How Analytics Can Help Coaches Develop the Most Optimal Championship-Meet Lineups

coach-write-plan-practice-splits-times Lineup

How Analytics Can Help Coaches Develop the Most Optimal Lineups

By Michael Kidd & Kate Burk

In this study submitted by Michael Kidd and Kate Burk, the authors analyze putting together an optimal lineup for a championship competition.


Coaches have long understood that the team advantage painstakingly developed by training, technique, and strategy over a swimming season can evaporate should the coach make sub-optimal lineups for the championship meet. At the same time, the number of swimmers and available events can quickly create tens of thousands of options for coaches to choose between when developing their entries. Applicable research into optimization mathematics has been the providence of computer scientists and statisticians and outside the comfort level of most high school and club coaches. Though meet optimization has not advanced to the smartphone application, common spreadsheets such as Microsoft Excel and Google Sheets now have the capabilities for coaches to build and test their optimization models.


Using Microsoft Excel, we set up a binary evolutionary optimization model. The model provides the ability to estimate individual and team points under multiple situations. More importantly, based on many external scenarios, we can use the model to create lineups in all but the “swimmer pick” and “top 3 events” scenarios. We used different sets of information, or increasing levels of knowledge, to create lineups and then quantify the improvement of each lineup against the actual lineup for a six-team championship swim meet. More specifically, after we created a lineup with a given amount of knowledge, we transferred our team’s entries to the model containing the actual entries for each other team. We could then calculate the meet’s estimated results and record the predicted points of every team.

Evolutionary optimization is a powerful algorithm inspired by the process of natural selection. In simple terms, it mimics the process of evolution by iteratively improving a set of potential solutions to a problem. In Microsoft Excel, you can implement evolutionary optimization using the Solver add-in, which allows you to find the optimal values for a set of decision variables.

Although this is not a tutorial article, a basic approach to developing an Excel-based model includes:

  1. Define Decision Variables: Identify the variables that influence your problem. In the context of swim lineup planning, these could be the selection of swimmers for specific events.
  2. Set Objective Function: Define an objective function that represents the goal you want to achieve. In swim coaching, this is maximizing the total points scored by the team.
  3. Constraints: Establish any constraints or limitations. Constraints may include limits on the number of swimmers available to score in an event or the number of events in which they can participate.
  4. Evolutionary Solver: Access the Solver add-in in Excel and choose the evolutionary solving method. The algorithm will generate and evaluate various combinations of decision variables, evolving towards an optimal solution.

Levels of Knowledge:

The levels of knowledge studied included several approaches. Each adds levels of complexity to the model. Though Excel can handle the additional computational complexity, each level requires significant increases in the data collection and analysis required of the coach who is designing the lineup.

  1. Swimmers Pick: Each swimmer selected the events they want to swim, without input from coaches or parents. This was the worst lineup that we tested and was orders of magnitude inferior to any other approach. The overarching problem we observed was that nearly every swimmer favored certain events. Since each team could only score two people per event, most swimmers were not in a position to score points, and many events were left uncovered, with points remaining on the table.
  2. Top 3 Events: This was our baseline lineup. Each swimmer was entered in their three best events based on overall rankings within the league, without regard to what any other swimmer on the team entered. Coach, Isabel Gomez, of the Sigonella (Italy) Swordfish, explained that they chose to seed their entries based on the development of each swimmer, and meet entries were largely an extension of the overall training approach and program. We observed this approach on several smaller teams who did not have the size to compete for the meet win but were looking to best advance individual efforts.
  3. Distributing Swimmers to Avoid Point Cancellation: With a deeper understanding, coaches can strategically distribute swimmers to avoid point cancellation within the team creating a more balanced and effective lineup. Under this approach, we are likely to see swimmers distributed evenly over all of the events, perhaps in some events that they would not consider their best or favored. From a technical approach, we conducted our optimization without any input from possible other teams – we just made sure to look for the best possible distribution of our athletes – essentially assuming that no one else was competing.
  4. Understanding All Possible Opponents: Knowing all possible swimmers who may compete against your team provides a broader perspective. This knowledge allows for more comprehensive lineup adjustments, considering the strengths and weaknesses of potential opponents. While the prior approach assumed that no one else was swimming, this assumes that everyone on the other teams is swimming every event – even knowing that would not be a legal entry approach.
  5. Estimating Opponent Swimmers in Specific Events: As coaches delve further into the opposition, estimating which swimmers will compete in specific events becomes crucial. This knowledge enables precise adjustments, optimizing your team’s chances of success in targeted races. To make this estimation, we conducted evolutionary optimization of the swimmers on each of the other teams, using the results of that approach to feed our optimization. Other teams rely on their experience to predict other teams’ entries. Coach Kate Morgans of the Kaiserslautern (Germany) Kingfish revealed that she spends weeks pouring over seasonal league reports to predict what other teams will enter. The results show better results than the prior model but did not accurately predict each swimmer’s entries. We did not have advance knowledge of which of the possible swimmers the opposing teams were bringing, and in many cases, other teams were not optimizing their lineups for total points. However, as the graph below shows, the estimated knowledge produces superior lineups than if no assumption of the other teams’ intentions were made.
  6. Knowing Opponent Teams’ Entire Lineup: The pinnacle of knowledge involves understanding the entire lineup of opposing teams. This level of insight allows coaches to make highly strategic decisions, maximizing the overall team performance. Running our optimization in this method uses the exact lineup of other teams to create an optimized lineup for your team and doubles the improvement over the baseline when compared to any other approach.

