A new hypothesis and theory article published in Frontiers in Sports and Active Living argues that elite sport selection is fundamentally a predictive task rather than a simple evaluation of current performance. The study, authored by Tyson Wellock, distinguishes between procedural compliance and the actual accuracy of the future-facing judgments made by coaches, selectors, and high-performance directors. While existing governance systems can verify whether a selector followed established criteria and acted within their authority, they rarely assess whether the embedded performance forecast was accurate or improved upon a simple baseline. The paper proposes that predictive-judgment benchmarking is a necessary component for earning deference in elite sport governance, suggesting that high-consequence discretion should be matched by systematic records of decision quality.
The research combines published selection-policy examples, public forecast records, and talent-identification research to address an audit gap in Olympic sport systems. In many national sport organizations, a single pathway controls access to coaching, funding, competition starts, and institutional legitimacy. This concentration of authority means that a false-negative judgment can do more than deny a single opportunity; it can remove the developmental conditions required for an athlete to prove a judgment wrong. The study posits that while procedural review is essential for legality and fairness, it does not measure the predictive validity of the decisions that shape athlete careers over time.
The empirical analysis serves as a proof of concept for auditing future-facing judgment using public, attributable, and scoreable forecasts. The primary Olympic sport anchor for this analysis is biathlon, where the International Biathlon Union format made some coach and sport-leader judgments public and scoreable. A scoreable row in the biathlon dataset required a named athlete, a numeric actual rank, a numeric target boundary, and a forecast that could be scored against the end-of-season World Cup total-score result. This approach allows for a direct comparison between expert predictions and a continuity baseline, which assumes that the previous season’s competitive hierarchy would simply continue without additional expert interpretation.
Across 209 scoreable biathlon forecasts, the study found that expert accuracy was 52.2%, while outcome-scored continuity accuracy was 58.4%. This resulted in a forecast-minus-continuity difference of -6.2 percentage points, indicating that the simple continuity rule was more accurate in the recorded mixed-source sample. In a more specific subset of 42 rows involving coaches and sport leaders, expert accuracy was 52.4% and continuity accuracy was 73.8%, a difference of -21.4 percentage points. Although these leader samples were small and concentrated within one season, the results suggest that unaided expert judgment did not measurably outperform the prior-status baseline in this Olympic-relevant context.
To probe the incremental value of expert judgment more sharply, the analysis examined cases where expert and continuity predictions differed in correctness. In the all-scoreable biathlon record, the two methods differed on 101 rows. Expert judgment was correct in 44 of those cases (43.6%), while continuity was correct in 57 cases (56.4%). In the coach and sport-leader subset, the predictions differed on 21 rows, with coach and sport-leader forecasts being correct in only six of those cases (28.6%), compared to 15 correct continuity calls (71.4%). These descriptive results favored continuity, highlighting that when experts departed from the prior hierarchy, they were less often correct than the baseline they overrode.
The study also included comparator contexts from professional and collegiate sports to test the measurement approach across different task structures. In the National Basketball Association General Manager Survey, expert forecasts showed a positive separation from continuity in some categories. For NBA champion prediction, GM forecasts were correct in 43.5% of cases compared to 13.0% for continuity, a difference of +30.4 percentage points. However, in the Most Valuable Player category, there was no advantage, with GM forecasts correct in 26.1% of cases and continuity correct in 30.4%. These mixed results suggest that expert advantage may be more visible in bounded team or contender-identification tasks than in individual projection tasks where feedback is slower and outcomes are more confounded.
In collegiate sports, NCAA football coach-consensus preseason polls retained 61.5% of preseason Top 25 teams in the final Associated Press Top 25, compared to 53.0% for prior-final-poll continuity. This season-level paired comparison favored the coach-consensus poll with a Wilcoxon p-value of .008. In contrast, NCAA women’s volleyball showed higher raw accuracy for experts at 67.5%, but continuity was already strong at 65.2%, resulting in a +2.3 percentage-point advantage that was not statistically detectable. These findings generated, rather than tested, a task-structure hypothesis that expert advantage varies depending on the information conditions and the nature of the prediction task.
The paper emphasizes that these findings do not imply that sport leaders lack expertise or that selection decisions are usually wrong. Instead, they show that high-consequence discretion can operate without a systematic, scoreable record of the prediction it contains. The thinness of the public Olympic sport-leader record is itself a governance finding, as it limits the ability to audit the quality of judgment that controls scarce athlete opportunity. The study argues that if discretion is valuable enough to preserve, it is valuable enough to measure, requiring systems to document the forecast embedded in discretionary decisions and compare it with transparent baselines.
A proposed solution is a lightweight override ledger that records discretionary departures from objective criteria or transparent baselines. This ledger would not require the disclosure of confidential medical or tactical information but would capture the predictive claim, the expected outcome, the time horizon, and the primary basis for the judgment. By recording these elements prospectively, sport organizations can build a scoreable record that allows for the review of judgment quality over time. This approach aims to structure expertise rather than replace it, providing a feedback system that helps decision-makers compare stated expectations with later outcomes and revise their assumptions.
