracy, I’m pivoting to predicting Division Winners rather than the NBA Champion directly. This adjustment follows last year’s scenario where three Eastern teams ranked in the Top 3, yet only one could proceed to the finals.
By analyzing the team dynamics within the conference, the goal is to refine the forecasts despite the increased complexity.
Predicting the 2024 NBA Division Champions:
Following extensive hours of hard work and dedication, I excitedly tease the NBA Champion 2024 prediction, only to encounter a disappointing lack of results. Despite my relentless efforts, there are no noticeable improvements. I just have the impression that the NBA is becoming more and more unpredictable.
This is why I don’t want to make a clear prediction for a Champion this year.
But wait,
before I dive into why I didn’t see any improvement after the adjustments, I want to steer your attention to something even more captivating — a truly exceptional video & article.
Over a year ago, I had an insightful conversation with JakeAllenData about this very topic. Since then, his fervent devotion to predicting the NBA Champion based on numbers has evolved into what I consider the Best Forecasting Technique.
Understanding the Absence of Progress
Coaching Stats are useless for this Prediction
Don’t get me wrong, a good coach has some correlation with long playoff runs, but it’s impossible/very hard for a ML model to properly interpret that in the context of predicting just the NBA Champion.
Allow me to illustrate this with an example:
Erik Spoelstra is considered by many to be the best playoff coach in the league, and this is also reflected by the Playoff-Score (see chart).
Logically, the model should favor high values for the coach statistics. However, this is not the case, as there have been many coaches in the past who were able to win the Larry O’Brien Championship Trophy directly after their first or second attempt. In practice, this means that assigning a high or low number for the Playoff-Score to the ML model does not have a significant impact on its predictions, as this metric does not provide any clear added value to the champion’s prediction.
Division Prediction is impossible
Initially, it seemed like the logical choice, but after countless hours of painstaking work, I can confidently attest that making a reliable and consistent determination of Division Winners is incredibly challenging. The outcomes and models often resemble blind guesswork rather than objective number crunching.
To emphasize the unpredictability, picture writing the names of the top three teams of one division on three identical balls. Place them in a spinning drum and randomly pick one ball to determine the Division Winner. 😅
Conclusion
Given that my previous ML model aligns with Jake’s predictions, both favoring the Denver Nuggets and Boston Celtics as primary contenders, I’d like to offer my own perspective.
In my view, the Western Conference appears notably stronger this year. Consequently, every game is likely to be tightly contested, which could/will lead to increased fatigue and injury risk.
Therefore, I predict the 2024 NBA Champion will be the Boston Celtics, securing the title in 6 games.
Feedback & Questions
If you have any feedback, please check out my email address on my Website or connect with LinkedIn. If you want to study the really old code in detail, check out the GitHub Repository.
