Lineup Data Just Got A Whole Lot More Useful

Introducing a game-changing new lineup projections, grades, and optimizer tool at EvanMiya.com

Every team is looking for the perfect lineup: five players on the court with an ideal balance of skills, playing off each other with unthinkable synergy, carving up defenses while having few defensive holes.

Finding effective lineups is more complicated than just throwing your five most talented players on the court together. There should be positional balance, offensive skill sets that complement each other, defensive understanding, and more. One of the most common methods for optimizing rotations is analyzing five-man lineup data, like what is already found at EvanMiya.com. A team’s per-possession efficiency can be grouped based on unique lineups used. Based on the possessions played by a specific set of five players, we can see how good a lineup has performed offensively and defensively, even adjusting for the strength of opposing players faced by that lineup:

Duke’s most used lineups in 2024-25, sorted by Adjusted Team Efficiency Margin

There has always been one massive flaw with five-man lineup data: lack of sample size. In 2023-24, the average starting lineup in Division 1 played 320 possessions together across the season, equivalent to about 4.5 full games at a typical tempo. While this is a large enough sample size to determine if a starting lineup works, it gets way more dicey for lesser-used lineups. If you order every team’s lineups based on utilization, a team’s 10th most used lineup only plays around 45 total possessions in a season, or about 26 minutes of game-time. Now, consider that most teams feature between 65 and 90 unique lineups in a season. Analyzing lineup data for those lesser-used groups is essentially worthless due to a lack of sample size.

One way around this is by analyzing how smaller sets of players perform together. 2-man, 3-man, and 4-man lineup combination data (also found at EvanMiya.com) helps alleviate some of these sample size concerns, but it still fails to answer all our questions.

Lineup Projections

The solution to this problem comes in the form of a groundbreaking new feature at EvanMiya.com: A lineup projections, grades, and optimizer tool.

Instead of only using the observed on-court performance of a lineup to judge its viability, we instead calculate forward-facing projections of how efficient a particular lineup will be using several factors:

1. Individual Player Skills

   Even if a group of five has never shared the court together, we can get a good sense of how effective the lineup will be based on the strengths and weaknesses of the individual players. At EvanMiya.com, we can characterize the skills in a lineup using the player projections available for each player, which grades individual skills like shooting efficiency, rebounding, and playmaking for all college players. With this data available for all five players, we can predict how efficient a lineup will be in all of these categories, including an estimate of the exact values for all of these lineup skills and grades compared to all other D1 lineups.

2. Observed Lineup Performance

   The on-court success of a lineup is still really crucial! If a particular lineup plays a lot of minutes, that group’s actual offensive and defensive efficiency becomes a much more critical part of the final lineup projection. However, suppose a lineup has barely played together (or not at all). In that case, we can still form a reliable estimation of how good a lineup will be based on the individual player profiles.

3. Position and Role Balance

   A lineup should be appropriately balanced across the five positions. No matter the individual skill levels of players present, if the player types don’t complement each other, there will be significant issues, especially defensively. For example, Hunter Dickinson is a great player, but if you had five Hunter Dickinson clones together on the floor, the lineup would struggle in multiple departments. Every lineup projection includes a check to ensure that the positional and offensive role balance in that lineup falls within the range of reasonable lineups that succeed in college. If a selected lineup shows balance issues that are way outside the norm, it will get flagged, and the predicted efficiency will drop, sometimes to a large degree. Examples of an unbalanced group would include having a lineup with three centers, too many ball-dominant players, no players who can create offense for others, or a lineup of players all 6’4” or shorter.

Here is the complete list of features used to calculate a lineup’s projected efficiency:

• Observed lineup performance:

  • On-court offensive and defensive efficiency, adjusted for opponent strength

  • Possessions played

• Individual player skills:

  • Offensive and Defensive Bayesian Performance Rating, which predicts per-possession value on both ends of the court

  • Projected 2PT shooting

  • Projected 3PT shooting

  • Projected FT shooting

  • Projected scoring volume

  • Projected shot frequency rates (such as three-point attempts per possession)

  • Projected offensive usage

  • Projected offensive and defensive rebounding

  • Projected assist rate

  • Projected turnover percentage

  • Projected steal percentage

  • Projected block percentage

  • Projected foul percentage

• Lineup balance:

  • Player heights

  • Estimated player positions

  • Estimated offensive roles (does a player create offense, or is he a secondary player?)

The Lineup Projections tool uses three factors: individual player skills, on-court lineup performance, and position and role balance adjustments.

Demo

All new lineup features are in the Lineup Metrics section at EvanMiya.com and are available to premium subscribers.

Projections / Grades Page:

Let’s start with the Projections / Grades page, which sorts every lineup in D1 by Predicted Team Efficiency Margin and displays strengths and weaknesses for each lineup in percentile or letter grade form. Here are Baylor’s top predicted lineups, with lineup skill grades displayed as “Percentiles” using the toggle in the top left:

You can sort by any skill category to see the best-predicted lineups in that area. Here are BYU’s lineups sorted by projected three-point shooting, displayed as letter grades:

Because you can view more than one team at a time, you can find the best lineups in the country or a specific conference based on available lineup skills. Here are the best projected defensive lineups in the country right now, with the actual projected efficiency values displayed:

Lineup Analysis Page:

While the Projections / Grades page gives a high-level overview of multiple lineups at a time, the Lineup Analysis page drills down to a single lineup to show projections in greater detail. You can click on any lineup name from the other pages to go directly to a lineup’s analysis page, or you can select the individual players yourself:

And this is where it gets REALLY fun: You can choose any hypothetical lineup of five players in the country to see how good it would be, whether the players are on the same team or come from different teams. For example, here’s how a lineup of Zakai Zeigler, Kam Jones, Dillon Mitchell, Cooper Flagg, and Johni Broome would grade:

If you submit a lineup that has characteristics very outside the norm, such as a lineup of all point guards, not only will the predicted lineup efficiency get docked, but the extreme concerns will be listed at the bottom of the page. Here’s how a lineup of elite point guards would fare:

Offseason Roster Building

These lineup projections are vital for in-season lineup optimization, but the tool will be equally valuable for roster building during the late spring and summer. Once the calendar flips to around February, you can view hypothetical lineup projections for the 2025-26 season. Not only can you select players from different teams, but you will also be able to include incoming freshman recruits in the lineup analysis tool.

So How Much Does It Cost?

These tools, along with individual player projections, are available for Premium subscribers at EvanMiya.com for just $29.99 a month or $179.99 per year. I want these features to be as widely accessible as possible: The premium features to be a steal-of-a-deal for any athletic department, and diehard CBB fans should also be able to dig in without it costing a fortune.

It’s a huge privilege to build college basketball models as my full-time job. By being a paid subscriber at EvanMiya.com, you are getting some fantastic CBB data tools at your fingertips and supporting my work, allowing me to invest more in the CBB community.