It should come as no surprise that the more knowledge a model had, the better lineup it could produce. Furthermore, coaches can decide what level of improvement they require to judge if the extra time to gather and synthesize more knowledge is justified. We can see that using an optimization approach without any knowledge of the completion produces a 14% better lineup for the team than just putting every swimmer in their best events, and for the next several levels each additional knowledge step increased the lineup quality by about 5%. The biggest jump, however, is going from an estimate of the other team’s entries to the perfect knowledge of the other team’s entries.  This last step produces an increase in lineup quality of 16 percentage points. This begs the question of how to better estimate the other team’s entries. Options include leveraging relationships or clandestine methods to discover the other teams’ intentions. Even a few points of knowledge about what they intend can permit an estimate that is closer to what is actually seeded. By finding out who may not attend, or confirming a few entries, the remaining lineup will be simpler to predict.  Two years ago, we would have recommended using a data scientist to generate a neural network to improve predictions beyond our optimization model. With the growth of artificial intelligence, we predict that coaches could train a model to take into account prior years’ entry patterns, seasonal results, distance from the meet, and other factors to predict an opponent’s intended lineup.


To thrive in a competitive swimming environment, coaches must embrace the principles of decision science. The intricacies of crafting swim meet entries, whether in a dual or multi-team setting, surpass the capabilities of even the most seasoned coaches. The notable enhancements in performance achieved through optimization cannot be underestimated. In any human performance arena, inherent mathematical uncertainties arise, with swimmers experiencing both exceptional and challenging days. These fluctuations pose challenges to any lineup, yet, over time, the unpredictable nature of human performance tends to balance across teams. Ultimately, success favors the team with a superior lineup. Coaches, considering factors such as team size, stated objectives, and the anticipated competitiveness of opponents, have the discretion to determine the extent of opposition information to incorporate into their optimization models. This strategic approach enables coaches to navigate the dynamic landscape of competitive swimming with a tailored and informed decision-making process.

Further Reading

For coaches interested in exploring these approaches, we recommend:

  • Kidd, M. (2023). Swim Analytics; Unleashing your Inner Geek. Naples: Kindle Direct.
  • Mancini, S. (2018). Assignment of swimmers to events in a multi-team meeting for team global performance optimization. Ann Oper Res, 325-337.
  • Nowak, M., Epelman, M., & Pollack, S. (2006). Assignment of Swimmers to Dual Meet Events. Computers & Operations Research. doi:

Author Bios:

Michael and Kate are a coach/swimmer team representing the Naples Tiger Sharks in Naples, Italy. They are interested in combining scientific approaches into club swimming.

Notify of

Welcome to our community. We invite you to join our discussion. Our community guidelines are simple: be respectful and constructive, keep on topic, and support your fellow commenters. Commenting signifies that you agree to our Terms of Use

Inline Feedbacks
View all comments
Would love your thoughts, please comment.x