Evaluating pedaling strategy using a principal component analysis-based analysis of crank force vectors collected during outdoor cycling
Abstract
Purpose
Cyclists adapt their pedaling movements according to workload, a behavior defined here as a “pedaling strategy.” This study aimed to develop a quantitative method for evaluating pedaling strategies using pedal force vector data recorded during outdoor cycling.
Methods
A principal component analysis (PCA)-based framework was introduced to verify the linear relationship between instantaneous pedal forces and effective loads, reduce angle-specific slopes derived from this relationship to a small number of principal components, and assess the feasibility of modeling a pedaling strategy from these components. Outdoor data from 91 cyclists (including 16 professionals) were analyzed. Slopes representing load-dependent changes in pedal forces were computed for each crank angle, and PCA was applied. Statistical comparisons of principal component values between professional and nonprofessional cyclists were conducted.
Results
Instantaneous pedal forces showed strong linearity with the load, enabling accurate slope estimation. PCA reduced the pedaling strategy primarily to four components: PC1 reflected the coordination between downstroke and upstroke forces, and PC2 captured the differences in peak-force timing. Professional cyclists exhibited significantly higher PC1 values and lower variability than nonprofessional cyclists, while variability in PC2 also differed between them. These components effectively characterized individual pedaling strategies, with patterns of positive PC1 and near-zero PC2 indicating desirable strategies.
Conclusions
This study demonstrated that cyclist-specific pedaling strategies can be quantified from outdoor force vector data using PCA, offering an evaluation approach that is not achievable through conventional efficiency indices. Differences between professional and nonprofessional cyclists further support the validity of the method.
© 2026 Masahiro Fukuda, Ryo Koyama, Takashi Inomata, Tomoki Kitawaki, published by Wroclaw University of Science and Technology
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License.