Using Advanced analytics to cut through the smoke
The regular season is over. All that’s left is the conference tournament and then any post season tournaments Mountain West teams get into. It has been a wild season with a ot of parity throughout the league. The talent this year is arguably better than it’s been top to bottom in a long time.
The past few seasons I have looked at who were the best players in the conference, using a combination of advanced analytics. This entry marks the second installment for the 2021-22 season.
Three different advanced analytics will be used. The analytics are: Player Impact Plus Minus, Points Over Expectation, and Bayesian Performance Rating.
These 3 contributions put together should give us a good idea of who’s performed the best this season, as they measure different things, such as impact vs. efficiency. They all also measure contributions on the offensive and defensive sides of the floor, enabling them to paint a full picture.
2022 Mountain West Tournament Bracket, TV Schedule Announced
Player Impact Plus Minus – Also known as PIPM, this is an impact stat that primarily takes it’s measurements from box score stats. Basically, it measures how well a player has performed in the role they’re in. A player being used in the way that best suits their skill set will have a higher score than a player who is talented in certain areas but not able to show that talent off. As an example, if Hunter Maldonado was asked to shoot 3 pointers all game he would hurt his team, as that’s not his skill set. This statistic is important because no matter how purely talented a player may be, if the player doesn’t use the talents correctly it will hurt the team and prevent winning. PIPM also makes adjustments for the quality of opponents. For more on PIPM click here.
Points Over Expectation – Also known as POE, this is an efficiency stat. It takes into account the number and type of shots a player takes (or defends) and compares the outcome to what an average player would’ve done with the same number and type of shots. A score of zero is the equivalent of an average player. Since POE takes into account the number of shots, than the higher usage a player has, the more likely they are to be farther from 0. So players that are really efficient on large volume are the ones that get good scores here. It is also a per game stat, as opposed to a per 100 possession stat. Since basketball is about scoring more points than your opponent, someone who can score, and defend, at an efficient level is a valuable player. For more on POE click here.
Bayesian Performance Rating: Bayesian Performance Rating, or BPR, attempts to qualify the value a player gives their team while on the court primarily by measuring offensive and defensive ratings while a player is on the floor. It is an impact stat in the vein of PIPM, but uses different inputs to estimate the impact a player has. Similar to PIPM, it makes adjustments for the quality of teammates as well as opponents in it’s calculations, so that fans can better determine who is good vs. who plays with good teammates. A score of 0 is considered average. To learn more about BPR click here.
Combining the different methodologies of who helps their team when they’re on the floor, who looks good in the box score, and who is efficient should give us a pretty good feel for who has perfromed well this season, as these metrics will help cover up each other’s weaknesses.
Simply taking the average of these numbers won’t work though, as they measure different things. So Z-scores will be used. Basically, Z-scores measure how far away something is from average, with zero considered to be average. Once the Z-scores for all three metrics are calculated, the average of those numbers will be taken to determine who has been the best so far.
To give you a feel for Z-scores, last year, using the same methodology, Jordan Schakel led the league with a score of 2.777, and Neemias Queta came in second with a score of 2.694. Bryce Hamilton was considered pretty average with a Z-Score of 0.074.
With the boring explanation out of the way, lets look at the results.