Teams underperform their potential for many reasons. Roster quality, unclear roles, poor tactics, slow collective adaptation, incongruent game plan understanding, compromised decision-making, fitness, etc. One of the dominant coaching responses to solving these challenges has been to become more data-informed and led. Watch more film, track more player stats and workloads, analyze game statistics and opposition plays in order to identify weaknesses, sharpen preparation, and make better decisions about training, tactics, and selection. The proposition is compelling.
What many coaches have started to realise is that:
1. Having more data and having more understanding are different things.
2. Converting deeper understanding into practical coaching practices is hard.
3. They have gotten better at understanding how teams play, but less why they behave and think the way they do.
Sports analytics has typically focused on what is easiest to measure. Things that can be seen or observed and thus metricized: player strength and speed, pass rates, rushing yards, offensive patterns, possession rates, shots on target, tackles made, etc. What has emerged is a much clearer picture of what players do, patterns of effective play, and competitor insights, but much less the relational conditions that help them perform together.
Data challenges are well documented. Collection scales faster than people's capacity to interpret; description is mistaken for explanation, aggregation removes meaningful signal providing differences. Potentially the most difficult being that data does not supply its own context. A number rarely explains the circumstances that produced it. The same result can arise through different processes and require different responses. Are running yards a response to winning or the cause? How much possession versus what you did with the possession, is running distance attack or defense dominant, etc. In response, the analysts have created modeled proxies and derivative metrics that attempt to make sense of many of the "direct" measures. Corsi is a proxy for possession, which is a proxy for territorial control. Field tilt is a modeled construct that quantifies territorial dominance by measuring where the ball spends its time, then treats that as evidence of control. PPDA (Passes Per Defensive Action) is an inference about passing intent.
While immensely useful, performance data is by its very nature lagging and offers very few clues as to the conditions that are enabling the performance. It fails to measure signals like collective confidence and intra-player win belief variances, player motivation, stress levels, nature and type of communication, energy levels, attention levels, internal intelligibility, and core team sense-making narratives. Furthermore, it fails to surface what's driving these phenomena, namely player differences, perceptions, player connection densities, relationship quality, information flows, and common know-how. These are behavioral conditions that are always in play. The "current" that binds the collective into coherent and cohesive states, or not.
While performance analytics metrics measure observable execution, behavioral data is subtle, ever-present, harder to see and make sense of. Most importantly, it's a leading measure. Conditions form, then shape behaviour, then show up in results. A team's belief in itself erodes over weeks before it shows in third-down conversion rate. These signals give coaches an earlier opportunity to investigate, before a problem becomes evident in results.
Behavioral data is also very personal to this team. Behavioral data provides baseline and trajectory. These signals mean little against a league benchmark. They mean a great deal against this team's own numbers, tracked week over week, with a threshold set in advance for what a widening spread will be taken to mean.
The role of behavioral data is to monitor and highlight process and event interpretation, difference, and dispersion within the team. The signal lives in the spread, not the average. Behavioral data makes explicit, team and situation-specific observations without telling you what to do about it. Their interpretation and remediation sits in the hands of the coach and the team itself.
1. Player confidence levels staying stable but collective efficacy scores declining. (Every player still believes in himself. Fewer believe in the team.)
2. Players having increasingly divergent (same metric) post-game interpretations.
3. Aggregate team wellness metrics staying stable but with greater standard deviations.
4. Players' stress level scores rising for three consecutive weeks.
5. Players and coaches disagreeing on the same post-game metric interpretations.
6. Players having a wide array of MVP picks, and different to coaches.
7. Coaches evaluating identical in-game criteria fundamentally differently to each other.
Difference is not automatically dysfunction; it is a reason to explore what players and coaches are noticing and valuing and why.
Surfacing behavioral proxies gets coaches closer to the inescapable phenomena that influence and produce team dynamics, and the performance it helps enable or constrain. Behavioral data captures how the collective mind is forming and responding throughout a season. The data delivers information that is less analytically prescriptive and algorithmically definitive than it is informative and forewarning. It doesn't scream the answers, it whispers what's brewing and where to look.
Behavioral data sheds light on where the team is heading, what it's thinking, how aligned they are, where they are disagreeing, and by how much. Figuring out how to respond is where "this team" advantage lies.
Circle-In gets teams to perform closer to their potential by making team behavior more visible: who is connected to whom, the quality of their relationships, what players are thinking and feeling, and the collective behavioral state this affords, at any point in the season. We make players more intentional, coaches more informed, and the team collectively smarter.
Graphic created by Iconins
