A weighted approach to identifying key team contributors: Individual productivity in professional road cycling

Assessing an individual’s contribution within a team remains a fundamental challenge across many domains, particularly when recognition for collective achievements is limited to only a few members. This issue is especially important in professional road cycling, where personal success depends on both individual talent and group effort. Existing points-based ranking systems tend to disproportionately reward high-scoring team leaders while undervaluing domestiques - riders who sacrifice personal success to support group performance. To better capture a rider’s impact on the team, we propose a weighted measure of cycling productivity that factors in race points, a redistribution metric, and an adapted version of the CoScore formula. This formula assesses an individual’s productivity relative to their teammates’ performance. Using data from the 2023 season, we show that our approach offers a comprehensive evaluation of professional cyclists, addressing key limitations of existing ranking systems.

💡 Research Summary

The paper addresses a persistent problem in professional road cycling: the difficulty of quantifying each rider’s contribution to the team, especially when traditional ranking systems heavily favor team leaders and marginalize domestiques. The authors propose a novel “productivity” metric that integrates three distinct performance indicators: (1) the official UCI points earned by each rider during a season, (2) an egalitarian redistribution of a team’s total points weighted by each rider’s participation (measured in race‑days), and (3) a weighted adaptation of the CoScore formula originally introduced by Flores‑Szwagrzak and Treibich (2020).

The CoScore component assigns to each rider a share of a team’s points proportional to the rider’s contribution in each race (p_i(r,t)/p(r,t)). By incorporating participation‑based weights, the authors ensure that riders who consistently support the team, even without high individual scores, receive credit. To combine the three components, the authors employ two convex‑combination parameters: β∈