The study identifies several boundary conditions that narrow its claims. Public forecasts are imperfect proxies for private selection decisions because they do not reproduce the incentives, information access, or accountability conditions of formal selection. The evidence contexts are heterogeneous, and the biathlon leader samples are small and context-specific. Furthermore, the mobility gradient used to interpret the results is a structural hypothesis rather than a statistically tested cross-context effect. The paper does not argue for the removal of coach discretion or the automation of selection, but rather for a governance standard that matches concentrated authority with proportional mechanisms for evaluating decision quality.
In conclusion, the research highlights a critical gap between process compliance and decision-quality measurement in elite sport. While existing arbitration systems can review whether selection criteria were followed, they do not usually ask whether the embedded performance projection was later accurate. The evidence supports a practical governance standard where national sport systems that concentrate discretionary authority should also document the forecast embedded in that discretion. By comparing these forecasts with simple baselines and reviewing later outcomes, sport systems can ensure that the deference given to expert judgment is earned through a demonstrated record of predictive performance.
The theoretical framework underpinning the study draws on the distinction between kind and wicked learning environments, a concept from judgment literature that clarifies why sport selection is particularly resistant to standard feedback mechanisms. In kind environments, feedback is rapid, accurate, and repeated, allowing experts to refine their intuition through consistent corrective signals. Sport selection, however, often operates as a wicked learning environment where feedback is delayed, noisy, and incomplete. Selected athletes receive the coaching, funding, and competition starts that improve their chances of later success, while excluded athletes lose access to the very conditions needed to generate comparable evidence. This creates a survivorship problem: systems evaluate the athletes who remained inside the pathway and use their success to validate the process, while the counterfactual performance of excluded athletes remains permanently unavailable. The paper argues that this structural asymmetry means that unaudited discretion becomes a learning problem, not merely a fairness issue, because the system rarely learns what excluded athletes might have achieved under equivalent developmental conditions.
Published selection policies from various national sport organizations illustrate how formal criteria can authorize broad discretion over high-consequence decisions. Examples include USA Gymnastics, where discretionary criteria controlled most final Olympic team positions, and Biathlon Canada, where final nominations were at the sole, full, and absolute discretion of the Performance Pathway Director. These policies often include safeguards such as named committees, conflict-of-interest rules, and appeal processes, which ensure procedural compliance. However, these safeguards typically review whether a procedure was followed or whether reasons were given, rather than whether the embedded performance forecast was later accurate. The audit gap identified in the paper is the absence of a requirement for the predictive component of these decisions to be stated in advance, compared with a transparent baseline, and reviewed against later outcomes. This distinction is critical because a procedurally valid use of discretion can still leave the predictive quality of the decision unmeasured, allowing authority to be exercised without a corresponding record of decision quality.
The empirical design also included a classification-validity comparator using National Football League quarterback tiers, which provided a second professional sport-leader judgment context. Unlike the direct forecast contexts, this dataset involved anonymous ratings by NFL coaches, executives, and evaluators who placed veteran quarterbacks into five tiers annually. Across the 2019 to 2025 articles, the panels ranged from 50 to 55 voters. The analysis examined whether these insider classifications carried forward into later top-12 performance. Expert top-12 classifications had a 47.2% hit rate, compared with 41.7% for prior-tier continuity and 43.1% for prior-performance continuity. These results corresponded to point-estimate advantages of +5.6 and +4.2 percentage points, respectively. This context was retained to test whether repeated insider classification of known performers showed different validity patterns than direct named-outcome forecasting, further supporting the idea that task structure influences the observable value of expert judgment.
The proposed override ledger is designed to be integrated into existing selection minutes rather than creating a separate reporting regime. It would require decision-makers to record only cases where discretion is used ahead of objective criteria or where a selector departs from a transparent baseline. The record would include five fields: the baseline being accepted or overridden, the expected outcome, the time horizon, the primary basis for the judgment, and the review date. A developmental flag could be added when a projected-capacity override rests on expected maturation rather than present performance, making that basis visible for later review. This design keeps the burden proportional by asking decision-makers to state the predictive claim clearly enough to be reviewed later, without requiring public disclosure of confidential medical, tactical, or interpersonal information. The goal is to build a prospective, scoreable record that current practice rarely leaves, allowing sport systems to adjust how much deference a process earns based on demonstrated predictive performance over time.
Subject of Research: Predictive judgment benchmarking and governance in elite sport selection
Article Title: Predictive judgment, continuity baselines, and earned deference in elite sport selection
Article References: Wellock, T. (2026). Predictive judgment, continuity baselines, and earned deference in elite sport selection. Frontiers in Sports and Active Living, 8, Article 1935646. https://doi.org/10.3389/fspor.2026.1935646
Image Credits: AI Generated
DOI: 10.3389/fspor.2026.1935646
Keywords: elite sport governance, predictive judgment, biathlon, selection policy, continuity baseline, decision quality, talent identification, sport leadership, Predictive, judgment, continuity, baselines
News Source: Ophelia Keating. (November 13, 2026). Biathlon Data Shows Continuity Beats Expert Forecasts. Scienmag.
