Notes & Details
A few notes and details for more context.
Seems Kinda Biased...​
​
I understand that at first glance, it may seem biased to create my own variables to use for this experiment. However, when creating these variables, I made sure that each variable accurately represented what they were intended to measure. Additionally, the variables I created are all based on existing measures that are tracked by the NBA. So I can't simply give a team a subjective competitive rating. The stats from all teams are put into a single standardized formula I created (for the exact formulas, see here). This means that each team will have the same advantages and disadvantages as other teams with similar stats.
​
Team Score is meant to represent the degree to which a team is successful during the regular season. A team cannot necessarily have a "perfect" Team Score considering the factors that make up Team Score.
​
Team Score is made up of a team's win %, net rating, and my measure of performance score (which is made up of a combination of existing metrics based on the Four Factors by Dean Oliver). These are all different metrics that are consistently used to measure degrees of winning and success for a team.
​
I designed Team Score based on how video games typically allocate points to a player's score. You start with a Team's Performance Score (times 10). This serves as a baseline for a given team. Then, you just factor in a team's net rating (times 10) and win % (times 100 since it's initially expressed as a decimal). A lower win percentage of course means less points added onto the baseline and net rating can either be positive or negative, therefore teams can actually lose points from a negative net rating.
​
Example: 2020 Milwaukee Bucks
The 2020 Bucks have a Performance Score of 29.204. Multiply it by 10 to get a baseline of 292.04. Then take the Win % and Net Rating and multiply them by 100 and 10 respectively to get 0.767 * 100 = 76.7 and 9.6 * 10 = 96. Lastly, add these to the baseline to get the Team Score which would be 464.74 (the numbers in Excel are more precise).
​
Competitive Score is a bit simpler in comparison and only applies to the postseason. You take the number of wins a given team has multiplied by the Team Score of their opponent (for each round). Then, you just divide that by the number of total games played. Here, Team Score is used as a way of measuring an opponents strength. Teams that have higher Competitive Scores tend to have strong matchups, more wins, and less games played.
​
Example: 2020 Milwaukee Bucks
The 2020 Bucks won 4 games against the Magic (329.43 TS) and 1 game against the Heat (374.88 TS) with them playing a total of 10 games. This results in a Competitive Score of 169.26.
​
NOTE: Teams that either do not qualify for the playoffs or don't win any games in the playoffs receive a Competitive Score of 0.
What About Injuries? Trades? Etc.​
​
So one notable flaw that people may have with my project is that it doesn’t necessarily take into account (star) players that may have contributed a lot in the regular season but didn’t play in the playoffs due to injuries or other reasons. However, I’d counter that
-
The data will reflect “discrepancies” like this if a particular team has a high team score but a low competitive score. So any significant results where this is the case will likely stand out anyways.
-
The data/project in its entirety is meant to reflect the performance of teams as a whole. This means that if a team solely revolves around a given player (or group of players), and that player does not play (much or at all) in the playoffs, then accordingly, that will be reflected in the team’s competitive score.
It also doesn’t take into account trades that may have occurred where newly acquired players don’t play much during the regular season.
I guess a blanket statement would be that if any really good, all-star caliber players do not play (much or at all) during the regular season, then it may appear somewhat skewed at first. However, as intended, the Team Score is meant to measure a team’s strength and capability based on their performance…NOT POTENTIAL. This project eliminates (most) of the wild speculation that often takes place throughout the NBA every year.
Additionally, with the way that the playoffs are structured, if a star player does not make consistent contributions to their team, the team will have a lower win percentage, meaning more difficult matchups in the postseason, which the Competitive Score does in fact account for (this seems like common-knowledge but you’d be surprised how few people understand that a big name, star player does not suddenly/automatically make an entire team competitive if they weren’t before). So while it’s possible that a star player barely contributes during the regular season (resulting in a lower Team Score) but balls out during the playoffs to the point where their team gets wins, then yes, this would create a discrepancy between the evaluated Team Score and Competitive Score.
Please note, that this scenario is extremely unlikely as teams that find themselves without star players that consistently contribute and perform during the regular season either won’t make the playoffs, won’t have chemistry with the team, will have a much harder time in the playoffs if they do qualify, etc. There is no player that I’ve seen in all my (albeit short) time watching and analyzing the NBA (almost 20 years) where they can lift up a truly garbage team to the point of extremely high-level playoff contention. Lol.
Late rebuilds I of course can’t account for. If a team completely revamps their team mid-season, and they go from a well-below .500 team in all games before the rebuild but end up well-above .500 in all games after a rebuild/trade deadline, then this can create a discrepancy as well. However, since playoff structure is linked to how team’s do in the regular season, I’d say that a team will still have to play fairly well and therefore will still need to earn a decent Team Score before being playoff contenders.
This experiment also doesn’t account for injuries that occur during the playoffs.
With all these possible discrepancies though, it will be interesting for others that are even more basketball savvy than myself to review the data and find causations that I myself will most likely overlook or simply be unaware of. In my opinion, that makes this project a lot more inclusive and gives it a larger sense of purpose and scale than just myself and my limited knowledge.
​
Phew! Thanks for reading all that.