Over the past few years, my thinking around performance has shifted significantly. I began my career as an Athletic Trainer before moving into Strength & Conditioning and eventually stepping into a Director of Sport Science role to tie these layers together. However, I eventually realized that "together" didn’t mean what I first thought it did. It wasn’t just a matter of stacking roles or collecting more data. Through a combination of books, conversations, and experience, I stopped seeing my job as delivering one-off inputs and started thinking more like an engineer fine-tuning a complex system. It became less about a single intervention and more about an entire orchestra of interactions, where various inputs compound or sometimes conflict with one another.
"In a nonlinear environment, the specific input you deliver is often less important than how the system as a whole internalizes that information to either adapt or fall apart. I wish more Strength and Conditioning coaches understood this. I get why they do, though."
Early in my career, I found a lot of comfort in treating athletes like machines because it made the world feel manageable. This was largely a byproduct of my own background competing in powerlifting and Olympic weightlifting, where training is highly specific and variables are relatively controlled. In those worlds, if you program X amount of volume at Y intensity, the stimulus is predictable, and you generally expect Z result. This view of cause and effect served as a professional safety blanket, giving me a sense of control that worked well in the weight room but began to fracture when applied to the field or the court.
Eventually, the anomalies in team sports became too frequent to ignore, and I had to admit that athletes aren't machines and performance rarely follows clean, linear rules. Inputs don’t always scale as expected, delays are present in every physiological process, and feedback loops shape every adaptation and breakdown. This reality is less like programming a robot and more like conducting a jazz ensemble in real time, where each player responds to the others in a mix of harmony and tension. Despite this, sport is still filled with simplified models because they are easier to sell and control, even if they don’t reflect the world we actually work in. I believe the next evolution of high-performance systems requires models rooted in complexity science—a practical framework for designing systems that can adapt and evolve in the face of uncertainty.
Core Features of Complex Performance Systems
The trap with systems thinking is that it can remain too abstract. To move from theory to actual practice, a performance model should account for five core features of complex systems:
- Distributed Control: No single actor should act as a bottleneck. Coaches, medical staff, and S&C specialists need the autonomy to make decisions within shared guardrails, similar to the leadership concept of Decentralized Command.
- Feedback Loops: The value of data lies in the insights it provides about the environment rather than the numbers themselves. Without these loops, it is impossible to study cause and effect to inform future action.
- Delayed Effects: Because athletes do not adapt on command, systems need a form of "memory" with long enough timelines to observe the real response to a stimulus.
- Multiple Scales: Local actions, such as a single lift or a practice session, must remain coherent within broader goals like seasonal health and availability.
- Adaptability: Since we are guiding an organism rather than optimizing a machine, the model must be capable of evolving as the athlete adapts.
In practice, this means your model needs to be more than a dashboard; it must reflect the actual structure and constraints of the athlete, the environment, and the task. By mapping out subsystems—physical, medical, tactical, and psychological—you can identify their objectives and see where tensions emerge. The goal of a model that represents a complex system is not to exert total control, but to provide infrastructure that is robust enough to guide behavior while remaining loose enough to allow for self-organization.
Turning Systems Thinking into Action
Moving past static metrics requires a shift toward dynamic feedback structures where the model functions as a continuous loop: objective → behavior → feedback → adaptation → updated objective.
In a robust player development system, this process involves creating athlete subsystems that feed data into a centralized controller or hub. For example, during a multi-week acceleration development block, you practice distributed control to expose the athlete to various forces across the sport demands. The S&C coach might monitor force-velocity metrics using electronic resistance, while medical staff track hamstring symmetry via dynamometer and sport scientists track peak in-game accelerations. Rather than siloed reporting, all of this data flows into a hub that guides progressions and ensures everyone is aligned toward the shared goal of improved explosive output.
We can then establish feedback loops at multiple time scales, such as monitoring daily load, weekly adaptation, and monthly trends. This layering allows for both micro and macro decision-making without losing sight of the multiple time scales that the athlete is training within. We can also design "thermostats" that actually influence behavior, such as setting progressive on-field sprint volume thresholds for a player returning from a hamstring strain. These tools guide workload in a similar manner to how we might progress an athlete's squat.
Finally, the model must be updated based on the actual response of the system (i.e. adaptability). If an athlete hits their external workload targets but shows delayed effects such as declining recovery metrics in serial neuromuscular testing, the team should pause and reassess. The plan changes because the model is adaptable, acknowledging that if inputs are sound but outcomes are flat, the loading strategy may need a tweak. Instead of parallel, disconnected workflows, this approach creates a dynamic model designed to coordinate variables rather than control them.
In Thinking in Systems, Donella Meadows warns that when subsystems optimize at the expense of the whole, collapse is near. The answer is better coordination and living models that adapt in real time. High performance is about learning to work with complexity and knowing where to look when things start to break, which begins with an organized performance ecosystem.