NCAAB Efficiency System
Everything we offer for college basketball starts with efficiency. Efficiency is a metric that measures how strong each team is per possession. Offensive efficiency is the number of points we project a team to score per 100 possessions against an average opponent. And defensive efficiency is the number of points we project a team to give up per 100 possessions against an average opponent. These efficiency metrics start with a preseason projection (detailed below) and then iterate throughout the season based on performance. We use them to fuel our power ratings as well as pair them with a tempo (number of possessions we expect a team to have in a given game against an average opponent) projection to get our gameday score predictions. However, it is important to remember that these score predictions assume general playing conditions. For example, it may not factor in if the game is being played on an Aircraft Carrier like Gonzaga and Michigan St. did on Veteran’s Day. It is important to do additional research on a game on top of the projection if you are using them to place bets.
The key to strong in-season efficiency numbers is a solid pre-season projection for each team. Preseason projections are so strong that there have been studies where these numbers outperform end-of-season metrics when attempting to predict what happens in March Madness. It turns out that our minds not only become biased during the season but in some cases, the numbers can also become biased. There will never be enough games in a given season to build a large enough sample size so that you can be 100% confident in your numbers. So if you weigh the small sample size too much in your model, it can bias data to those specific games. That is why we include some value of the preseason projection for our efficiency and tempo numbers throughout the season. Here is where we start…
Offensive efficiency involves the most extensive process for our projections because several factors come into play. The first is program rating. We find a program rating for each school and use that as a starting point for projection. This is done by taking the offensive efficiencies from the past few years and weighing each year to get an average. The most recent season is weighed the most and the weight diminishes the father back we go. From there, we take a closer look at the team from last year. Who did the team rely on? What’s new about the team (players in and players out)? How much different will returning players perform? Those first two factors are pretty easy to pull and understand, but projecting how players will perform is the hard part. There is a multitude of resources that give insight into how to project player development, but the factors that we focus on are usage, player ages, recruiting profile, and individual efficiency. For example let’s look at a player who was a freshman last year, a 5-star recruit out of high school, and performed at a particular clip and usage. Rather than enter the draft he returns in 2022 for his sophomore year. From freshman to sophomore year we expect him to make a leap in production. Additionally, since he was a 5-star recruit we also expect him to develop and hone his skills in a specific way. However, someone that he split minutes last year has now graduated, so his expected usage will rise. Generally speaking, a player's efficiency goes down as usage goes up. We'll perform this type of analysis for each player which when combined with a school’s program rating, culminates into a rating for each team.
Defensive efficiency starts the same way as offensive efficiency–with a program rating. We take the last few years of data and weigh them into a number that acts as our starting point. From there we take a look at % of returning minutes as well as individual recruits/transfers for each team—specifically 4- and 5-star players. Last year’s defensive player of the year, Walker Kessler of Auburn, was a transfer out of North Carolina. He, along with top recruit Jabari Smith, was able to lead Auburn to their best defensive finish since Bruce Pearl started coaching there in 2015. On top of being impact talents, both were 5-star recruits out of high school, something that both of these players have in common is their height. Height is linked to a team’s defensive ability due to its impact on shot blocking/rebounding and court coverage. To evaluate height's impact on defensive ratings, we combine and analyze the number of minutes each player is projected to play and the height of each of those players. For example, a tall player who plays a lot of minutes will have a large positive impact on the team. These individual numbers are combined for each team and used to alter their defensive ratings. Lastly, we evaluate the coach and if a change has been made in the offseason.
Tempo involves the most straightforward process for projecting the next season as a lot of it is rooted in previous program numbers. Most coaches have particular playstyles that lead to unique possession lengths on offense and defense. For our preseason tempo projections, similar to our offensive and defensive efficiency numbers, we take a weighted average of the last 3 years to get a program number. We then consider if there has been a coaching change. With a coaching change, we take the time to analyze that coach’s style and put it into a tempo projection. For example, let's look at a team that went through a coaching change last year: Oklahoma. Lon Kruger coached OU since 2012. Over his 10 years at Oklahoma, the average tempo for the team was 69.7. If we were to project Oklahoma for 2022 under Lon Kruger, looking at the last 3 seasons of 68.8, 69.7, and 67.1, we would likely settle around 69. However, Oklahoma went through a coaching change to Porter Moser. Moser had spent the last 10 years at Loyola Chicago. Over those 10 years, Moser deployed a very specific system leading to one of the slowest-paced teams in the nation. On average over those 10 years, Loyola Chicago played to a tempo of 64.04. So where do we go from here? It is important to blend the two numbers, but the coach’s tempo will be weighed much more than the school’s. Therefore, our projection for tempo would come out to around 64 as opposed to 69.